You are on page 1of 508
V.A.KRECHMAR eo tey ANY BOOK } IN ALGEBRA a : 4 yi MIR PUBLISHERS :- MOSCOW ABOUT THE BOOK ‘The object of this collection of probleins and exercises in elementary algebra is tu provide teachers and pupils with material of a higher degree of difficulty than that in stan- idary schuol texthouks, so js who wish to deepen their tical knowledge can do su, The oxercis the collect can be tackled by any reader who has studied elementary mathematics, but nol gone beyond il. Though some probl are in trigonometry, they wre in fact all of an algebraic ehara- eter, The problem grouped, and each group is provided with notes on how to go about solving them Answers are given for almost all prob- Jems, with a few exceptions. If the reader's knowledge and appree of mathematics is fostered by working through the book, the author's will have been justified, 3AJIAUHMK 10 AJITEBPE Uanateanerno «Hayxar V. A. KRECHMAR A PROBLEM BOOK IN ALGEBRA Translated from the Russian by Victor Shiffer Translation editor Leonid Levant MIR PUBLISHERS MOSCOW First published 1974 Second printing 1978 TO THE READER Mir Publishers would be grateful for your comments on the content, translation and design of this book. We would also be pleased to receive any other suggestions you may wish to make. . Our address is: USSR, 129820, Moscow 1-110, GSP Pervy Rizhsky Pereulok, 2 MIR PUBLISHERS Ha aneauiicrom aaixe © English translation, Mir Publishers, 1974 Printed in the Union of Soviet Socialist Republics 40. CONTENTS . Whole Rational Expressions . . . . Solutions to Section 1. .....-.- . Rational Fractions... . . Solutions to Section 2... 2. 2... ee ee ee ee . Radicals. Inverse Trigonometric Functions. Logarithms Solutions to Section 3 . Equations and Systems of Equations of the First Degree Solutions to Section 4... 2... 2.2... ee eee . Equations and Systems of Equations of the Second Degree Solutions to Section 5... 2... . Complex Numbers and Polynomials ........-. . Solutions to Section 6.......... . Progressions and Sums... ......-....205 Solutions to Section 7... . 2... 2... ae . Inequalities See a area Pa eee cee eee aaa Solutions to Section8 ...... Reese eee eee . Mathematical Induction... ...... Solutions to Section 9... ......... see Mantes eee ee ceratstite ater ratte ea HateRE eeeeee react . PROBLEMS 1. WHOLE RATIONAL EXPRESSIONS The problems presented in this section are mainly on the identity transformations of whole rational expressions. These are the elementary operations we have to use here: addition, multiplication, division and subtraction of mono- mials and polynomials, as well as raising them to various powers and resolving them into factors. As regards trigono- metric problems, we take as known the definition of trigonc- metric functions, principal relationships between these functions, all the properties connected with their periodi- city, and all corollaries of the addition theorem. Attention should be drawn only to the formulas which enable us to transform a product of trigonometric functions into a sum or a difference of these functions. Namely: cos A cos B=-4 [cos (A+B) + cos (A—B)], sin A cos B=-4 [sin (4 + B)-+sin (A—B)], sin A sin B=-+ [cos (4 — B)—cos (A+ B)]. 1. Prove the identity (a? +b) (x? + y?) = (ax — by)® + (br + ay). 2. Show that (a? + B+ 8 + a) (a + y+ 22 + A) = = (ax — by — cz — dt)? + (bx + ay — dz + ct)? + + (ex + dy + az — bt)? + (dx — cy + bz + at)?. 3. Prove that from the equalities az —by—cz—dt=0, br +ay—dz+ct =0, ex + dy+az—bt=0, dx—cy+bz+at~=0, 8 Problems follows either a = b =c=d=0,orr=y=z=t=0Q, 4. Show that the following identity takes place (a? +b +c?) (a? + y? + 2%) — (ax + by + cz)? = = (bx — ay)® + (cy — bz)* + (az — cz). 5. Show that the preceding identity can be generalized in the following way (@ tat... +0) (BHR+...+ 0) = = (aby + Gaby +... + dnb)? +. (4b, — aebs)? + + (abs — ashy)? +... + (Gn-tbn — Onbn-t)?. 6. Let n@+P+ej r+...+~—= =(atb+e+... 49% where n is the number of the quantities a, b, c,..., 1. Prove that then a=b=c=...=l. 7. Prove that from the equalities aitat+...taz=1, B+R+...¢b=1 follows —1 + B8 + c8) = Sabe (a? + b? + c%); 2° 5 (a? + be + c) (a? + B+ c4) = 6 (a5 + B+ C4); 3° 10 (a? + b? + c?) = 7 (a? + B? + c*) (a> + BF + €), 27. Given 2n numbers: a, dg, . . ., Gn} by, bay. - +) By. Put Cees eee re Prove that ayby + dade +... + Gybn = (a, — a2) 81 + (a2 — 4) 82+ tee + Gn-t — Gn) Sn + Ons. 28. Put a+a+...+4a,=755. Prove that (s — a)? + (s — a)? + oe. + (8 — a) = =aQ+a+...- a3. oe Given a trinomial Az* + 2Bzy + Cy’. u w= z= an’ + By’, y= ya’ + by’. Then the given trinomial becomes Ata? 4 2B'x'y’ + Cly"?. Prove that BY? — A'C’ = (BP — AC) (a5 — By)’. 30. Let Pta=1 @=1, 2,..., 7) and p= Put pate + Pn ers Tie nee st Prove that Pigs + Paget «+ «+ Pngn = rpg —(P1— py? —(P2— p)?— ...—(Pn— Pp). 42 Problems 31. Prove that 1 1 tmatp mst +RoT eit A (eget tty) 32. Let m=t+t44t4 0.44, Show that de nan (atat ee ll n—1 2 ns = nt (74 — ae 33. Prove the identity 4 4 1 aed (—p4p-gt tg moar tate t+ 4 +3 34. Prove (1+a2a) (4-gar) (14+ eer) x x 4 1 x (t+ (2n—f)a—1 \(1- ani) = ___(mttya (n-+2)@ (nna ~atha—t ""(n+2)a-1 °°" (@fnja—t 35. Let [a] denote the whole number nearest to @ which is less than or equal to it. Thus, la] > ee 7 20 Problems Prove that two of the three fractions must he equal to +4, and the third to —1. 25. Show that from the equality S44 +o-oRe follows tet eos if n is odd. 26. Show that from the equalities bz--cy ___capaz ay }-br z(—az-+-by-+os) ~ y(ax—by oz) 2(az--by—ca) follows generac eee esas cee ap e—at) Date) (a2 F522) © 27. Given a+Bp+y=0, a+b+c=0, eee Prove that aat+ Bb*-+ yc! =0 28. If a + b+ ¢%=(b+¢)(a+c) (a+b) and P+e—a)c=(?+a—B) y= (e840? —c’)z, then | t B+p+e=(e+y) (+2) y+2). 29. Consider the finite continued fraction wrth. a ay 2. Rational Fractions 24 Put Po=d, Qo=1, Pr=apayti1, Qi=a and in general Prii=QruPat Pras hts = OniQn + Ons Then, as is known, Pa A On Oa te, fede an (n=, 4, 2, 3, ...). Prove the following identities (St —1) (1B) = (Se) (1B): Past (=1)r4 | o Pn Po __ 1 Pea art tae 3° PaszQn-2— Pr- ae = (Gn42@nisdn+ Anse t an) (— 1)"5 Pr 4 Pea e ages 1 aaa Oni 4 Ont ta te, 4 seagate 30. Put for brevity wtey, = (do Hy +62 n) = eo , fees and let the fraction be symmetric, i.e. Q=An, &=—An4,.... Prove that Pr1t=Qn 31. Suppose we He a fraction Ftt yt ‘ + Geet > 22 Problems Prove that Prt Prot =PnsPasit PnPrive 32. Let 4 = — 4 ates. ete “TT4+744 ate+.. 1 +a Prot be, respectively, the last and last and let 2" and Qn n but one convergents of the fraction 4 ath o+:, 4 ‘+7. Prove that PrQn+PnPn-s r= Ont PnOn- 33. Consider the continued fraction a, TH De Put Po=by Qo=l, Pr=bobita, Qa=dy-.. and in general Pris = dnssPat QrPrar Qhat = baiQn + Anns Prove that a nO 1 an ~Hon, by 2. Rational Fractions 34, Prove that r HT spr = Tet vA ., rnp ei ae FHT (the number of links in the continued fraction is equal to n). 35. Prove that 4,4 1 tw teeta 1 aaa “i ub win — 2 Ug ug — = . Mn} Unatun * 36. Prove the equality i yay eee = crag ate 4. Geert eabe tiaele ete pte op Saeiintn on nbn where ¢, Cy, ...,¢n are arbitrary nonzero quantilies. 37. Prove the following identities 4° sin(n+1)z _ sinnt (a total of n links); 2° 1+-b,+ bobs+ ...+ bobs... bn = 24 Problems 38. Prove that 1° sina+sinb+sinc—sin(a+b+c)= gen at) . ate . bte , = 4sin 3 sin—— sin 5 2° cosa+cosb+cose+cos(a+b+c)= =4cos £t* cos PAE cos SHE ‘ 39. Show that sin(afb+c) _ tana + tanb + tan ¢— 2 pcos = tan atanb tanec. 40. Prove that if A+B+C=n, then we have the fol- lowing relationships 1° sin A +sin B-+sin€ =4c0s4.cos 3 cos $; 2° cosA+cosB+cosC =1 +4sin A sin sin $ ; 3° tan A+tan B+ tanC =tan A tan B tanC; ve tan A tan 3 + tan tan $+ tan 2 tan = ; 5° sin 2A + sin 2B+ sin 2C =4 sin Asin BsinC. 41. Find the algebraic relations between the quanti- ties a, b and ¢ which satisfy the following trigonometric equalities 1° cosa-+cosb-+cose=1-+4 sin sing sin-$; 2° tana+tanb+tanc=tanatanb tanec; 3° cos? a+ cos* b+ cos® c-—2 cos acosbcose = 1. 42, Show that = y z Azys ia timp time Waa xy+azt+yz=1, 2. Rational Fractions 25 43. Show that the sum of the three fractions be c—a a—b Thee? Tpac? Tyab is equal to their product. 44, Prove that tan 3a = tana tan ($+2) tan (4-2). 45. Prove that from the equality sinta , costa 1 icara get aprean ao follows the relationship + et c ar : 46. Suppose we have @; COS Oy + a2 C08 A+... +anCosa,=0, ay Cos (a; ++ 8) ++ a2 Cos (a2 -+ 0) + ... +n C08 (An + 0) =0 (8A kn). Prove that for any 4 Gy COS (044+ A) + az COS (&_ +A) +... + an Cos (An +A) =0. 47. Prove the identity sin(B—y) _, sin(y—a) , sin(a—B) _ 9 cosp cosy ' cos ycosa ae cos @ cos 48. Let in a triangle the sides be equal to a, b and c, and let Poin eee > r= poy Foe where s is the area of the triangle and 2p=a+b+c. Prove the following relationships 2 ee BR arte tres Tar Ter Te ary bry cre pe a 2 aera t+ Oa baa t+ ae = a6 Problems o atbfe e Tatrotre ° be wayanoe t aes t va +p4t)=4 2 ae mete 4 git (a+) re Pp (e—ay(e—by or * 49. Prove the identity sin (a+6—c—d)= sin sin (b—c) sin (b—d) a sin (b—a) 50. Given ce afb a cosO=7——, cosg= b Phe ape? COSP= (9, p and wp lie between 0 and x). Knowing that a, b and c are the sides of a triangle whose angles are A, B and C, correspondingly, prove that 4° tan? 2+ tan? £4 tan? B= 1; 2 tan 2 z ° tan tan t=tan4 tan aE 2 tan <. 51. sin (@—b) sin (@—<) 2. Rational Fractions 27 Prove that aaecay t ampSaeeaa + +saESanS A 4 = = 2003 ==? cos $=" cos 2 2 52. Prove the identities ° 2° 53. ° = 2° 3° ° s sina sind ain (@—ay sin (@—e) + ~sin(b—aysin—ay i ieietrem erence Gs sin (@—a) sin(e—b) if cosa cos b sine sin (ae) + ain ayant cos ¢ gin (¢—a) sin (¢—b) =0. Prove the identities sin asin (b—c) cos (b+ ¢—a) + + sin b sin (ec —a) cos (¢ +a—b) + +sin ec sin (a—b) cos (a+b—c)=0; cos a sin (b—c) sin (b++-c—a) + + cos b sin (ec —a) sin (c++ a—b) + + cos ¢sin (a— b) sin(a+b—c)=0; sin asin (b ~c) sin (b-++-c—a) + +sin 6 sin (c—a) sin (c-+a—b) + +sincsin (a—b) sin (a+b—c)= = 2 sin (6 —c) sin (c—a) sin (a—b); cos a sin (b—c) cos (b4+-¢ —a} + + cos b sin (c —a) cos (ec + a— b) +. + cos ¢ sin (a —b) cos (a + b—e) == = 2 sin (b—c) sin (ec —a) sin (a6). 1 28 : Problems 54. Prove that 1° sin? A cos (B—C) + sin’ B cos (C — A) 4- + sin’ C cos (A— B) =3sin Asin B sin C; 2° sin? A sin (B — C) + sin? B sin (C — A) + + sin? C sin (A — B) =0 if A+ B+C=n. 55. Prove the identities 4° sin 3A sin? (B — C) + sin 3B sin? (C — A) + + sin 3C sin* (A — B) = 0; 2° sin 3A cos® (B — C) 4+ sin 3B cos® (C — A) + + sin 3C cos? (A — B) = sin 3A sin 3B sin 3C fA+B+C=n. 3. RADICALS. INVERSE TRIGONOMETRIC FUNCTIONS. LOGARITHMS The symbol j/ A is understood here (if n is odd) as the only real number whose nth power is equal to A. In this case A can be either less or greater than zero. If n is even, then the symbol VA is understood as the only positive number the mth power of which is equal to A. Here, neces- sarily, A >0. Under these conditions, for instance, VA=A if ADO, VA@=—-A if A= —1)? if k is even, cos 0 if k is odd. Further, we use the symbol arcsin z to denote an arc whose sine is equal to x and which lies in the interval between -> and +H: Thus, in all cases —$ 0, b> 0); 1 4 1 2 (3 [(@ +0) — al [a + 64)% —B)}F = 2 =(a+b)> — (a?—ab+ b) 8. Compute the expression oe 3 1 4 (4 —azx) (1+ ax)7} (1 4-b2)® (1—bay 2 at 4 2 z=at (24-1) (O1 Va+2Va—1t +Vr—2Va—1 is equal to 2 if <2, and to 2Va—1 if e>2. 12. Compute V a+b+c+2Y ac+be +V a+b+c—2V ac+be (a>0, b>O0, c>0). 13. rove that the trinomial e+ Petd vanishes at at 14. Express x in terms of a new variable so that Vata and YVa-+-b become rational. 15. Rationalize the denominator of the fraction 4 ab fearon et 416. Prove that 7/2 cannot be represented in the form p+Vq, where p and q are rational (¢>0 and is not a perfect square). 3+1225 34 Problems 17. Prove the following identities tan (3 a) cos ($a) 4° cos Oa) + cos («—})sin (a—a)+ -+ cos (1 +) sin (2-$) =; 2° [1—sin (31 — a) -+ cos (3-+-@)] X x [4—sin (3-2) +c0s (#2) ] + sin 2a =0; 3° [1—sin (1+ @) 4 cos (n+)? 4- +[1—sin (3 40) + 3a. +c0s (2-2) =4—2sin 2a. 18. Let a=2kn +a, where U 2 (mod 5), A,=0 if n , 3, 4 (mod 7), Ay =0 if n=1, 2 (mod 5), Ay =0 if n==1, 2, 3, 5, 7 (mod 11), Ay; =0 if n=2, 3, 5, 7, 9, 10 (mod 18), Ay, =0 if n=4, 3, 4 (mod 7), Ay =0 if n=0 (mod 2), Ay, =0 if n=4, 3, 4, 6, 7, 9, 18, 14 (mod 17), and that A>, A3, Ay, Ag, Ag, Ag, Ato, Ais and Ayg never vanish for any whole n (S. Ramanujan. Asymptotic formulae in combinatory analysis). 20. Let un P(n)= A (n 4-3)? + B4-C(—1)"+ Deos 3 (nan integer). Prove that there exists the following relationship P(n) — p(n — 1) — p(n — 2) + p(n — 4) + + p(n — 5) — p(n — 6) = 0. 24, Show that 4° sin 15° Vo V2 cos 15° Ver V2 ; 2° sin 18° = —1+V5 , cos 18°=4.V 1042/5. 22. Show that sing? — V20—8 V5 —V6+2V5 ee oe gee V 1846 V5 +V 10-25 wD tb, 23. Show that cos (arcsin x)= /1—a, sin (arccos z) = V1—2*. tan (arccot x) = 4 > cot(arctan x) = £ . 38 Problems cos (arctan z) = » sin (arctan x) = oe Vite” cos (arccot x) = Vi’ sin (arccot x) = ve : 24. Prove that n . m arctan z+ arccot z= zr arcsin x + arccos z = z: 25. Prove the equality zy T—2y where ¢=0 if 2y1 and <0, e=+1 if sy>1 and c>0. arctan z+ arctan y = arctan +en, 4 1 au 26. Show that 4 arctan arctan y= 27. Show that arctan 4 -+-arctan + + arctan 7 + 4 x ++ arctan I-T: se 28. Show that 2arctanz-aresing “=z —=m (x >1). 29. Prove that arctan z+ arctan t= + if z>0, 4 Lan arctan z+ arctan > = —— ifz<0. 30. Prove that arcsin z+arcsin y=mnarcsin (zt V1—y*+y V 1—2*) + en, where n=1, e=0 if zy <0 or 2+y'<1, n=—1, e=—1 if @4+y>1, 2<0,y<0, no—4, e=4t if s4y5>1, 2>0, y>0. 3. Radicals, Inverse Trigonometric Functions. Logarithms 39 31. Check the equality arccos «+ arccos (= + + V3= 32) =% if f and B-== arctan z then prove that cos 2A:=sin4B. 33. Let a?+ b? = Tab. Prove that +b cf log “3 = 5 (log a+ log 8). logan __ 34. Prove that Mogami 1+ log, m. 35. Prove that from the equalities z(y+2~2) _ y(@te—y) _ 2(z+y~2) log x logy log z follows x¥-y* = 24 .y*? = a*-2*. 36. 1° Prove that log, a-log.b=1. 2° Simplify the. expression log(log a) q lea (logarithms are taken to one and the same base). 1 1 37. Given: y=10!—-le*, z=401-ley (logarithms are taken to the base 10). Prove that 1 z= 10 T-loeF, 38. Given. a+ b= 40 Problems Prove that logorc @ + logy. a = 2 loge41, a loge-1 a. 39. Let a>0, c>0, b=Vace, a, ¢ and ac41, N>0. Prove that logy N __ loga N--logy N Tog, VW ~ ‘logy V—log, N * 40. Prove that logajay...0,° = —F T Tog, = 1 Tog, = 4 Tog, = 41. Given a geometric and an arithmetic progression with positive terms @, Ay, Ga, 0) Any ve b, by, Be, . ey Day eee The ratio of the geometric progression and the common difference of the arithmetic progression are positive. Prove that there always exists a system of logarithms for which log a, — b, = loga — b (for any n). Find the base B of this system. 4. EQUATIONS AND SYSTEMS OF EQUATIONS OF THE FIRST DEGREE The general form of a first-degree equation in one un- known is Ac+B=0, where A and B are independent of z. To solve the first- degree equation means to reduce it to this form, since then the expression for the root becomes explicit B z=—s. A 4, Equations and Systems of Equations of the First Degree at Therefore the problem of solving the first-degree equation is one of transforming the given expression to the form Ax + B =0. In doing so great attention should be paid to make sure that all the equations involved are equivalent. The problem of solving a system of equations also consists to a considerable extent in transforming a system into an equivalent one. This section deals not only with equations of the first degree in the unknown z, but also with the equations which can be reduced to them by means of appropriate transforma~ tions (such are equations involving radicals, trigonometric equations and ones involving exponential and logarithmic functions). Here and in the following section we consider a trigonometric equation solved if we find the value of one of the trigonometric functions of an expression linear in zx. Indeed, if i¢ is known that tan (mx +- n) = A, then we find max -+- n = arctan A +- kn, where & is any integer. Consequently, all the required values of z are given by formula ___aret {hse m Likewise, if it is found that cot (mx-+n) =A, then mz+n=arecotA+kn and eee ™m But if it is known that sin (mz + n) = A, then all the values of x satisfying the last equation are found by the formula ma + n = (—41)* aresin A + kn, where k, as before, is any integer. 42 Problems Analogousty, from the equation cos (mz +n) =A follows mz +n = -arccos A + 2kn. When solving exponential equations one should remember that the equation a =1 (a>0 and is vol equal to 1) has the only solution x = 0. 1. Solve the equation ¥ z—ab | zac . t—be app tape | bye =a+b+e. 2. Solve the equation r—a , -x—b zc 1 4 1 fends tana (z+7+4)- 3. Solve the equation 6x4 2+ate — Ir+bapb+3e 62+; 24—3b—c Qn ba—b—se* 4. Solve the equation afb—z , a}c—z , bte—a , 4a ieee eaap cuts apoyo 7! 5. Solve the equation Pore Woy giRean ABs / 6. Solve the equations 1° Ve+i+Vae-1=1; 2 Veti-Ve-i=1 7. Solve the equation ' WV atVitV a—Va=Vb. 4, Equations and Systems of Equations of the First Degree x 8. Solve the equation Vi- VY t— 9. Solve the equation 10. Solve the equation Vere+V=* _y5 doy, 11. Solve the system x y try a+2 y 12. Solve the system 2+ a2 + ty + ty = 2ay ay + ty — 23 — Z, = 2a, n Ty — 12+ 23 — 4, = 2a ty — 22 — 23+ r= 2a,. 13. Solve the system ax+m(yt+z2+u =k by +m(r+z24+v =1 cz+m(zr+y+v)=p du+m(r+y+2=q. 14. Solve the system 24 04 aos my mg te TH+ oat... +rp=a. 43 44 Problems 15. Solve the system Mepaeiee dl ipa tertu eerste ed tiene 1 4 4 vtatE 1 1 4 epilegi ap ate 16. Solve the system ay+br=c ca + az =b bz + cy =a. 17. Solve the system cy ++ bz = Qdyz az + cx = 2d’2r bx + ay = 2d"zy. 18. Solve the system zy eid pee ayo ~° “Ge-pex =6 Bey ah 49. Solve the system ype—ra ke zpe—y=E z+y—2=AF A 20. Solve the system if (b+ ¢)(y+2)—ar=b—e (¢ + a) (+2) — by =c—a (a+bd)@+y—ca=a—b at+b+c#0. 4, Equations and Systems of Equations of the First Degree 45 21. Solve the system (c+ a)y+ (a+ b)z—(b+ 0c) a = 20° (a + b) zt (b+ 0) — (c+ aby = 20° b+)zr+(c+a)y—(a+ d)z=2 if b+c#0, a+cK0, a+ b40 22. Solve the system . att : arp y tet eat hey ees 23. Solve the system 24-ay+a@x+a=0 2+by+Br+b>=0 ztey+ex+=0. 24. Solve the system z+ay+a%x-+a%t +at=0 2+ by-+bx+ bt +b4=0 atoytert+cert+c=0 z+dy+-@x+dt+dt=0. 25. Solve the system atytz+u=m az+by+cz-+du =n @xrt by +-es+@usk ax + by +32 + Bu =. 26. Solve the system D+ 2ay.4+3r3+ ... +nrn ty 4-2r343a,+ «6. nay Tp+ 22,4 8a,+ 2. +nra-+ 46 Problems 27. Solve the system Xy—t_ — %— ... —My=2a 244+ 8x%,— %3—- «1. —ty=4a —&4— %+T#3— ... —an= 8a 28. Solve the system T+ %_4-X3+ -- Uy+as+ oe Ly+ MM oF Upp ayt os + opy=n. 29. Show that for the equations azr+b=t, ar+b'=0. to be compatible it is necessary and sufficient that ab’ —a'b=0. 30. Show that the systems ax+by+ce=0 az+by+te =0 and Uar+by +o) 4 (a'x +b'y +e) =0 : m(azx-+ by +c)+m' (a'a+b'y+c’)=0 are equivalent if Im'—l'm #0. 31. Prove that the system ax+by +e =0 a’z+b'yte’=0 has one and only one solution if ab’ —a'b0. 4, Equations and Systems of Equations of the First Degree aa 32. Prove that from the equations ax+by =0 a’'z+b'y=0, if ab'—a'b=0, follows r=y=0. 33. Show that the following three equations are compatible ax+by +c =0, a’z+b'y+c'=0, a'r + "y+ c"=0 if a" (be — b’c) + b" (ca” — c’a) +c" (ab! — a’b) = 0. 34. Let a, b, ¢ be distinct numbers. Prove that from the equations: x+ay+az=0, a -- by + bz = 0, rtey+ce%=0 follows z=y=2=0. 35. Prove that from the equations Ar+By +Cz =0, Ayz + By +Cz2=0 follows x y z “CyB—CB, ~~ CAy—C,A ~~ ABy— A,B if not all of the denominators are equal to zero. 36. Prove that the elimination of z, y, z from the equations az + cy + bz =0, cx +- by + az = 0, br + ay + cz =0 yields a 4b +c? — Babe = 0. 48 Problems 37. Given the system St+i=4 (144) 5-i-4(1-4) SE (1-4) Ay) Prove that the equations are compatible and determine z, y and z, 38. Determine whether the equations of the system (at b)x + (apt bq) y = ap? +67 (ap + bg) x + (ap? + 69°) y= ap? +- bg? (ap"* + bg) 2+ (ap" + og") yap" + bgt are compatible. 39. Solve the system By ry =a Ty 23 = Ag Ty-+ aX, = a3 pap nS An Ent =an 40. Solve the system 4, Equations and Systems of Equations of the First Degree _ 49 Al. Solve the system (x + a) (y +) = (a—n) (Ld) (y + b) @ + m) = (6—D(m— oe) (z + ¢) (@ + n) = (c — m) (n — a). 42. Determine k for the system tity =0 (A—khatky =1+k (+k) e+ (12 —k)y = —( +4) to be compatible. 43. Solve the system azsina + y sin 2a + 2 sin 3a = sin 4a x sin b + y sin 2b + zsin 3b = sin 4b zsine + y sin 2c + z sin 3c = sin 4c. 44. Show that from the equalities wea ewe, AtB+Can follows a=bcosC +¢ cosB, b=ccosA +acosC, ec =acosB + bcosA. 45. Show that from the given data a=bceosC +c cosB, b=ccosA + acosC, e¢ =acosB + bceosA, O0, b>0, ¢>0, follows leet earosHe teeta) sin A sin ~ sine and A+B4C=n. 44-1225, 50 Problems 46. Given a=bcosC +ccosB a = b? + c® — 2be cos.A b =ccosA + acos C (1) b? = a® + c? — 2ac cos B (2) c acosB + bcosA c? = a? + b? — 2ab cos C. ll Show that systems (1) and (2) are equivalent, i.e. from equations (1) follow equations (2) and, conversely, from equations (2) follow equations (1). 47, Given cos a = cos bcose + sin b sinc cos A, cos 6 = cosacose + sina sine cos B, (+) cosc = cosacos b + sina sin bcos C, where a, b, c and A, B, C are between 0 and a. Prove that sinA_sinB_ sinc sina sinb~— sine * 48. Prove that from the conditions of tne preceding problem follows 1° cos A = —cos BcosC + sin B sinC cosa, cos B = —cos A cosC +- sin A sin C cos b, cos C = —cos A cos B + sin A sin B cos ¢; 2° tante Vf tan ¥ tan ife=A+B+C—n and 2Zp=at+b+te. 49. Solve the equation (6 — c) tan (x + a) + (c — a) tan (x + B) + + (a — 6b) tan (x + y) = 0. 50. Prove that sin z and cos z are rational if and only if tan > is rational. 51. Solve the equation sintz + costz =a. Equations and Systems of Equations of the First Degree St 52..Solve the following equations 4° sin x + sin 2x + sin 3x = 0; 2° cos nz + cos (n — 2) —cosx = 0. 53. Solve the equation 4° m sin (a — z) = nsin (b — 2); 2° sin (x + 3a) = 3 sin (a — 2). 54. Solve the equation sin Sz = 16 sin® z. 55. Solve the equation sin z + 2 sin z cos (a — z) = sina. 56. Solve the equation sin z sin (y — z) =a. 57. Solve the equation sin (a + z) + sina@ sin z tan (2 + z) = mcosa@ cos z. 58. Solve the equation cos? a + cos? x + cos? (a + 2) = 1 + 2 cosa cos (a+2). 59. Solve the equation (4 — tan z) (1 + sin 2z) = 1 + tang. 60. Show that if tan z + tan 2z + tan 3z + tan 42 = 0, then either 52 = kn, or 8cos 2a =1+ Vii. 61. Given the expression ax* + 2bay + cy?. Make the substitution z= X cos 8 —¥ sin 8, y = X sin ®@ + Y cos @. rn 52 Problems It is required to choose the angle @ so that to ensure the identity ax® + 2bay + cy? = AX* + BY*. 62. Show that from the equalities z y 2 Tan (OF ay tan(OFP) tan O-F yy follows Zt sint (@—p) + 42% sin? (B— 9) + $45 sin®(y—a)—0. ye 22 63. Solve the systems ge Sinz _ siny _ sing a b ce: etyte=T; ge fanz _tany _ tans a b c cty+2=n. 64. Solve the system tanztany=a aty=2b. 65. Solve the equation -1 x 4 ais Fag 8p, 66. Find the positive solutions of the equation Father 67. Solve the system a*b’=m z+y=n(a>0,b>0). 68. Solve the system way* a* =". 69. Solve the system (ax) * = (byy#® plog x — logy, 5. Equations and Systems of Equations of the Second Degree _58 70. Solve the system way amy". 5. EQUATIONS AND SYSTEMS OF EQUATIONS OF THE SECOND DEGREE The present section contains mainly problems on solving quadratic equations and using the properties of the second- degree trinomial. It should be remembered that if the roots of the trinomial az? + bx + c* are imaginary, then this trinomial retains its sign at any real values of z. As is easily seen in this case the sign of the trinomial coincides with that of the constant term (i.e. with the sign of c). Thus, if c > 0 and the roots of the trinomial az? + bz + c are imaginary, then az*+ br +e>0 for any real zx. When solving systems of equations the following proposi- tion should be taken into account. Let a system of m equa- tions in m unknowns be under consideration, the degrees of these equations being, respectively, key, hay. oy Kms Then our system, generally speaking, allows fork,k,. . . k,, solution sets. To be more precise, the product of the degrees of the equations is the maximal number of solutions. Sometimes this limit is reached (see Problem 23), but some- times it is not. Nevertheless, this proposition is of impor- tance, since it prevents the loss of solutions. 1. Solve the equation Ot aye+e 1 pp C+ O42) og (ebay (edd) eH a9 +E Gag bay + eae OT * In this section the letters a, 6, c, p, g and other constants in the equations denote real numbers. 54 Problems 2. Solve the equation . a (b—c) (c—b) (c—c) + B3 (c—a) (x—c) (n— a) + +c3 (a —b) (tx—a) (x--b) = and show that if the roots of this equation are equal, then exists one of the following equalities 4 = tata =0. Vit Wt V 3. Solve the equation (a—z) Va—z—(b—2) Vz—b coat Va-z4Vz—6 4. Solve the equation V 4a+b6—52+)46+a—5x—3 Va+b—2z=0. 5. Prove that the roots of the equation (x — a) (rx —c) + A(x — b) (x — a) = 0 are real forany 4 ifa0. 5. Equations and Systems of Equations of the Second Degree _58 41. Let a+y+tz=a. Show that then ety tea". 12. Prove the inequality atyt+20, B>0). Express {/a-+{/ in terms of the coefficients of the equation. 15. Show that if the two equations Av +Be+C=0, A’c®+ Br +C' =0 have a common root, then (AC’ — CA’)? = (AB’ — BA’) (EC’ — CB’). 16. Solve the system riety +)=@ y@@ty+)=0 a(e¢tyta=c. 17. Solve the system a(e+y+2=a-—yz y@@tyta=b—-xz a(a+y+2) =c—azy. 18. Solve the system y+2e+2=a(y+2)(¢+2) z+2y+c=b(2+y)(e@+y) e+z+y=c(y +2) (+2). Problems 19. Solve the system y+st+yzma c+z+a2=b r+ytay=me, 20. Solve the system yz = ax w=by (a>0,b>0,¢>0). zy = cz 21. Solve the system e+ yf crys a2 = ba yt a 22. Solve the system r(y+2)=a? y (x + 2) = z(@vuty)=c. 23. Solve the system w= ax + by y® = bx + ay. 24. Solve the system z=at (y—2)* y=b+(e—2) B@=e+ (x —y)% 25. Solve the system b(z+y) a e(z4 2) SH+y+ery ' zp2}bcr ~ % ets) 4 _a(ety) ytztayz © z+y+ery a(z+z) , bt) apatbezs * ytatays 5. Equations and Systems of Equations of the Second Degree 57 26. Solve the system a—y=a y—a=b at?_— ay =c. 27. Solve the system ypte?—~ytarra e+2—(¢+2y=b e+y—(r+y)2z=c. 28. Solve the system e+yt+ayad 2+o24 a= 0? yt e+ yz =a’. 29. Solve the system P+pP+eaa a+ y+ 2 = a? zgty +2 =a. 30. Solve the system ai+yt+atut=at P+y¥4+7+4+ 0 = a3 + yt 2 +u7 = at z+y +z+u =a. 31. Prove that systems of equalities (1) and (2) are equi- valent, i.e. from existence of (1) follows the existence of (2) and conversely. a@+h+c? =1, aa’ +bb’ +cc’ =0, a?4+b%4c%—=41, a’a’+b'b" +e'c" =0, (4) a4" 1, aa" + bb” +cc” =0; ata %pa%=1, abta’b'+2's"=0, PHOT D a1, bet bie! +b” =, (2) ef +c%+c% = 14, ca+c’a’-+c"a" =0, 58 Problems 32. Eliminate x, y and z from the equalities Pytaj=a, yY(et2z=b, F(aty=c, cyz=abe. 33. Given z 2 zy pas eee gee ates eee eer ere zy 8 ye Eliminate z, y and z. 34. Eliminate z, y, z from the system y+ 2 — 2ayz = 0 2+ 2? — 2baz = 0 a+ y? — Qcry = 0. 35. Show that the elimination of z, y and z from the system yYptyte=a 2+az+a? = e+aytypac zy +yz+az =0 yields (a+ b+c)(8+e¢—a)(a+cec—b)(a+b—c) =0. 36. Eliminate z and y from the equations zgty=a, P+yY=b, P+ y¥=c. 37. Eliminate a, b, c from the system zy fae bc el @?4+R+cei=1 atb+c=1. 38. Given ayes Speatecanticg tan z z atetyas 5. Equations and Systems of Equations of the Second Degree 59 Eliminate z, y and z, 39. Prove that if zrty+z+w=0 az + by +ez+dw=0 (a — d)? (b — ¢)? (aw + yz) + (b — d)? (c — a)® (yw + 22) + + (¢ — O(a — bzw + zy) =0, then z os y a Gh G—90=9 = G9 -a ew z w ~d=a d—H @—) (bc) (C—a) (@—8) * 40. 1° Let O— = tan z" 50. Show that if cos z = sete tO __ 608 (%+28) _ cos (2430) cease @ ¢ then ate b+d eects 51. Let 29 cO8a 2,,_ cosy tan@ tana coat 0 — cosp ? 08 = cosB tang tany Prove that tan? $- tan? f = tan? +. A. 52. Prove that if cos 0 = cosa cos B, cos @ = Cos a cos B, tan $ tan $=tan$, then . 4 t sic = (sea—!) (sea ')- 53. Let x cos (& + B) +cos (a — B) = xcos (B + y) + cos (B —¥) = = cos (y-}+ @) + Cos (p—a@). 62 Problems Prove that tana _ stan tay tan 4 o+n tang (a+y) tan + (+B) 54. Prove that if sin(@=P) cosa , cos(at@)sinB _ Gy sin(g@—a)cosp ' cos(p—b)sina and tan @tana | cos(a—f) _ tan @ tan Bp +728 (a+B) a then tanO= 7 (tanB--cota), tang= + (tan a —cot B). 55. Given n® sin? (« + B) = sin? a + sin? B—2 sina sin B cos (a —B). Prove that 56. Eliminate @ from the equations cos (a — 38) = mcos* 6, sin (2 — 30) =m sin’ 0. 57. Eliminate 0 from the equations (a—b) sin (6+ @) = (a+ d) sin (8—@), 9 g atanz—btan poe 58. Show that. the result of elimination of 0 and @ from the equations in sin sinB cos p= st sina > cos (8—)=sinBsiny tan? @ = tan? 8 + tan? y. 59. Eliminate 0 and @ from the equations asin? 0 + bcos? 0 = acos?p + b sin? y = 1, atan@ = btang. 60. Prove that if cos (0 — a) =a, sin (6 — B) = b, 5, Equations and Systems of Equations of the Second Degree 63 then a® — 2ab sin (a — B) + b? = cos? (a — ). 61. Solve the equation cos 3z cos* z + sin 3z sin? x = 0. 62. Solve the equation sin 2z + cos 2x + sinz + cosr+1=0. 63. Solve the equation 1—cos« tan? z =———— 1—sing 64. Solve the equation 32 cos’ z — cos 6x = 1. 65. Solve and analyze the equation sin 3z + sin 2z = msin z. 66. ‘Solve the equation | py 008.2008 (2z—a) __ . (+4) Sie—e)y =1+kcos 2x. 67. Solve the equation sin4 z+ cost z—2 sin ae +3 sin? 2x =0. 68. Solve the equation 2 logy a + logax @ + 3 logge, a = 0. 69. Find the positive solutions of the system vay, yo (a@>0). 70. Find the positive values of the unknowns z, y, u and v satisfying the system woyt=a*, ulvP=a", wvt=b, uv*=c (a, b, ¢>0 and p?—g 0). 64 Problems 6. COMPLEX NUMBERS AND POLYNOMIALS We proceed here from the assumption that the principal operations with complex numbers (i.e. addition, multipli- cation, division and evolution) are already known to the reader. Likewise, we take as known the trigonometric form of a complex number and de Moivre's formula. In factoring polynomials and solving certain higher-degree equations an important role is played by the so-called remainder theorem (stated by the French mathematician Bézout), usually considered in textbooks of elementary algebra. Let us recall it: if f (z) is a polynomial in z and if f (a) = 0, then f (x) is exactly divisible by z — a. Hence (assuming that the polynomial. has one root) follows the possibility of resolving an nth-degree polynomial into n, equal or unequal, linear factors as well as the following proposition used here repeatedly: if it is known that a certain nth-degree polyno- mial in vanishes at n + 4 different values of z, then such a polynomial identically equals zero. Consequently, if two polynomials of the nth degree ni z attain equal values at n+ 1 different values of z, then such polynomials are identically equal to each other, that is, the coefficients of equal powers of x coincide. Finally, let us mention the relationship bet- ween the roots of an nth-degree equation and its coeffi- cients. Let the polynomial a pet t pot? 2+... + Pret pn 7 have the roots 21, Z2,.. ., Z,, so that there exists the facto- rization : 2b py) pyt? +... + pn = (4@— 24) (U— 42). - (4 In). We then have the relations: my+t2+... +2 = —p, UyEq + ayy +... + 2yt_ + FQt3 +... + pat, = Pr ~ ayrgtg +... + Spent, = —Ps 6. Complez Numbers and Polynomials 65 1. Let z and y be two complex numbers. Prove that l2+yP+le—yP =2 {| 2 + ly}. The symbol | @| denotes the modulus of the complex num- ber a. 2. Find all the complex numbers satisfying the following condition The symbol z denotes the number conjugate of x. 3. Prove that VGit ay... FanP Orbs. Fon SV a+b +V@+h+...4Va+e, where a; and 6; are any real numbers (i = 1, 2, 3,..., 7). 4, Show that (a+ b +c) (a + be + ce’) (a + be? + ce) = =0 +B +46 — 3abe t if e@teti=0. 5. Prove that (a? + b? + c® — ab — ac — be) x X (+ y? + 2 — zy — az — yz) = : = X?4 Y?4 2— XY —xZ—YZ i X =axr cy bz, Y = cx + by + az, Z = be + ay + cz. 6. Given tty +z =A, z+ye +2e=B, xtye+ze =C. 5-125 66 Problems Here and in the next problem ¢ is determined by the equa- lity e+tet+i=0. 4° Express x, y, 2 in terms of A, B, and C. 2° Prove that [AP + 1 BP + 1 CP =3{/2P + ly? + [zP}. 7. Let Asatytz, A’aa'tyt+2', AA =a"+y'+2", Ba=artye+ze, Bo=a2'+y'e +28, BB=2"4+y'e+2'e%, CHatye+z, C=a2'+y'e+z'e, CC =2"+y"e?+2"e. Express x", y" and 2” in terms of a, y, z and 2’, y’, 2’. 8. Prove the identity (ax — by — cz -- dt)? + (ba + ay — dz + ct)? + + (cx + dy + az — bt)? + (dx — cy + bz + al)? = H(A +B +e? + a) (+ y? +24 0%). 9. Prove the following equalities ae Serb 1— (5) tanto +( 7) tante— wee tA, where n (-1)? tan"@ if n is even, eet fea Saree ete! A=(—1) ey tang if n is odd; o sin n n\,, n Tare = (7) tamg—(§) tantot (5 tanto +... A, where ' Nee A=(—1)? (4) tan"1@ if n is even, n- A=(—1)” tan"@ if n is odd. 6. Compler Numbers and Polynomials 67 Here and in the following problems n\_ pk a(n—1)...(n—k41y eee 3 oer ae 10. Prove the following equalities lemme 2m 1° 2°" cos*™ z= > 2( )cosaim iy ( ); a k m hom 2 2"sin'™2= >) h=0 2m (aye ( k ) cos 2m — hy + a) k=m 3° 2’ cos™ x — SD} pe ‘ cos (2m — 2k + 1) 2; k=0 & aoe 2m+1 4° 2" sin mt (—4m( es ) sin @m—2i- +1). k=0 14. Let Un = Cosa --r cos (% +0) +7? cos (a+ 20)+ ....4 +r” cos (a+ n6), vn =sina-+rsin (a +6)+r* sin (@+ 20)+...4+ +r" sin (a+n6). Show that cos & —r cos (a —0)—r"*1 cos [(n-} 1) 0-+ ae] + rt? £08 (nO) 1—2r cos 0-f 2 vp = Si &_rsin (a0) —r"44 sin (m+ 1) 0-+ a] +r"? sin (0+ +a) ie 1—2r cos 0-72 Un= 12. Simplify the ana sums zn 1° S=1+nceos04+74 02 )) cos 20+ ...= kon =>) Cheosko, (C2 =1); k=0 68 Problems hon 2° S’=nsinOd+ h=0 13. Prove the identity sin?” @ + sin?” 2a -+ sin?? 3a 4+ ...+4+-sin??na= 4 =ytn if aa and p< 2n (pa positive integer). 14. Prove that 4° The polynomial x (x"-! — na") + a” (n — 1) is divi- sible by (x — a)*. 2° The polynomial (4 — x”) (4 + x) — 2nz™ (4 — x) — — nz" ({ — 2)? is divisible by (1 — 2). 15. Prove that 41° (w+ y)"—a"—y" is divisible by ay (x + y) x X (2? + zy + y?) if n is an odd number not divisible by 3. 2 (x + y)"—a"—y” is divisible by ay (x + y) x x (x? + zy + y*)? if nm, when divided by 6, yields unity as a remainder, i.e. if m = 4 (mod 6). 16. Show that the following identities are true @ + yp — 2 — y = day (@ + y); 2° (w + y) — ab — y® = Say (x + y) (2* + ay + y%); 3° (a + y)? — a? — y? = Tay (2 + y) (@* + zy + y?)* 17. Show that the expression (@+y+2"—2™ ~ y™—2™ (modd) is divisible by (@+y4+28—2-—y—2. 18. Find the condition necessary and sufficient for «* + + y3 + 2 + kayz to be divisible by x + y +2. 19. Deduce the conditon at which x” — a” is divisible by 2” — a? (n and p positive integers). 20. Find out whether the polynomial at + 2 4 + a? + a (a, b,c, d positive integers) is divisible by O+24+24+1, 6. Complec Numbers and Polynomials 69 21. Find out at what n the polynomial 1 + x? + 2# +... +2" is divisible by the polynomial 1+ 2+ 2? +t... ta"), 22. Prove that 4° The polynomial (cos @ + xz sin )" — cos np — — x sin nq is divisible by 2? + 1. 2° The polynomial x” sin @ — p"™! x sin np + + p"sin (x — 1) @ is divisible by 2? — 2pz cos » + 0%. 23. Find out at what values of p and q the binomial x4 + 1 is divisible by 2? + px + q. 24. Single out the real and imaginary parts in the expres- sion Va + bi, i.e. represent this expression in the form a + yi, where z and y are real. 25. Find all the roots of the equation feceni 26. Find the sum of the pth powers of the roots of the equation az" = 1 (p a positive integer). 27. Let €= cos 24 isin ms (na positive integer) and let Ay =a + ye + 2e* +... + wer? ® (k =0,1,2,...,2— 4), where x, y, 2,..., U, W are n arbitrary complex numbers. Prove that hen-1 , 4eP antler tly Pt leP+ +1 P} (see Problem 6). 28. Prove the identities h=n-1 1° 2™f=(@—1) > (2220s +1); k=i kon 2° at 4 = (eA) [Jf (x? 22008 wy +1); k=1 70. Problems kon 3 ot 4 = (+4) T] (242i cos st +4); kat hon-t £2r+i= TJ (2?—2c0s PED +4), 29. Prove the identities 4° sin L sin & ... sin CV" — 7 2 cos cos te SE egg teeta 4° Bap 8 Batt ++ 8 SFT a if n is even. 30. Let the equation 2" = 1 have the roots 1, a, B, y,-.-, Show that d@—a4—p)(1—y...d—N =n. 31. Let Ty, lq, . + 0 Ly be the roots of the equation atom +...4+2e24+1=0. Compute the expression 4 ae + 32. Without oS the ae * tte was +e ga =1, y 2 Sttetaeat 2 Stptetpect find ery t 2 6. Complez Numbers and Polynomials um 33. Prove that if cosa +i sina is the solution of the equation a pat +... + pn = 0, then p; sin ap, sin 2a+. ..+p, sin na =0 (py, Pa. -- , Pn are real). 34. If a, b,c,..., & are the roots of the equation at py? + pot? ? +... + Py at + Pa = 0 (Ps, Po, . . ») Pn ate real), then prove that (+a A+ d)...@+h) 5 =(1—petpe—---)?+(pi— ps +ps—.---)? 35. Show that if the equations @ + pr+q =0 2+p'a+q' =0 have a common root, then (pq — ap’) (Pp — P'P = (9-7. 36. Prove the following identities : V cos my V cos V cos SK 45-377); ey cos a V cos 4 V cos 2 = V 4079 37. Let atb+c=0, Put ak LU 4 chy. Prove the following relations (see Problems 23, 24, 26 of Sec. 1) 2s, = 82, 655 = 5sy83, 6s; = 7s38,, 10s, = 75955, 25s783= 218;, 50s? = 495,82, 1 Sn43 = bCSn + S28n44- 72 Problems 38. 1° Given z+y uty, Ptypawtov Prove that a fy" =u" $v" for any n. 2° Given gcty+z=ut+v+t, gtyp+Pawpre r+pLe, P+¥+H8=—U4U4 6. Prove that a ty tataum poten for any n. ‘ 39. Let A =a + xe + 238%, B= 2, + x92? + ze, where : eteti=0, and 2, £2, 23 are the roots of the cubic equation P+ pe+q=0. Prove that A? and B® ate the roots of the quadratic equa- tion 2 + 27qz — 27p> = 0. 40. Solve the equation it (@ +4) (t+ b)(@+e)(e+a=m atb=c+d. 41. Solve the equation (c + a) + (x + dB) =e. 42. Solve the equation (@+b+e(@ tate (@+at+b)(atb+o— — abex = 0. 6. Complex Numbers and Polynomials 73 43. Solve the equation x + 3ax* + 3 (a? — be) z + a + B® 4 C8 — Babe = 0. 44, Solve the equation az* + ba + ca*9+dzr+e=0 if a+b=b+e+d=d+e 45. Solve the equation (a+ b + 2) — 4 (a + B8 + 2) — 12abz = 0. 46. Solve the equation a2x2 @+ apa =m (2 and m>0). Deduce the condition under which all the roots are real, and determine the number of positive and negative roots. 47. Solve the equation (54 4 1022-1) (Ba + 1002-4 1) (8+ 1022+ 1) (a8 10025) 48. Solve the equation 4 yx aga” 4 \ t+ slat Gaye a) + Gay ea era Ft : yyq.Z2m-t _ 2pxm— p2 Yea) Ea) (Fam) Ea)» am)” 49. 1° Solve the equation e+ pe +qe+r=0 if ci = ryx3. 2° Solve the equation P+ pe + qr +r=0 if ay =a. +25. 50. 1° Solve the system Pte +a = 3ayz 2+ 23 + D = 3ber P+y+e = decry. 14 Problems 2° Solve the system t—a=yi—b=24-c = ut —d = zyzu ifat+b+c+d=0. 51. In the expansion 1 + (4 +2) +...+(4+ 2)" in powers of z find the term containing 52. Prove that the coefficient of-z* in the expansion in powers of x of the expression {(s — 2) 2? + nx — s} (x+41)” is equal to ney, 53. Prove that for c>1 pat— gz? —p+q>0 (p,q positive integers and q > p). 54. Let x and a be positive numbers. Determine the greatest term in the expansion of (z + a)”. 55. Prove that 4° mie —1yn 4 ED Gay y+ (ty =0 ifi>m. 2° m™— m (m— 1+ AED (may 4 + c —1)"tm=m! (i and m positive integers). 56. Prove the identity (24 a)" = {2 —C22™#a? + Che 4a — .. +{Cha*41a— Chea +... 57. Determine the coefficient of x’ (L=0, 1, ..., 2n) in the following products 1° (1+er424...4 2%} (44 r42°24+ ...42"}; 2° Atete+...+2"} f—24+2?—234...4- +(—1)" 2}; 3° (A+ 204 327+ ...4+(n+1)2"} (14-2e432% + ...4 +(n+4) 2%; 4 (A+ 204 322+ ...+(n+1) 2") (1-22 + 327— ... + +(-1)" (n+) 2", 6. Complez Numbers and Polynomials 75 58. Prove that PALCR+CR+ --. = Cat Cab... = 205 2 Chat Cint.- + Cin! 3°14 Cnt s+ Cin = 59. Prove the identities rctc+et+... 2n-2 if n is even; DiC CMC hue 3B CR+CR+ CR +... 60. Prove that n PCR+ Chet... (2°142? cos =); 2 CHEETA 4 = y (2242 sin E); poh 3° Ch+ CR + C+ = rr (2 —2° cos ); ry Cr+ Cht ON + ...= nl (2-1-2? sin ). 61. Prove the equality QPL mt =Chgs +2 (Cat Cnt +. +03). 62. If a, a2, a3 and a, are four successive coefficients in the expansion of (1+ 2)" in powers of z, then ay a3 an aay! Gs+ a, Oy ay * 63. Prove the identity 4 1 4 4 Qa-1 Tair + aay tao tb Roar =a (n even). 76 Problems 64. Find the magnitude of the sum s=Cr—3Cn + PCR BCL +... 65. Find the magnitudes of the following sums o=1-C4+C4—C84+..., 0 =Ca—CatC,—Cat.... 66. Prove the identities 4° CLA QChA BCR AACR +... (m1) Ch = (nm 4-2) 2°; 2 cL 2071-303 +... +4(— 1)" nCh =0. 67. Prove that 2. (—1)"t an sented Cease = a= TET’ 68. Prove that ea 1 nn nett PdtzCat gent ..-4 wa = 3 cae Cn 23c2—-24c3 aniCh —gntt_t 2 2Cn Sore hep ececan ore 69. Prove the identity 2, ft —1yn-2 1 Chae Cnt Zap $M Chart Sage. et. 70. Prove that Crt Cngit Cnye+ ..- are | 2° Ch—Ch4-CR+... +(—Ayh oh = (Ay Ch. 71. Show that the following equalities exist 1° CACRACKOR +... 4+08CL = Chains 2° cacn+-chontt+ ... penton = 72. Prove the following identities 1° (Ca)?-+ (CAP (CP +... + (CRP = Chas 2 (Can)? — (Cin)? + (Cin)? — «+ (C38) = (—1)" Chas 2n! 6, Complec Numbers and Polynomials 717 3° (Cats)? — (Cingt)? + (Conga)? --- — (Cant = ou n2y9 myo 2n—1)! 4 (CLP +2 (C24 ee 73. Let f (z) be a polynomial leaving the remainder A when divided by x — a and the remainder B when divided by « — b (a b). Find the remainder left by this polyno- mial when divided by (2-— a) (x — 6). 74, Let f (x) be a polynomial leaving the remainder A when divided by x — a, the remainder B when divided by a — b and the remainder C when divided by z — c. Find the remainder left by this polynomial when divided by (x — a) (x — b) & —c) if a, b and ¢ are not equal to one another. 75. Find the polynomial in x of degree (m — 1) which at m different values of z, io q,.. +, 2m, attains respecti- vely the values yi, Yo, - - -; 76. Let f (x) be a polynomial leaving the remainder A, when divided by x — a;, the remainder Az when divided by z — az,..., and, finally, the remainder A,, when divi- ded by z — am. Find the remainder left 7 the polynomial, when divided by (x — ay) ( — a2)... (& — am). 77. Prove that if 2, 22,.. +, 2m are 7 m different arbitrary quantities, f(z) is a polynomial of degree less than m, then there exists the identity (z—~ag) (t—23) 1) = (2) aM (22) (e—a9) « +4 2) ey aa) +1 Gm) —24) m2) 78. Prove that if f (x) is a polynomial whose degree is less than, or equal to, m — 2 and x, %,..., 2m are m arbitrary unequal quantities, then there exists the identity F(x) f (z2) Bo G A) mam) + Gay Gam) em fm as et (@m—21) @m—22) «-- (Zm—*m-1) =O. 78 Problems 79. Put a ah at Sn Gay ee) aan) | Gay ay ee ag) T Zin FPG) ema) im Emad (21, Z, ..+, 2m are m arbitrary unequal quantities). Show that s.—0 if Om. 80. Compute the following = =" =” $0 = Gayman | Gm Taal zm si FAG Say ee ce Gm ate) MAE 28. 81. Show that if f(z) is a polynomial whose degree is less than m, then the fraction £@) (=) (t— 2)... (2m) (21, 22, . . +) Cm» are arbitrary quantities not equal to each other) can be represented as a sum of m partial fractions rn za r—2 z—Im where A;, Az,..., Am are independent of z. 82. Solve the system of equations atnt. tet! ay—bp 4 a 3% Ch +35 vee. 6. Complez Numbers and Polynomials 79 In particular, 1 2m 38a 4a We ETT Ate HE oe 84. Prove the identity ny (44 —b5) (ay Br (b= bs) (ays) (aa—2 aba b)- (ay—bn) Bia biy On 85. Prove the identity (+B) (enB) (@—B) - @—nB) as (ner) (a8 1%) (n? He _gyner (nb r) (n®§—12) (x2 ah 4)" roe 86. Given a series of numbers Co, C1, C2, -- +) Chy Chtiy ++ 0° Put Ac, = cz45 — ¢y, So that using the given series we can form a new one Aco, Acy, Acg,.. 6s We then put Are, = Acar, — Ac so as to get one more series: A®c9, A%c,, A%co,... and so forth. Prove the following formulas 5 n n(n—1) ao 1 Chen = Ca + Aen + 12 APcy + —1)(n—2) ef + nn) Arey + see FAR; OAR, a1 2 AM, = Cran ——E Chant FS Chena tee + (—N"en- 80 Problems 87. Show that if f (x) is any polynomial of nth degree in z, then there exists the following identity 1 (a)=F (0) + ZAf0) +2459 ayy... + tent stent) A'Y(0), where Af (0), Af (0), . . ., Af (0) are obtained, proceeding from the basic series: f.(0), f (1), f(2),... . 88. Show that if aa Ay + At (e—1) + 4B (eA) (@—2) HE + AB (w—1) (2-2)... (@—n), 4 then A,=(s-+1)"—Cls" + C2(s—4)"-+ 2... +(—4)' CsA 89. Prove the identity nl 4 z(@+1) aaa (eteeit: tiaa}= at cho ay wa ari terme HO eae 90. Let Qn (at) = 2(x—1)(e—2) ... (2—k +1). Prove that the following identity exists Gn (EY) = Pn (X) + Cn Pn-1 (2) G1 (Y) + Cn Pn-2 (2) Ga (Y) + oe + + CR; (2) Ons (Y) + On (y)- 91. Prove the following identities apy ph pray SD pags. (anys Sa 4 OMI pages +4 yt ee H(A Cp rg + 6. Complex Numbers and Polynomials 8t where PHITY, G=xy. 92. Let z+y=1. Prove that aA ACY + Cyt 6. HOM YT) + +y"A+C,ae4- ... +CR a") =1. 93. Prove that the following identity is true 4 Ch 1 1 1 wears Gone (aoa tape t Cha ae + Gramt eae + 1 1 Ch +a { e—bat GoH@i~@sh te ym Cinta +e cami} 94. Show that constants A,, Ag, A; can always be chosen so that the following identity takes place (ety) =a" ty" Agry (2 $y) + HAs y (24+ y™4) + 00. Determine these constants. 95. Solve the system My+%_ =a TY 1 TaY2 = a2 2 2 Tayi + 2y2 = as a 3 TAY, + TaY2 = %- Show how the general system is solved Mey agt+ ... + onitm=a4 (4) TYitLYat +++ + Lryn= ae (2) yi tryst --- + tnyn=ads (3) ayy + rays (2n) 61226 82 Problems 96. Solve the system atytetut px+-qy+rz+su+tv=3 Pat gy +r%2+ stu + Pv = 16 Peat q’y +r3z+ su + tv = 31 Piet gtytrizt stu + tv == 103 pix 4+- gy +r'2+ seu + tv = 235 pea + g®y + 182 + s8u + thv = 674 pict qg’y+riz+stu+ tv=1 669 Pec + g’y +182 + stu + t8v = 4526 peat gy +182 + su + tv = 11595. 97. Let m and p be positive integers (u<2 an integer) can be represented in the form of a sum of n successive odd num- bers. 8. Let the sequence a, a2,...,a, form an arithmetic progression and a, = 0. Simplify the expression that “= i eee ). 41 Safty... tf a (L4z as Tat 9. Prove that in any arithmetic progression 4, G2, a3,--~ 7. Progressions and Sums 85 See VanatVan Vart+ Van 10. Show that in any arithmetic progression Gy, Az, A3,..~ we have S= Haipat—at+...$ah.,—ah 5 11. Let S (x) be the sum of the first n terms of an arithme- tic progression. Prove that 1° S(n +3) — 3S (n + 2) + 38 (n + 1) — Sq, = 0. 2° S$ (3n) = 3 {8 (Qn) — S(n)}. 12. Let the sequence aj, dy,...,@n,Qn41,--- be an arithmetic progression. Prove that the sequence S,, Sj, S3,..., where Sy =a +ag+... +4, So = anti t.-- + Gon, S3 = Gens, tees + dans ee is an arithmetic progression as well whose common diffe- rence is n? times greater than the common difference of the given progression. 13. Prove that if a, b, c are respectively the pth, gth and rth terms both of an arithmetic and a geometric progres- sions simultaneously, then a2.p'-4 ch 4, 14. Prove that (1+a24+2?+...+2"f—2"= =((1+2+24...+2™) (142424 ...42™). 15. Let S, be the sum of the first x terms of a geometric progression. Prove that S, (Ssp — Sen) = (Sen — Sn)?- 16. Let the numbers aj, a2, @3,..-. form a geometric progression. 86, Problems Knowing the sums Saatatat eta, Sat+iy.tt, ay a find the product P= aya, .. 417. If a,, a,, ..-, dn are real, then the equality (@-+at+ ... +43) (@tad+ ...-+a3)= = (Aid -+ G9Q3 4 --. + An—1dn)* is possible if and only if a, a, ..., @, form a geometric progression. Prove this. 18. Let a,, a2,...,a, be a geometric progression with ratio g and let S, = a, +... + dm. Find simpler expressions for the following sums 1° Spt Sat +--+Sni 0 4 4 4 Cte are ee 19. Prove that in any arithmetic progression, whose common difference is not equal to zero, the product of two terms equidistant from the extreme terms is the greater the closer these terms are to the middle term. 20. An arithmetic and a geometric progression with Positive terms have the same number of terms and equal extreme terms. For which of them is the sum of terms grea- ter? 241. The first two terms of an arithmetic and a geometric progression with positive terms are equal. Prove that all other terms of the arithmetic progression are not greatet than the corresponding terms of the geometric progression. 22. Find the sum of n terms of the series Sy, = 1-2 + 22? + 323 +... + nz 23. Let a, a2,...,@, form an arithmetic progression and uw, Ug,...,U, a geometric one. Find the expression for the sum $= ayuy + Agua +... + ayy. 7. Progressions and Sums 87 24. Find the sum 4\2 1 \2 cgay (+5) +(@+a) +--+ (+z) 25. Let S,= t+ 2*4 384 ...4n". Prove that t rn 4) (2n +4 3 4)? s,=20t9 Us $= R EN ery S=eee 26. Prove the following general formula (Hef) Sp +-EED* 5, PEO EEED Set + (E+ A)S1+ So= (n+ 1) — 27. Put th 2h 4 1. 4n*=S)(n). Prove the formula Sp (1) = Srst (M) + Sp (M—1) + Sa (M—2) +--+ : + Sp (2)+ Sa (1). 28. 1° Prove that Pa 2h43h4 0. tnt An*14 Bn*4+Cnt4...4Ln, i.e. that the sum S, (n) can be represented as a polynomial of the (k + 1)th degree in n with coefficients independent of n and without a constant term. 2° Show that A =a and Bat. 29. Show that the following peal take place 4g, = RlmstA) (2m) (Bn + Bn —1) ‘=D 5, = (n+AP? (An? + 2n—1) = Sqm OME BAM 2nd —Tn8 on 2 _, M (m+ A) (2a) [3n2 (n-4-1)2— (Bn? + 3n—1)] ee eee 88 Problems 4g, — Bb 12? tins Tab Bn ee _ nr? (n+ 4)? [3n? (n+ 1)? 2 (2n? + 2n—1)] eee] geese aree eee aE 30. Prove that the following relations take place S3=S}?, 48}=S3+4+3S;, 258;+S3=35}, S,+S,= 28}. 31. Consider the numbers By, By, Bo, Bs, By, ... deter- mined by the symbolic equality (B+ 4/4 — Bet =k +1 (k = 0,1, 2,3,...) and the initial value By = 1. Expanding the left’ member of this equality according to the binomial formula, we have to replace the exponents by subscripts everywhere. Thus, the above symbolic equality is identical to the following common equality Bass }ChatButChatBaat +--+ Che1By + Bo—Bay=k+1. 1° Compute Bo, B;, By, ..., By with the aid of this equality. 2° Show that the following formula takes place 1h4-2F4 384 0 tnt = Che Bin + Ch Bah. Ch Ban}, 32. Let 2, 22, ..., Z, form an arithmetic progression. Tt is known that B+ a+...+4n=a, a+ap+...Fah=0. Determine this progression. 33. Determine the sums of the following series 1° 144249274... 4-n?2™; 2° 134 232-4 3827+ ...4n32"1, 34. Determine the sums of the following series ° 3 5 7 2n—1 Pitstptyt + pe : Based ny 2 1-S+q—gte +(-)" = 7. Progressions and Sums 89 35. Determine the sums of the following series 1° 4-243—-44...4(—1)"3n; 2 222439. (Nyy 31-347. .—(4n—1) 42 2143-24 veka EA) rt, 36. Find the sum of n numbers of the form 4, 11, 111, 4411, ... 37. Prove the identity gina yanea = {a1 Dame 4 gem 3yd | (1) Qaryenye = {yr — 2yt tg 4 Dyen-Bgd_ | L(A)" Qy any, 38. Find the sum of products of the numbers 14, a, a, ..., a", taken pairwise. 39. Prove the identity (e+) 42 (24s) +. +) (2+ pei eS an—1\2 “grt (FS 40. Prove the identity ° ; 1 1, 0 potaataat stages to nyt? of 2 1 2 rratesat +a eEpwrD = Eeceea (ate 1 i =a (z-apners)! o 4 2 A 8° trstasat + gop ehoeds _ n(n ty =F On+h On+s)" 44. Compute the sum n4 3h gate t + (2n—1) Qn+f) * 90 Problems 42, Let a4, a2, .-., Qn be an arithmetic progression Prove the identity 4 4 feecee. 1,4 1 a4an + tina tS t Gna, tan (a : a tite): 43. Prove that 4° n+4 n+p 4 1 wo tape tet orem “ai Weer 2 apn teem t--taapar< 4 4 Zz 4 b and b>c, then a>e, 2° If a>b, then a+m>b+m, 3° If a> b, then am > bm for m > 0 and am < bm for m <0, i.e., when multiplying both members of the inequa- lity by a negative number, the sign of the inequality is reversed. 4 Ifa>b>0, then a > d* ife>0. This last inequality is readily proved for a rational z. Indeed, let us first assume that x = m is a whole positive number. Then a” —b" = (a—b) (a4 a4... 4 bm), But either of the bracketed expressions on the right exceeds zero, therefore a*—b">0 and a™>b™. We now put z=. Then a*—b'=9/a—/ 0b. We have (a—) = (9/2 5) (TT. EPA, Hence, actually, it follows that a—V/b>0, ie. Ya>VYo. 94 Problems Let, finally, oat. We have P P a’ — b= at 09 = VY PY BP. But a? > b? (as has been proved), consequently, ~/a® > > ¥ &. To prove this inequality for an irrational x we may consider x as a limit of a sequence of rational numbers and pass to the limit. 5° Ifa>1 and e>y>9, then a* >a"; but if0O< y>0, then a* 1 ifa>0 anda>1 and can be obtained from 4°. 6° logs > loguy if z>y and a>1; and logsx< < log, yife>yand0F (n, @ positive integer). 2. Let n and p be positive integers and n >1, p>1. Prove that 1 4 4 4 apt wapet Spe + ae te tae < 1 4 0) and the angle C is obtuse, then tan A tanB< 1. 10. Let tan®@=ntang (n> 0). Prove that tan* (n—12 4n 11. Show that if 4 cosa cosp t ane tanB=tany, then cos 2y<0. 12. Let us have n fractions OAH essere z, b>0 (i=1,2, ...,7). « aytagtes tan ; : . Prove that the fraction Rtbt th, §8 contained bet: ween the greatest and the least of these fractions. 13. Prove that ™*"*--+?/ab...1 is contained between the greatest and the least one of the quantities Vad... Wh 14. Suppose 0 0). Prove that a>yh +e ifA>2, a (a + b —c) (a +c — b) (b+ —a). 18. Let A+B+C}=n. Prove that tan? 4+ tan?-2 + tan? 4. 19. Let A+B+C=n(A,B,C>0). Prove that in4 B Greed sii zsinzsing< zy . 20. Given A+B+C=2n(4,B,C>0). Prove that 4° cos A+ cos B+-eosC<3; 3V3 o A B 2 cos-4-cos-$ cos <3 VE 3 2 24. Prove that V@FoC+)>Vab+Ved (a,b,c and d>0). 22. Prove that a8-+-b3 b S(4F)’ @>0, 6>0. 8. Inequalities oT 23. Prove that vr sya (a 6>0); 1 (a—b? bb — 4 (a—o2 . oe Pee ce Vb g o™ it ade. 24. Prove that SRT! SY abe (a, 6, e>0). 25. Prove that Vaan + Vast +V dueitn <*S* (ay + ay +... tan) (a: >0; i=1, 2, ..., n). 26. Let a; >0 (i = 1,2,...,”) and aa,...a, =1. Prove that (1 + a) (1 + a2)... + ay) > 2". 27. Prove that 1° (a+-b) (a+¢)(b-+¢)>8abe (a, b, c > 0): a b c 3 2 bye ape a45=T° 28. Prove that VOTH OFD EC Hm) SV abe + / kim (a, b,c, k, 1, m>0). 29. Prove that 1 4 1 3 atet sea (a be>0. 30. Prove that Bbeeb ben / Fa en (a> 0; i= 1, 2... 0), the equality being obtained only in the case ©, = y=... = Lp. 31. Let a,, ag, ..-, da form an arithmetic progression (ai > 0). 71225 98 Problems Prove that V aan 1 + ad (a is any positive number; 4 > 1 is rational). 2d +-ar< ch (a> 0 real, 4 rational and posi- tive, a < 1). A. Let u, = (1+4)", n is a positive integer. 1° Prove that Upti > Un. 2° Prove that u, isa bounded quantity, i.e. there exists a constant (independent of n) such that u, is less than this constant for any n. 42. Prove that V2>V3>V4I>V5>V6 >... >Vu> >"WnFi>.... 43. Prove that 2>VE>VA>V5>...>"VWan>Vnti>.... 44. Let us have yy + Agnte +... + Aint = YW Gyyty PF Agate +... + Gen = Yo Gy + Anat, +... + nnn = Yas where a;; > 0 and rational, x;; > 0. 100 Problems Furthermore, it is given that ay + dae +... + ain = 1, Per eee ey ee Prove that WY2 ++ Yn SUly.- . Eye 45. Let a,>0,b:>0 (6 =1,2,...,n). Prove that W (ai + by) (da Fa) «(nF On) Gay + ay + 4 W/ Diba. One 46. Prove that rfaetee+enyh_ obtahy tah (ites) cnt Fin 7 ’ nand k are positive integers, x; > 0. 47. Let the function (é) defined in a certain interval possess the following property tt te @ (t1) +9 (ta) o(*)< 7 for any two t, and f, not equal to each other. Then tyttet... ttn @ (ts) +. (ta) +. + P (tn) g (AoE) x , where ¢, t2,...,¢, are m arbitrary values from the given interval not equal to one another. 48. Find the greatest value of the sum S =sina, + sinag+...+ sina, a,;>0anda+a,+...+4a, =n. 49. Let x, p and q be positive, p and q being integers. Prove that é wP—t | 29-1 > P q | =...=m Ht. ] 8. Inequalities 101 if p>q (« #1). 50. Let z > 0 and not equal to 1, m rational. Prove that ma™* (zg — 1) > 2" —1 > m (x — 1) if m does not lie between 0 and 1. But if 0 —1). 52. Prove that 4 sak ) P <( 2] +094 n seas n q=pD, both q and p being positive integers. 53. Find the value of x at which the expression (x — a)? + (@ — a)® +... + (@ — a) takes on the least value. 54. Let 2 + 22 +...-+2, =C (C constant). At what 1, 2_,...,%, does the expression 22+ 224+... + 2% attain the least value? 55. Let 2,;>0 (6 =1,2,...,m) and 2% +224 see ta He, At what values of the variables 2,, 22,..., 2, does the expression : titaht ah (A rational) attain the least value? - 56. Given z;>0 (i = 1,2,...,m) and the sum x, + +2,+...+2, =C =const. Prove that the product @%Z_...2, Teaches the greatest value when 2, = 2, = 102 Problems 57. Given xz; >0 (i =1,2,...,m) and the product T%L_03... L, is constant, ie., 42%2...2, = C. Prove that the sum 2 + 22 + +2, attains the least value when ==... == VC. 58. Let 2, >0 (i ,2,...,m) and the sum x + + a,+...+2, =C = const. Show that fteh? 2. alte takes on the greatest value when fitediig Heese ts cir pes tent Hn Wate Fon” uw; >O0 (i =1,2,...,n) and rational. 59. Let a,>0,2%;>0 (i =1,2,...,n) and Ayr, + ata +... + Ont, = C. Prove that the product 2,22...2, attains the greatest value when Cc QZ, = Oy%g=... = Ann =—. 7 60. Given a>0, 2, >0 and azt+ ark? +... tana" =C (A, >0 and rational). Prove that pak ane takes on the greatest value when Ayayzh! — Agagah? Ananth” “OB Leciareni ali 64. Let ziz}? ... ci” =C =const. Show that ayxtt + ark? +... + ana” 8, Inequalities 103 attains the least value if apt 2 a yyy aft nbn (ai, x; > 0; A; and p; > 0 are rational). 62. Find at what values of z, y, z, ..., t the sum e@+yt+At+.. +8 takes on the least value if ax+by+...+hkt =A (a,b,...,% and A constant). 63. At what values of x, y does the expression u = (ax + by + 4)? + (age + bey + on)? +... + + (Gut + bay + en)? take on the least value? 64. Let xo, 24,..., 2, be integers and let us assume ee ee Prove that any polynomial of nth degree x” + ayz"-! + +...+ 4 and n be a positive integer. Prove that ttt nod a"—1>n(a? —a7), 104 Problems 69. Prove that n 4 4 1 gsttgtyt tm <” (n a positive integer). 70. Prove that (a, b, c, d>0). 9. MATHEMATICAL INDUCTION This section contains problems which are mainly solved using the method of mathematical induction. A certain amount of problems is dedicated to combinatorics. 1. Given Unt1 = BV, — 2p and vy =2, 4 =3. Prove that v, = 244, 2. Let Unts = 3Un — 2uy4 and Uu =0, uw =1. Prove that u, = 2" —14. 3. Let a and A > 0 be arbitrary given numbers and let and(er4), andlatt)rn om 1 _ A =7 (amit). Prove that wo VA_ (m= Vaye ant VA NatVA 9. Mathematical Induction 105 for any whole n. 4. The series of numbers G9, MH, Az, « is formed according to the following law. The first two numbers dp and a, are given, each subsequent number being equal to the half-sum of two previous ones. Express a, in terms of ap, a, and n. 5. The terms of the series 4, Az, Ag, ..- are determined as follows a, = 2 and a, = 3a,4 +14. Find the sum a +a,+... +a. 6. The terms of the series 4, 42,... are connected by the relation a, = ka, +l(n = 2,3,...). Express a, in terms of a;, k, l and n. 7. The sequence a;, d2,... satisfies the relation a,4,; — — 2a, +a,4 = 1. Express a, in terms of a, a, and n. 8. The tern. of the series Qy, G2, d3,--- are related in the following way an43— 3@ns2+ 30n41—y = 1. Express a, in terms of a;, a2, a3 and n. 9. Let the pairs of numbers (a, b) (as, 54) (ag, ba). . + be obtained according to the following law a+b 7? atte, be Eh, Qe BE, ya BEN, 106, Problems Prove that 10. The terms of the series To, Yor Tits Yt» Ley Yar s+ are determined by the relations Ly = Ly + 2m SIN? Yn = Yu -t “b 2tty-1 COS* @. Besides, it is known that tp = 0, yy = cos @. Express z, and y, in terms of a. 11. The numbers Tos Ly Tay - + +1 Yor Yt» Yar - + + are related as follows Ty = Biya + BYn-ss (a5 — By #0). Yn = Vn 1 8Yn-1 Express z, and y, in terms of x, yo and n. 42. The terms of the series Lo, Ty, L2,... are determined by the relation Ty = Ay 1 + Pry —2- Express z, in terms of xo, 2; and n. 13. The terms of the series zo, 2;,... are connected by the relation — Pen at gtn-2 cn Ptrq@ : Express z, in terms of xo, 2; and n. 14. The terms zo, 41,22, ... are determined by the equa- lity = nai th = Rina ee ire Soren 9. Mathematical Induction 107 Express z, in terms of x and n. Consider the particular cases Trt Zpatt "Det? 8 Gy tS 15. The numbers: x, Qo, At, Gar ++ oy bo, by, bay. are determined by the following law +b; 2anb, ans =F, bn a) and by are given, and ay>b)>0. Express a, and b, in terms of do, by and n. 16. Prove the identity : q 1 1 4 1 tpt} Boat + Gappcm api tapet tar 17. Simplify the expression (Q—a—2)...@—2% 400-2 (1-2)... x x (1-2) +22 (1-23)... (4-2) 4 ...4 taka)... (Loa) 4... pa (t— 2) 4 2" 18. Prove the identity trig tiet + Se ES 19. Check the identity (L4+-2)(1+ 2°) (1+ 24)... (1+ 22") = =t4e4e4e34...422"1 20. Prove the validity of the identity 1 Sp ett Cee ry sy 12 Tee abe 4 oth (b- A) o. (e+4) (R44) (a@-+4) (O41)... (+4) (144) abe... skl abe... kl ie 408 Problems 241. Prove the identity btetdt+...tk+l 4 re a(aybyey... +h+D “Ta at wp erETS + apregesorery tet L 22. Let lt “ee oo « (4—g"2z) = Fp (2). Prove the ae 1+ Pr (2)— Fn (g2) = (1 —92) (1—g?2) ... (1—g"2). 23. Prove that he (han) (1a)... (anh) sy = 2 Tak = kat 24. Compute the sum Sn= 54 FG=Nt HESHO SD t HON aE (6 is not equal to 0, 1, 2, ...,—1). 25. Let Sn=a1+ (art 1) a+ (21+ 1) (@t fast... + + (a+ 1) (a+) ++ (@ntf) an, Prove that Sn = (a; +1) (a, +4)... (Qn +1)—4. 26. Prove the following identities: eon 1° Srey) e+) aay nth)... (+941); amt =n 4 0 4 4 4 2 2 er)... @tH @ a OY. FO) }- 9. Mathematical Induction 109 27. Prove the identity 44 1 4 4 (t-z-4)+(a-g-e) t+ 28. Let us have a sequence of numbers (Fibonacci’s series) 0, 1, 4, 2, 3, 5, 8, 13, 21, .... This sequence is determined by the following conditions Uns = Un + Unt and up=0, u,= 14. Show that there exist the following relations 1° tnsg = Uo + Ut uet...+un+4; 2° Unda = Ui t+ Ug + s+... + Wendi 8 Weng = 1 ua tug t + + Uans 4° Wang tL = ey — Unt Ug te + Mant = Uani 5° Uan-2 + A = uy Ug + Us— Mgt +t anti 6° untinsr = UEP Ua ++ Uni To Why = Walle + Uallgs -- + p Uae —stlans — 1) yay 8° UnsUnsa— Unlngs = 9 U3 —Unpiting = (— 4 10° uf — Upon sensing 4. 29. Compute the sum ee 30. Prove the relations 1° Ung pt = Uns + nlp} 2 ant = Unt Una BP Unt = Unlingt — Un-2en-1+ 110 Problems 34. Prove that up --u},,—ui_,)=Usn- nt mFS 32, Prove thal wa= SY Choate h=0 33. Find the number of whole positive solutions of the equation 2,+2, +...-+2, =m (ma positive integer). 34. Prove that the total number of whole nonnegative solutions of the equations a+2y=n, 22+ 3y=n—1,...,ne+(n4+ 1) y=1, (n+ ft) a+ (n+ 2)y¥ =0 is equal to n + 4. 35. Show that the total number of whole nonnegative solutions of the equations xt Ay =8n—1, 4¢+ 9y =5n—4, 9a + 16y = = In —9,...,Mx+(n4 1)? y =n (n+ 1) is equal lo n. 36. There are n white and x black balls marked 1, 2, 3,..., n. In how many ways can the balls be arranged in a row so that all neighbouring balls were of different colour? 37. In how many ways is it possible to distribute kn distinct objects into k groups, each consisting of n elements? 38. Tlow many permutations can be made up of un ele- ments in which the two elements a and b never stand side by side? 39. Find the number of permutations of n elements in which none of the elements occupies the original position. 40. In how many ways can n distinct letters be arranged in r squares (first, second,..., rth square) so that each square contains at least one letter (the order of the letters inside the squares is disregarded)? 10. LIMITS We take as known the concept of a variable and its limit, as well as the basic theorems on limits which are usually treated in elementary textbooks of algebra (the limit of a sum, product and quotient). Let us here remind the reader » 10. Limits 444 of one of the indications for a limit to exist: if a variable increases but remains smaller than a certain constant, then such a variable has a limit (likewise, a variable which, when decreasing, remains greater than a certain constant also has a limit). When dealing with an infinitely decreasing geometric progression and, in general, with simple infinite series, one should bear in mind that the symbolic notation uy tug+ugt...+u, +... denotes none other than lim (uy + u,g+...-+ u,) if nro such a limit exists. If there is no limit, then the series Uy + Ug + ug +... Up t..- is said to be divergent, and it is useless to speak of its nume- rical value. 1. Let x, =a" and |a| <1. Prove that lim 2, = 0. n-r00 2. Prove that n lim <-=0 nooo for any real a. 3. Find «. agnk aynk-1+ ... ap dim ynh-p Opp Bn (a #0, by #0). 4, Let 4 n3—14 as m4 2 Prove that limP,== . nc 5. Prove that lim pete = ca (k a positive integer). n-+00 6. Prove that tepoka... tnk n 4 lim {+ nh a rar haa ns 00 é 112 Problems (k a positive integer). 7, Let us have a sequence of numbers x, determined by the equality ty = Zn tnce and the values 2 and 2. Prove that . +2: lim zy =a ‘ n-+00 8. Let N>0. Let us take an arbitrary positive num- ber zo and form the following sequence Prove that limza=VN. nc 9. Generalize the result of the preceding problem for the extracting a root of any index from a positive number. Prove that if then 10. Limits 413 10. Prove that lim + =0 aoe i 41. Let ken . Si= 3 (VW 14+-2-1) k=1 Find lim Sp. ae 12. Let the variable x, be determined by the following law of formation 1. m=Va+Va, m=V at+Va+Va, Find lim zy. n-r00 13. Prove that the variable malta tag te tae hn has a limit as n — 00. 14. Let us be given two sequences Loy Ty, Zo, - ey Yor Vis Yar e+ + (o> Yo > 0), where each subsequent term is formed from the preceding ones in the following manner p= Enctt eect, Yn = VEnYna- 81225 414 Problems Prove that z, and y, have limits which are equal to each other. 15. Let Spsattqt@+... lal 1. 21. Find the sums of the following series 12 14204 3a%4 00. fna™4 o.; 2 144049024 00.4242; 3° 14232-3822 0. tne 4... ([z] <1). 10. Limits 145 1\n 22. 1° Prove that the variable w,~ (4-1 =) (n= 4, 2, -) has a limit. He 2° Denoting the Jimit un by e so that lim (4 +) = oa prove that 4,4 1 0 e=t+ ltygtqa3t t+ qeace ties O<0<1). 23, Let 0. sing Knowing that Dn aa =4, prove that 24. 1° Prove that the series e+ art fet ia +... (0a; <9) is a convergent one. 2° Prove that for any real number o (0 << o <1) it is always possible to find, and in the unique way, a; (0 Pi(t—p)= % pi— DB pi=np— Dd ph, np = Prt Pat.-.- + Pa Further ian izn Dp np— Bi (pi— pt p= é é i=n = pn — 3iU(pi— PP +2ppi— P= np — 3 (Pe— pr i=n i=n —2pd pitnp=np— Dd (pi— py—mp & i But np — np? = np (1 ~ p) = npg. Thus, we get Pigs + Pada +... + Prdn = MPG — (Ps =P)? — — (Pz — pi — ++» — (Pn — py. 34. Indeed : _ tf Qn—t)t4 | (2n—3)43 7 . ied T 3 (2n—3) -& G t mo rtaor =L(t4f+e+- tay). 32. 1° It is obvious that walt. tha =nt[t—+(4-1)+ +($-1)4--4(4-1)]= (g+gt-444): 91225 430 Solutions hen hon han 2 a= 4, nad HRW (A+). k= hat hat Hence, n ny = n+ ( T +7 t.t+ ott): 33. Add to and subtract from the left member the follo- wing expression 2(p+a+et. +a): We get togtagcat tare bmet)—(gtet- tar) stg) + +(etat- tar)? (gtat- tar) = attytatat. ter tarn (tatgt te) ssertaget ta 34. We have 4 1 1 (t4+-g44 ) (1- 2a—1 ) (1+ 3a—1 ) ae 1 1 (1+-ar=pesr) (1 ~ mar) = (2a —2)-3a ... (2n—1) a (2na—2) (a—1) @a—1) a—1) ... na—1) 1-a-3-a-5-a ... (2n—1)-a-(24—2) (4.—2) ... (2na—2) x = (NaN)... (na— ie Nan 2a—T...[(npaja—t] tea Beas (2n—1) a+(a—41) (2a—1) “(ez ta=ty san) at (o —1 eanect ss Qn—1)-@ ~Ta+ha—h..- (ea na—t | Solutions to Sec. 1 134 But wherefrom we obtain the required identity. 35. Leta <2 —93)3 _ (p3 —2g8)8 r+¢ (3 5 —P Rae EGS It is evident that the right member can be obtained from the left one by permuting p and q. Let us reduce the left member to such a form, wherefrom it would be seen that after the permutation its value remains unchanged. Then the validity Of-the identity will become clear. heb have ae Pt 8+ Cr) = rae (te — Pe, Solutions to Sec. 2 137 2. We have p+ 98 3 gece 6(pt+a) way pa T Taye ( ete ) pra pg = Rammed By rae bya ( a) = _ Pape 3 pea ee teen ct rape Tae (F ae ) = eee (pq)? pag?" (p+-a)? p™ (p+ 9? Pa® 4 A x {P= pa+-¢ +39} = sa 3. Grouping the last two terms of the sum, we get Orin Gree eo reedane wae pas Tp ge page —_ 29=P) (2 ey ___2(a=P) wear eng (P49 + 2pq) = (p+ 4)? pq” Adding now the first term, we find ap oad aap Eee ESAs Ea J (p+) pigt (p+)? p3q5 pags 4. We have to prove that tte A4ty t—2 T-y t-2 Replacing z by its expression, we find pas a +. Since y and z are obtained from x by means of a circular permutation of the letters a, b, c, we have tty 6 Soyer c a itz TT Hence, the required identity is obvious. 5. We have atbte+d e—b+e—d atb—c—d a—b—c+d“ 138 Solutions 4 A+B _C+D But iff=y> then 5-8 tp’ there exists the second of these equalities, then the first one exists as well. Reasoning in the same way (putting A=a+b+c+d, B=a+b—c—d, C=a—b+e—d, D=a—b—c+d, we find a+b a—b a+b cd wad nd Gb a and conversely if Hence aie ab FHT THT: 6. The denominator has the form bey? + bez® — 2beyz + acz* + acx* — 2aczz + abz? 4- + aby? — 2abry = ¢ (ax? + by*) + b (ax? + cz”) + + a (c2* + by*) — 2beyz — 2aczz — 2absy = = (a+ b+ 6) (ax* + by? + e2%) — of? — dYyt — — az? — 2beyz — 2aczz — 2abay = (a + b+ c)x X (ax? + by? + 2%) — (ax + by + cz)*. Since, by hypothesis, az + by + cz = 0, the denominator turns out to be equal to (a + b + ¢) (ax® + by® + cz*), and our fraction is equal to jaca afbye * 7. Reduce to a common denominator the expression on the left. The numerator of the fraction obtained will be equal to aya? (a? — b*) + Bb? (2? — a) (y? — a2) (22 — a) — — a? (2? — BY) (y® — b%) (20%). It is obvious that (a? — 2?) (a® — y?) (a? — 2?) = = a8 — (22 + y® + 22) at + (vy? 4 2%? + y%e) ag? — — ay?2, Solutions to Sec. 2 139 Hence (o* — 28) — 9) (2) = = BFR (et yt tat) h (aty® + as? + y222) D? — 22y?2? Substituting these expressions into the numerator and performing all the necessary transformations, we obtain the required value of the fraction. 1 4 1 8. So= WHat + a ey a (c—a) (e—b) * Reducing the fractions to a common denominator, we have 4 So= Gop aa baa” {6-4 —(a—¢) + (a@—b)} =0, a b ¢ _ Si=Goyete + opera tHE = 4 = GH a Gaa {2 (b—e)—b(a@—¢) +e (a—d)} =O, S2= * et + e 1 = ope {a? (b— c) —b® (a—c) +c? (a—b)}. Consider the numerator. We have a(b—c) —B(a—c) +c (a— db) = =ab(a—b) —e(a? — 4) +8 (a—B) = = (a — b) (ab —ca— cb + oF} = = (a— bd) la(b—c) —c(b— od] = = (a — b) (b—0) (a@— 0), wherefrom it follows that S, = 1. S3, S, and Ss; can be computed analogously, but we shall proceed here in a some- what different way. It is easily seen that there exists the following identity ( — a) (t@—b)(@—e) =a — (a+b 4c) + + (ab + ac + be) x — abe. 140 Solutions Putting, c =a, x = b and « =, in turn, we get the fol- lowing equalities a — (a + b +c) a® + (ab + ac + be) a — abe = 0, B® — (a+b +c) b? + (ab + ac + be)b — abe = 0, (a+b +0) 2 + (ab + ac + be) c — abe = 0. Further, divide the first of them by (a — 6) (a —c), the second by (b — c) (b — a) and the third by (¢ — a) x x (¢ — b), and add them term by term. Then S;—(a+ b+ c¢) S, + (ab + ac + be) S; — abe Sy = 0. But since it is known that Sy) = S, = 0, S, = 1, we have: S,;=a+b+e. To compute S, let us take the preceding identity and multiply its members by z. We obtain a(z—a)(«—b) (@-Q=2et-(A+b+oxe84+ + (ab + ac + be) 2® — abea. Proceeding analogously, we find: S,—(a+ b+) Sy + (ab + ac + be) Sz — abe 8; = 0. Hence S, = (a+b +0) S,— (ab + ac + be) S, = = (a+ b+ c)? — ab — ac — be = =a + b? + c? + ab + ac + be. Likewise, for computing S; (multiplying the original iden- tity by x), we find S;— (a+ b+ c) S,+ (ab + ac + be) S; — abe S, = 0. Consequently Ss = (a+b +c) (a? + b+ c2 + ab + ac + be) — — (ab + ae + be) (a + b +c) + abe = = (a+ b+ c) (a? + Bb 4 ec) + abe = =@+ +4 ad + ae + ba + Be + + ca + cb + abe. Solutions to Sec. 2 AAt 9. This problem is solved analogously to the preceding one. Namely, the equalities So = S; == S,;=0, S;=1 are established by a direct check; and to compute S$, we may resort to the following identity (@ —0) («@—) @-)@—a@ = =at—(atb+e4+ag 2+ + (ab + ac + ad + be + bd + de) x? — — (abe + abd + acd + bed) x + abcd Hence S,=(a+b+c4dS;=atb+e4d. 10. Put as before am bm om Sm= (a@—b) (a—e) a (b—a) (be) ae ceeray cee Let us take the first term of our sum o,, and transform it m (a+b)(a+c) _ (a4+b+c) amt am-t.abc (@—b)(@—c) (@—8) (@—2) : Making use of a circular permutation, we get similar expressions for the second and third terms of o,. Adding now all these terms, we find: 0, = (a+ b+ c) Sms: + + abe S,-;. Hence (after some transformations) oy =(a+b+c)S8,+abeS)=at+b+e (Sp = 1, Sp = 0), 6, = (a+b +0) S3+ abe S,= (a+b +0), sincee S; =a+b+c, S;=0, 63 = (a+ b+ c) S,+ abe S, = = (a+ b+) (a? + B® +c? + ab + ae + be) + abe, 0, =(a+b+c)S;+ abe 83 = =(a+b+o0 [a+b +6) (a? + bt + c%) + Label. 142 Solutions 41. Transform the left member of our identity in the following way (a—a) (a—B) (@—¥) (6 ~a) (h—B) (@—¥) abe { Fecheay tbat s te—a) (=P) (ev), _(O—9) O—B)O—y) _ aby (e—0) (c—a) (¢—b) (0—c) 0—a) (e—b) abe Consider the first four terms of the sum in braces. Expand- ing the numerator of the first term in powers of a, we get @ — (a + B+ y) a + (aB + ay + By) a — ay. Performing an analogous operation with the remaining three terms and adding them, we find that the sum of the first four terms is equal to S3— (a + B+ y) Se + (@B + ay + By) Sy — aBySo, where S, is the known sum (see Problem 9, where it is neces- sary to put d = 0). Proceeding from the results of this problem, we find that the sum of the first four terms under consideration is equal to unity, and, consequently, the sought-for expression takes the form abe { 1-22} = abe—apy. be 12. Consider the following sum: = at Bt . ‘= apa ne—o + Ba Peo F yt 8 + G=aa-—Ha—8 + Ba G—H CW From Problem 9 we have: S, =a+6+y+6. Puta = = abe, B = abd, y = acd, 8 = bed. Then at it atbtct Hi (a—B)(a—y) (a—8) (abe —abd) (abe —acd) (abe—bed) a2h2e2 “=a b—a(a—d) * Using a circular permutation, we get analogous expression for the remaining three terms. Thus, the given identity is proved. Solutions to Sec. 2 143 13. 1° Transform one of the terms in the following way: 1 sine a(a—b)(a—c) a ( Then the required sum is equal to 1 \2 4 (=) 6.) But (see Problem 8) S, = 1, and, hence, we get: 4 ; 1 4 eae a(a—b) (a—c) © b(b—c) (b—a) + c(c—a)(c—b) abe © However, this result can be obtained in a somewhat diffe- rent way. Let us consider the four quantities: a, b, c and 0, and form So for them. We then have 1 4 1 6 ig (a= By (a epicleb (a9) O=e/ic cle al (eb 4 + =a 0H =H = 9 since Sy = 0. Hence we get the previous result. 2° Likewise the sum can be transformed as (=) 144 Solutions And so 4 4 4 __ ab-+ae+be A a2hre2 A similar method can be applied when computing other sums of the form ses ieee ak (a—b) (a—c) bk (b—a) (b—c) ch (c—a) (e—b) © 14. We have : ak bho hej ey + W797 oH T ck ak ay + [ae nea t eae ee =? at k=1 and at k=2 (Problem 9), Hence ak bk G—He@-jana + aj b-je—h + + yaae-h eg ae eee = (R= 1 2). 15. We have b+e+d _ (apb+e+d—2)+(2—a) CHa =a) (da) (=a) ~ (=a) (=a) a) (ea) 1 =(+b+e+d—1) Gea asa ena + 1 + (b—a) (c—a)(d—a) * Applying a circular permutation to the letters a, b, c, d and adding the expressions thus obtained, we find that the sum in the left member is equal to A (at+b+e+d—2) { yea e=HETy + 1 1 Gabo O-D Oa + Cae ea eHa } + 7 1 (d—a) (d—b) (de) d—a) } Solutions to Sec. 2 145 since the second sum equals zero. It remains only to make sure that eaarcaceSee e eer (@—b) (a—«) (@—d) (@ G—ab—a(6—do—a) | i 4 4 1 +a heh enay) +} Wah Wade 4 + G9e-pe-5 ew = It is possible to reduce these fractions to a common deno- minator and, on performing necessary transformations in the numerator, to obtain zero. But we can, however, proceed in a different way. Multiplying the left member by (a — 2) (b — 2) (c — x)x X (d — 2), we get 1 aay bea) da) + 4 + Gap baa (4 *)l + aes (9) 2) d— 2) + ea} (d—2) + 1 +a d= (a— x) (b—2) (c—z) +1. It is obvious that we deal with a third-degree polynomial in x. It is required to prove that it is identically equal to zero. For this purpose it is sufficient to show (see the beginn- ing of the section) that it becomes zero at four different particular values of x. Replacing x successively by a, b, c,d, we make sure that our polynomial vanishes at these four values of z, and, consequently, it is identically equal to zero. 16. Transposing x? to the left, we get there a second- degree trinomial in x. To prove that it identically equals zero it suffices to show that it becomes zero at three diffe- rent values of z. Putting « =a, b,c, we make sure that the identity is valid. 17. Solved analogously to the preceding problem. How- ever, Problem 16, as well as this one, can be solved by making use of the quantities S, (see Problem 8 and the following ones). W125 146 Solutions 18. Put c The left member of our equality takes the form (et+uts ($4444) =34 4442 ate. Consider the fraction yi . We have y+2 b—e c—a c b2—be-+ac— a2 ae rate) eben Beeb) (ab te) = c 2c2 =a (—a—b—e +20) = >> since a+b+c=0. Using a circular permutation, we find yt , ete, tty 2% | 22, me 2 frie tel Hecpae tara a cebrnibeutneacn ete tat tna): But if a+b+c=0, then a + b? + c* = 3abe (see Problem 23, Sec. 1). Consequently y+2 tye, aty copaet iy at ete and the equality is solved. 19. Miltiplying the given expression by (a + b) (b + c)x x (e+), we get (a — b) (a+ ce) (b+) + (a+¢) x x (a + b) (b—c) + (a + b) (C— a) (6+ 0) + + (a — b) (c — a) (b—0). This expression is a second-degree trinomial in a which becomes zero at a = b, a = c and a = 0 and, consequently, is identically equal to zero, i.e. a—b , b—e | e~a | (a—b)(b—e) (e—a) a+b * ble + ca + @FH) O40) (e-Fa) We assume here b #c. If b =c, then it is easy to make sure directly that the identity holds true. 20. We have bre (b—a)+(a—c) _ =H @=9 ~~ @—h @ a) Solutions to Sec. 2 447 Treating the remaining two terms in a similar way, we arrive at the proposed identity. 21. Answer. 0. Solved analogously to Problem 19. 22. It is required to prove that am (a—b) (b=e)-+™ (a—d) (c—d)_b=-d _ gy em (a—b) (a—d)+-a™ (b—c) (c—d) ae Reducing to a common denominator, let us prove that the numerator equals zero. However, if the numerator is divided by the product (a — b) (a —c) (a — d) (b—c) (b— —d) x (c — d), we get the following expression am bm @=He—j ed + GHG EH om am + a5 aa + Gay (d@—5) (d—<) * At m =4, 2 this expression is equal to zero (see Problem 9). 23. Let us first prove that 4—24 2a) sea) (ema) pg ay 4g yas x (z—wy) (Z—ag) .-. (T—Gn-1) _ Gita ee n (t= 04) (Z— Gt) «(7 —Gn) Tn Cidade tC Likewise, it is evident that the second bracketed expression is equal to (za) (a7) «.. (+ On) 402+. an 7 And the product of the bracketed expressions yields gyn (2 =a) (22 a8)... (2?) (oS ae Replacing here z by 2? and a; by aj and applying the equality (+) in a reverse order, we get the required identity. 24. Given Be c2@—a2 ch a2— be et) + (AF 1)4+ Lee + (“a+ 148 Solutions The first bracketed expression is equal to eRe? _ (b—c—ay (be) 2be a 2be fe the second to (ao? b)(a—c-4-b) Bac 2ae : Likewise, the third one takes the form (a+b2—c% _ (a+b+e) (a+b—e) 2ab ae 2ab z Consider the sum of these expressions (a+b—c)(a+c—b) (atb—e)le+b—a) a 2be iccietenicazaeaneeey 7" SeeeeeennoS 4 tb 9 (ab bbe) 2ab =" (a po4 c)—b(e+b—a)—a(a+e—b)}= _ (a+-b—c) (e+a—d) (c—a +8) a 2abe fA Thus, we are given that (a+b—0)(ate—b (e4+b—a) _ 9 2abe wherefrom follows that at least one of the factors in the numerator equals zero. Suppose a + b —c = 0; then all the three bracketed expressions in the equality (*) are equal to zero, and, consequently two of the given fractions are equal to 1, while the third one to —1. The remaining two possibilities yield the same result. 25. Reducing the original equality to a common deno- minator and cancelling it out, we get (after some trans- formations) (a+b) (atc) (b+¢e)=0. (4) But the second equality (which is to be proved) can also be reduced to the form (a +8") (a"-e%) ("+0 =0. (2) Solutions to Sec. 2 449 It is quite obvious, that with an odd 7’ equality (2) follows from (4), since if, for instance, a+b=0, then a=—b and a*+b"=a"4(—a)"=a"—a" =0. 26. Rewrite the given proportion in the following way (bztey)yz _ (cx-+az)2z — (ay+bz) zy az byez ax—by+ez az+by—cz * pete Gee E ay ee But from the proportion FP -DA=F follows Bip = CLE ALE “D+F ~ B+F =f=F-1 and expressing A, C and E in terms of A, B,D, F). Therefore we have (ett y%)+2 (ar+by) _ a(22+y?)+ 2 (by+c2) _ ¢ a Oe +2) ty (cz+a2) = + . (it is easy to check, putting A= Subtracting c? + y? + 2 from each term of this equality, we get. 2(az+by—cz) _ x(by+cz—az) __y (cz-++ar—by) EEEEEea Teng TEEEEPEE CECE cg ELEC o : Take the original equalities aytbr batcy ex-az z(az--by—cz) 2 (—az+byea) y(ax—by+cz) * Multiplying these equalities, we find aytbr — ba+cy _ cx-faz Ce aoe Hence c= (ay + bz) p, b = (cx + az) p, a = (bz + cy) p. Multiplying the first of these equalities by c, the second by b and the third by a, and forming the expression b? + +c? — a’, we find b? + c? — a? = 2uber. Analogously, we get co + a? — b? = 2ucay, a? + b® — c? = 2uabdz, 150 Solutions Hence, finally aE se eee eee eee a (2p c2@—ar) b(ar-+c—B) — c(a®pb2—cd) * 27. Since a +b+c¢c=0, we may write (a+ b+ c) (aa + bp + cy) = 0. Expanding the expression in the left member, we find aa + BB + cy + ab (% + B) + ac (% + y) + + eb (B + 9) =0. But a+ 6 = —y, a+ y=—f, B+ y = —a, therefore aa + BB + cy — aby — ach — ‘cba = 0, or an + BB + + cry —abe(% 4 1. & ++) = 0, and since + £8 et * =0 (by hypothesis), we have: a’a + b®B + c¥y = 0. 28. From the equalities (4+ 2% — a) x = (2? + a? — BD) y = (a + DF — cz follows x y 2 1 4 1 a Beat cz a2—b2 ap b2— ee Put for brevity P+ i—eG=A, P+ RP—-P=B,F+R-—P=C. It is evident that our problem is equivalent to the follow- ing one: if the equation <° + y® + 2 = (x + y) (ex +2) x x (y +2) has the solution r=a, y=b, z=, then it also has the following solution UN lll al =q v=R %=¢- We know the following identity (see Problem 19, Sec. 1). (ety+27%-—2-—Y—F=3(e+y (+H y+. Solutions to Sec. 2 154 Using this identity, we can easily prove that the equalities P+P+F=(et+y(et+a)y+y, (1) @+y +7 =4e+y42)= =4ee+y(e+2 yt, (2) (@+y—2) @+2—y9) Y+2—2) = —4ays (3) are equivalent, and the existence of any of them involves the existence of the remaining ones. Thus, it is sufficient to prove that (Lega dyas (S44) (444) (444), ive. that (AB + AC + BC) = 4(4 + B) (A +C) (B+ C)-ABC. But A+B=2c%, A+C = 20%, B+C = 2a. Therefore we have to prove (AB + AC + BC) = 320°b*c?-ABC. Let us first compute AB + AC + BC, and then ABC. We have AB + AC + BC=A(B+C)+BC= = (B® + c? — a?) -2a% + fa? + ( — c)] X Xx [a? — (b? — c*)] = 2a?b? + 2a%e? — 2at + + at — bt — ct + 2b%? = —at — Bt — ch + + 2a*b® +- 2a%c? + 2b%c? = 4a2b? — (a? + ? — 2)? = =(a—b+c)(—a+b+ec) (a+ b—c)(a+b+c). By virtue of equality (3) (a + ¢ — b) (b+ ¢ —a) (a+b —c) = —4abe. Therefore AB + AC + BC = —4abe (a + b + ¢). 182 Solutions Compute ABC. Put @+eP+ce=s, then ABC = (s — 2a?) (s — 2b%) (s — 2c) = =P 2+ PEAS +4 (CO + are? + We) s — — 8a*b*c? = 4 (a2b? + arc? + bc?) s — 8? — 8a2b?c? = = s (4a2b® + 4arc? + Ab%? — (a? + b+ 22/2} — —8a*b*c? = — s {at + bt + ct — 2a*b? — 2a®e? — — 2b%c?} — 8a*b?c? = s (a + ¢ — b) (b + c¢—a) X X (a + b—c) (a+b + c) — 80h? = = —4abe (a + b + 6) (a +b + c2) — Bate? = = —4abe {A +B ++ a (b+c)4+ 0 (a+ec)+ +c? (a + b) + 2abc}. But (a+b) (a+c) (b+ 0) =a*(b +e) + + 0 (a +c) + c (a + b) + 2abe. Therefore, by virtue of equality (1), the bracketed expres- sion is equal to 2 (a? + b® + c’). But, by virtue of equality (2), 2@ +R +A =Laro top Therefore ABC = —2abe (a + b + c)*. But, as has been deduced, AB + AC + BC = = — 4abe (a + b +c). Therefore, (AB + AC + BC)? = 32a*b*c?-ABC. 29. 1° We have: Py = QyPyt + Py-2y Py — Pn-2 = AnPar—y Qn = 42Qn-1 + Qn-21 Qn — Qn-2 = AnOn-1+ The left member of the equality in question is transformed by the following method Pnsg—Pn | Pnai—Pn-t Pr Prt Prat Pn = 4nt+2 “Onsip—— = ante" ant Pn Prt Solutions to Sec. 2 153 We get quite analogously that the right member also yields a,41-@n42. Thus, the identity is proved. 2° We have Py Prot PrOnt—QnPr-y_ (—1)P-4 On Ona QnQn-t nar” Putting here k =1,2,...,n and adding termwise, we obtain the required result. 3° We have Prt 2Qn-2 — Pr-2Qnt2 = (ant2Pati + Pa) Qn-2 — = Pree Qnitante + Qn) = use (PntQn-2 — Pn-20ngs) + + PaQn-2 — Pn-2Qn = = ange ((AntiPy + Prt) Qu-2— Pu-z (anti@n + Qn} + + @nPa-t + Pr-2) Qn-2 — Pa-2 (@nQn-1+Qn-2) = = Myt1dnt2 (PnQn-2 — Pn-2On) + + aute (Pa1Qn-2 — Pn-2Qn—) + +t ay (Pn-1Qn-2 — Pn-2Qn-1) = = An 41Ant2 {(anPu—t + Pa-2) Qn-2 — = Pry-2 (4nQn-1 + Qu-2)} + Qnt2(—4)" + ay (—1)" = = (Gynt 2tntidn + ante + a) (—A)". 4° It is known that P, =a,P,-1 + Py-2. Therefore Pn Pros 1 1 =4n+ == a, =an = Pag 8 T Pyay = On t Pag = Ot GPa FP Pra Ph-2 i 4 7 1 oe Pra ta te, 4 nt + cee - n-2 Po ate 1 = — mt ate, eat ! 1 tae The expression for oe is found in a similar way. nA 154 Solutions 30. On the basis of the results of the preceding problem we have P P; Prog = (4s nats ++ +4 Go) = (oy das +++) On) = G+ Consequently, Pr-; = Qn. © 31. We have to prove that Pht —PrtPrit = PrP aia — Pi or Prgi(Pngt— Pret) =Pr(Put2 — Pa) But Pats= @Pn t+ Paty Pate = @Pnii + Ph. Consequently, Pats — Pat = @Pp, Punto — Pp = aPais. Hence, follows the validity of our identity. 32. By hypothesis a 1 Pa 1 SEB bBo) On Gad Or os etsy, 44. op, 7 a T+gh- Thus, z is obtained from e if 1 is replaced by 1+ in P, Pn ; i i Pa _ Prost Pn-g + a in this fraction. But Oz 10 Ora Therefore Pn (14:Gr) Pat+Pa pad + PaPacs (FFE) Ont t Onn CAF PrGnt r= 33. It is obvious that at k = 0, 4 our formula holds true. Assuming that it is valid at k = n — 1, let us prove that it takes place also at k = n. And so, we assume Paws ~ Ont * by + eee ey iioyll fee Solutions to Sec. 2 155 However, according to the rule for composing P, and Q,, we have Prot __ bn-tPn-2+4n-1Pn-3 Ont bn-tQn-2F4n-1Qn-3 ” where Pp_o, Pn—3, Qn—2, Qn—3 are independent of an_; and 1. On the other hand, it is clear that the fraction a bot a+. en net gi bat +o is obtained from the fraction bo St OP ll | OT Ena by replacing by; by bn +e a in Therefore (ont + G2) PneatensPnes bot Ft = Peat (bnt+ 5) neat anciQn-a PatQn-atOnsOn-st Zo One QnttZ™ Ona — bnPn-ite@nPn-2 _ Pn ee 34. Denoting the value of our fraction by aa we have n Py=r, Qa=rt+41, Pp=r( +t), Q=Ptrti. Using the method of induction, let us prove that rat tT 156 Solutions At n = 1 these formulas are valid. Assuming their validity at n =m, let us prove that they also take place at n = =m+4. We have Prt = bmt+1Pm + amtiPm-t- In our case we find 35. Put A, 1 21 Up Ura pap Then we find ue Or Ere Therefore Further 1 4 ra where Thus Med 1 1 gig ag anes ~ ae up up a+ ry uy he a et Ug Using the method of induction, we also get the general formula. Solutions to Sec. 2 157 36. Let us denote the fraction ay Wee, fa Pa é by =, and put the fraction Qn a £46289 cabs + Sa bs, 4 natn nbn equal to ze. It is required to prove that 2 eH e “. for any in whole positive n. We have Pr) Pe __atbs ; Or bp 7 Qa baby pag? °°"? Py excotbe Gee Osho +a) * We may put Py = a4, Qi = by, Pp = aybe, Q2 = bbe + ao, and then the following relations take place (see Problem 33) Prt = bnstPn + OntiPr-ay nts = bnt1Qn + antiQn-t- Put P, = cay, Py = Cye2ayba; Qi = eibr, Oy = Cree (bib2 + a2) Let us prove that for any m we then have Ph = ey€_...CpPn, Qn = C1€2- ~~ CnQn- Let us prove this assertion using the method of induction, i.e. assuming its validity for a subscript smaller than, o1 equal to, n, we shall prove the validity for the subscript n+1. We have Prot = CntsbntaPa + CntneidntiPn ty Qn = Cus Wns iQ lula Hp 1Qnt- 458 Solutions Hence (with the asumption) Pats = Cnttbnsieice «+ + CaP + H+ CnentsAntil1ce «+ + Cn-1Pa-1 = = C4206 Cutt (OnttPn + ArtiPn-as) = = CC... CatsPrise Likewise prove that Qntt = C12 +» Cnt iQntae Now it is easy to find that Pn Pa Qn Qn” 37. 4° Put 1 P 2cosz— 1 =o OTD coalesce Qn 2cos x We have Therefore we may put sin 2x oie ainsi Further Pp 4 4costr—4 "Qj 28 Poon cosas: Consequently, we may take ». _ sin 3z sin 2e P= Sng? G2 Sng i 4 i a Let us prove that then P, = mee a 1Qn= ae for any n. Assuming that these formulas are valid for subscripts not exceeding n, let us prove that they also take place at n-+ 1. We have (see Problem 33) sin(n+4)2_sinnz 4 Pay= ——— = nu = 2 Cos @— sing ~ Sine sin (n+ 2)z. Solutions to Sec. 2 159 sin (n+1)2 , and there- sin a In the same way we find that Qniy= fore Pn _sin(n44)2 On sinna for any whole positive n. 2° Let us denote the continued fraction on the right by Pa, We have to.prove that Qn Dealt bet babs t+ set babs. - Dns We have Py _ A Pp _ beth a 17 @ 1° Therefore we may take: Py =1, Qi =1, Pr = bd. +1, Q2 = 1. Then, using the method of induction, it is easy to prove that Pr =1+ be + babs +... + babs... bas Qn = 1 and, consequently, our equality is also true. 38. 1° We have sina+sinb+sinc=sin(a+b+c)= =(sina+sinb)+[sine—sin(a+b+c)]= ee ied ae cos SEHR _ Fett os) a+b te. be =4sin = sin * z sin ). 2° Analogous to the preceding one. 39. Consider the sum tana + tanb + tance. 160 Solutions We have sin (a+b) cos a cos b _sin (a+b) cosc+sinccosacosb _ Fra AHCHEEEHEH GS-0 0s b Cos/eiiiiaittiit __sin(a-+0) cos e-+-cos(a-b) sin c—cos (ab) sin c-+sin ccosacosb _ C08 @ COs b COS ¢ _sin (a+b-++c)-+sin c [cos a cos b—cos (a+b)] _ ba COS a COs b COS ¢ fe tana+tanb+ tanc= sine + a cos c~ _sin(a+b+¢)+sin asinbsine 7 COs @ Cos b cos ¢ Hence follows the required equality. 40. The equalities 1°, 2° and 3° are easily obtained from Problems 38 (1°, 2°) and 39 putting a = A,b=B,c=C anda+b+c=A+B+C=n. Now let us prove 4°. Rewrite the left member in the follo- wing way A B Cc A B S=tan + tan>+ tan 5 (tan4 + tan ) . faa But since A+B+C=n, we have C_ an A+B\_.| A+B 4 lanz = tan (¢- >) t—— = AtB tan 2 Hence since A B in Gaal tan $+ tan eto ae 4, 5° —tan- tan > 2 2 Solutions to Sec. 2 464 5° Indeed sin 2A + sin 2B + sin 2C = = sin 2A + 2 sin (B + C) cos(B —C) = = 2sin A cos A + 2 sin A cos (B — C) = = 2sin A [cos A + cos (B — C)] = = 4sin A sin B sin C. 41. 1° It is necessary to find how a, b, and c are related if cos a+cosb+cosc—1—4sin + sing. sin=0. To this end let us reduce the left member of the equality to a forth convenient for taking logs, i.e. try to represent it in the form of a product of trigonometric functions of the quantities a, » and c. We have b —b cosa+cosb=2cos ot cos “F = es 2 cost 6 gine? gine 2 2 (cos 7 cos* + —sin? > sin’ z) cos c— —2sint Therefore the left member takes the form 28 cos?) _osin?* sin? —2 sin? £ — 2 cos 5 cos’ > —2sin? > sin® 5 — 2 sin? > ee eee eee —4sin5sin >sin5= 2 2 2 =2 [ cos? $ cos? 3 — (sin? $ sin? S 4 2sing sin Ssing4 +sin?5) ]=2 [ cos* + cos? — (sin sin S+sing)? ae =2[ (cos $cos 4 + sin sin 3) + sin 5 | x a b parte Pa x [ (cos $ cos $—sin $ sin 4) —sin5]= =2 (cos “+ sing) (cos * —sin $) = 11-1225 162 Solutions a—b nm cy" a+b _[%¢\)_ =2 [cos <5" + cos (=-3)] [eos *5*—cos ($-3) J= = —8sin tt ¢=4 gin Starenb x ~ 4 ssn 72D gy «a+b bens : By hypothesis, this expression must equal zero and, conse- quently, at least one of the factors must be equal to zero. But from the equality sina = 0 follows a = kn (where k is any whole number). Therefore, among a, b and c, satisfying the original relationship, there exists at least one of the four relationships a+b+e=(4k+1)n, a+b—c= (4k —1) a, at+te—b=(4K—1)n, b+c—a= (4k —1) x. 2° We have (see Problem 30) ; tana-+ tan b+ tanc—tanatanb tane = Seto to : By virtue of our conditions sin(a+ b-+c)=Oanda+b+c =kn, 3° Transform the original expression. We have 1 — cos* a — cos? b — cos? ¢ + 2 cosa cos b cosc = = 1 ~ cos? a — cos® b — (cos? c — 2 cos a cos b cose + + cos* a cos? b) + cos? a cos* b = 1 — cos? a — cos? b — — (cos ¢ — cos a cos 6)? + cos* a cos* b = 4 — cos* a) (1 — cos? b) — (cos ¢ — cosa cos b)? = = (sin a sin b — cose + cos a cos b)x X (sin a sin b + cos ¢ — cos a cos b) = = [cos e — cos (a + b)] [cos (a — b) — cosc] = = Asin QEb Ee sin SEO sin EM sin 5 Consequently, there exists at least one of the following relations a+b+e=2kn, a+b—c = 2kn, a+ec—b = Qk, b+ec—a = 2ka. 42. Put Solutions to Sec. 2 163 Then Qe 2y 22 Tog = tana, To = tanB, Tog =tany, and our problem takes the following form. Prove that tana + tan B+ tany=tana tan § tan y if % tan tant B tant tan> tan + tan > tan> + tan-5 tan > Rewrite the last equality as tan $ (tan $+ tan #)- (1—tan $ p tant) =0. = Dividing both members by 1—tanbtont, we get © san BEY % _ pot BEY x B+y tanZ tanPSY—1=0, tan F=cot Pt =tan (5 + ). Hence SBE 8 Lin (if tangents are equal, the corresponding angles differ by the multiple of x) and atp+y=(2k+ 1) a. And co the proposition is proved (see Problem 40, 3°). 43. Put b=tanB, c=tany, a=tane. Then b—c _ tanB—tany _ fae Tbe = Tptan Btany a 2" (B—V)> and, hence, our equality is equivalent to the following one tan (B — y) + tan (y — @) + tan (a — B) = = tan (B — y) tan (y — @) tan (« — §). Put B-y=2, y-a=y, «—-B=z2z. Let us finally prove that tan z + tany + tanz = tana tany tanz 464 Solutions if aty+z2=0, But then we have tanz+tany tan (z+ y) = — tanz, caren —tanz. Hence follows the required equality. It is obvious, that the last two problems can be solved by direct transformations of the considered algebraic expres- sions. 44. We have : : ata Serica tan Sq Sit 8a _ sina @—4sin?'a) 44, 3—4sin®a cos 3a, —4sin? a Divide both the numerator and denominator of this . 4 fraction by cos* « and replace wa by 4 + tan? a. We get 3—tan2a Vi T=Stanta = ‘ane ane V3—tana tan 3a = tana SS 1—V3tana 14+ V3tana Hence tan3a= tana tan(F +a) tan (Z— ). 45. Multiplying both members of the equality by a + b and replacing unity in the right member by (sin? « + + cos? a)*, we get sinta + costa + sin* ate costa = == sinta+costa-+ 2sin?@ cos", whence b.. . a z sinta—2 sin? a cos? a -+ > costa=0, F = 2 ( Esinta—/ = costa) =0, a 6 b.. a — sinta= = costa, a 6 Solutions to Sec. 2 165 or sinta costa Wo = OO, Substituting it into the original equality, we find 4 =a Therefore sina , cos8a a b 4 oe apoE TPO GEO 46. From the second equality we have (a, cos a + a, cosa, +... + a, cos a) cos 8 — — (a; sin a, + a,sina,+ ... + a, sin a@,) sin ®@ = 0. On the basis of the first equality and since sin 0 0, we get a sina, + a,sina, +... +a,sina, =0. (+) Multiplying the first equality by cos 4 and the equality (+) by sin’, and subtracting the second result from the first one, we have a; cos (a + A) + ay cos (ag +4) + ...4 + a, cos (#, + A) = 0. 47. It is obvious that the left member is reduced to the following expression (tan B — tan y) + (tan y — tan a) + (tan @ — tan B) = 48. 1° We have s sa. PP Pe-)" Hence _@ _ ap (p—a) Ta—T eee Therefore a be o= + Tar | rer £ {a(p—a)+b (p—b)+¢(p—o)}.- 4166 Solutions But s*= p(p—a) (p—b) (p—c). Hence a b c “oats tea 0—9 toa a}= (p—)+ (p=), (P—)-+(p—2)_, (P= a+ (p—B) = Goyges t oeapes + Gea t= =2(ratrotr): 2° We have ar, Bry Ore 9 = Gt Bg Ow F Gay (eB) az Be =8{g2ae nese t paae—aesat ie + parca} But (see Problem 9) az b2 @-Fe—H a5 + HH OJ Oat 2 2 toa oe B=9e— 9 C—H ~ PH P—H PJ Therefore jee ~(p—a)(p—b)(p—c) 3° We get Heed 1 1 5 (ab--ac+-be— p2) ret rites (Sat 55+ pa) “GeO Further Jt bt fad (a(p—ath(p—Yte(p—o}= =Lep—a—_ey= =2(—p abt ac -}-be). Solutions to Sec. 2 167 The rest is obvious. ae acta the first sum a (ehest Bege—a peat aea == (*[popecg tego tee] ~2pabe [ey e=at p=aesat eaecn)t a b e ey + abel eoyacat p=geca taco] } But (see Problem 8) 1 4 1 He Gea t Gg e—9t aH” a b ¢ ae (@—8) @—<) + (=e) (6—a) © (=a) (e—8) au Therefore p be ac ab . 0= 5 g=pesat ecaeca teas ai ab a ea y+ w= baa) t+e=nemy ee 1 = abe { (aecae aoatiecs baat cemayemyt 4 4 ae (O—a) (0—b) ora |ta } =1 And so p 1 O=-3=7- Let us ae over to the second sum. We have ar, Pry am {ene +e=50=5+ er, a + emo} Sa Leese t Be + e=ae—aeay t eoaecnesa}: 168 Solutions But a Ug (a—b) (ae) (@—p) T (b=e) (6—a) (b—p) pe (ep) © =a) (P—b) (P—) ~ +=He Therefore oa erp), = a (P—a) (p—b) (p—e) 5° We have _ ar, brs mt = Gah erat Oo bra t eae a b aoaesaera + a0 t t a + g=aesne=a} -—{ eaesaeent ; ¢ + ogo tae TH + : ii + =aG-H ea GOH E=H Sere, “(p—a) (p—)(p—-) oS Further ya Otte 4 Cebayry 4 (a+b) re = Gb a—a + Ga ba) * (ay (CH et (be) Peers ect ee) eeea . {@=pecae=a (=) Po + OH =a Ht (a) if ! atts} =s(atb+0) { aaesaecet 4 fl Helo b= a) (pb) teem) =D) pee : b —s (nwa ec t watageat ¢ T =a) (e—b) (p—e) }: Solutions to Sec. 2 169 But woes tee eit (a—b) (a—c)(a—p) * (b—c) (b—a) (b—p) 1 4 + ae-heen t =a —9 O95 =” Therefore, the first braced expression is equal to . The second braced expression is equal (p—a) (p—b) (p—e) pe to tan > tan = eoreanre pi ieaecaeares 8 9 vt 4y B c Hence, tan-5 tan tan-> = tan > tan > tan . 54. The left member of our equality can be rewritten as 1 ain (@—b) sin (aq sin (boo (Sin (b—e) —sin (a—e) + +sin(a—b)}. But we have sin (b—c) —sin (a—c) = 2sin 200s °+4=% | Therefore, the braced expression is equal to 2sin >> cos b+ o— Pe _ asin b= cos pd =4 sin 254 sin 2=” sin 25“ sin2< cos 2 2 The rest is obvious. 52. 1° The fraction in the left member has the form 4 . ‘i Fn aah sm era sin boa (Sin asin (b—e) + + sin bsin(c—a)-+ sinc sin (a—b)} = c >) sinasin(b—e), in (a—b) sin (@—c) sin(b—c) Solutions to Sec. 2 474 where summing is applied to all the expressions obtained from the one under the summation sign by means of a circular permutation. But sinasin (b -d=t [cos (a—b +c) —cos(a+b—c)]. Therefore we have > sin asin (b—c) =F {cos (a-+¢—b) — eos (a-+b—0) + + cos (b-++ a—c) — cos (b+ c—a) + cos (c+ b—a) —- —cos(c4-a—)}=0, and our identity holds true. 2° The given identity can be proved similarly to case 1°. But we can get the same formula immediately from formula 41°, replacing aby > —a,bby $ — b, and, finally, c by $ —e. 53. 1° We have to prove that > sin a sin (b — c)X X cos (b + c — a) = 0. Here summation is applied to all the expressions obtained from the original one by means of a circular permutation. But sinasin (b—c) = {eos (a—b+c)—cos(a+b—c)}. Therefore > sina sin (b~c) cos (b+ ¢—a) =+ >) cos (b4-¢—a) x X cos (a —b+g-45 cos (a-++b—c) cos (b+c—a) = =+ D>} [cos 2c + cos (2b— 2a) — cos 2b — cos (2e — 2a)] = =4 (cos 2c — cos 2b + cos 2a — cos 2¢ + cos 2b— — cos 2a-+ cos (2b — 2a) — cos (2¢ — 2a) + cos (2c — 2b) — — cos (2a — 2b) + cos (2a — 2c) — cos (2b—2c)} = 0. 2° Can be obtained from 1° by replacing a by 4 b by ot and ¢ by $7 172 Solutions 3° Likewise we find > sina sin (b—c) sin(b+ce—a)= =F {sin 2 (b—a) + sin 2(e—b) + sin 2(a—o)}. It only remains to show that 4 {sin 2 (b—a) + sin 2(e—b) + sin 2(a—o)} = == 2sin (b-—c) sin (c—a) sin(a—D). 4° Proved analogously to 3° or by replacing a by t-4 b by $7) and ¢ by poe. 54. 1° We have > sin? A cos (B —C) = >) sin® A sin A cos (B—C) = = 4 Disin’A{sin(A+B—C +sin(A—B+0)}. But since A + B+C =n, we have DI sin? A cos (B— C) = 4. 3\ sin® A (sin 2C + sin 2B) = = >| sin? A (sin B cos B + sin C cos C) = = sin? A sin B cos B + sin? A sin C cos C + + sin? B sin C cos C + sin? B sin A cos A + + sin? C sin A cos A + sin? C sin B cos B = = sin A sin B (sin A cos B + cos A sin B) + + sin A sin C (sin A cos C + cos A sin C) + + sin B sin C (sin B cos C + cos C sin C) = = sin A sin B sin (A + B) + sin A sin C sin (A + €) + + sin B sin C sin (B + C) = 3sin A sin B sin C. Solutions to Sec. 2 4173 2° We have Sy sin? A sin (B—C) = sin? Asin A sin(B—C) = = >) sin? A sin (B-+-C) sin(B—C)= =4, 5) sin? A {cos 2C —c0s 28}— 5} sin® A (sin* B—sin* C) = =sin? A sin® B sin®C >) (z45— garg) =sin* Asin® Bsin®€ x 1 1 1 1 bo. 1 : 1 . {saz C— smtBt Gnt a smec + sew sin A 55. 1° We have sin 32 = 3 sin — 4sin* z. Therefore >i sin 3A sin’ (B—C) =+ 5} sin34 3 sin (B—C)— —sin3(B—C)}=4 >) sin3 (B+C) sin (B—C)— — + 3) sin3 (B+-€)sin3(B—0) = = 5 {cos (2B + 4C) — cos (4B + 2C)} — —+ >) (cos 6C — cos 6B) = = 3 {cos 2 (B+ 2C)—cos2 (C+ 2B) +cos 2 (C+ 2A) — — cos 2(A + 2C) + cos 2 (A + 2B) —cos 2(B + 2A)}— — + (cos 6C—cos 6B-+c0s 6A—cos 6C + cos6B—cos 6A}. But cos (2B + 4C) = cos (2B + 4A), cos (2C + 4B) = cos (2C + 4A), cos (2A + 4C) = cos (2A + 4B). 174 Solutions And so, we finally have >) sin 84 sin? (B — C) = 0. 2° Since cos 3x = 4 cos* z — 3 cos z, we have > sin 3A cos? (B—C) = = + Di sin3 (B+C) {cos3 (B—C) +3 cos (B—C)} = = + D}sin3 (B+C) cos3 (B—C)+ +4 5) sin3 (B-+C) cos (B—C) = = 45) (in 6B + sin 66) +2 I {sin (4B-+.2C) + + sin 2B-+4C)} = + (sin 64 + sin 6B + sin 6C) = =sin 3A sin 3B sin3C. SOLUTIONS TO SECTION 3 1. The validity of the given identity can be checked, for instance, by the following method. From the formulas (x) (see the beginning of the corresponding section in “Problems”) we get V24V3-V 24+) 4, VI-V5-/3-Y 5. Therefore we have Fz T\2 24V3 (V3+V 4) _ U4 V3): V2 V24V24V3 y+ 2 Vt 23-43) + Y3)-V2 14-3 2V3(1+ V3) V6 Solutions to Sec. 3 475 Likewise we get 2-V5 Viva) V2-V 2 Va yi-f ays i 2 2 eed 3— 2(8— Vo Consequently 24V3 2-V3___\?_ /t4+-V3, V3—1\?_ (ater er) (eae) _(2V3 2» = (V5) 2 2. Let us prove the proposed identities by a direct check. 1° Put V2 =a, ie. a? = 2. It is required to prove that (1 —a + a)? = 9 (a — 14). We have (4A—a+te)?=1+ 07+ a4 + 20? — 203 — 24 = = 3 (a? — 1), since oF = 2, af = 2a. Hence (1 — a + @)§ = 3 (a? —a 4+ 1) (a? — 1) = =3(2—ati)(e+1)(2—1) = =3(@ +1) (a—1) = 9 (a —1). 2° We have to prove that (W247 DVB)" =9 (W5—77). Squaring the left member, we find WELW 00+ Y 62542 Y 40-2 Y 50—2 500 = =V F427 545 544 5-27 50-10,/ 4= =9(/5—/%). 178 Solutions 3° Proved as in the preceding case. 4° We have to prove that ( V544 y= 3425 Y5s—-1/ 3-275" Put We have (4 ‘= GEM — tte eat dotp at _ VYs-1 _ 3420-43024 20% B= 2a} 302203” since af=5, Further (au 4 3420-402 (3+2a) v4) ~ 82a a2 (3—2a) 5° It is required to prove that (1473-5) =5 (2-H), /3=a, ie. ab = 3, Put We have A+ a— a)? =1+4 a + at + 2a — 2a? — 2a? = = 14 2a — a — 20% + at, Further (+ a—a%)} = 1+ 3a — 5a* + 3a — a5 But a® = 3a, a = 3. Therefore (1 + a — ot)? = 10 — 508 = 5 (2 — 27). 6° Put ,“2 =a and prove the first equality which can be rewritten in the following form 5 fata)? = (1 + 0% Solutions to Sec. 3 477 The right member is equal to 4 + 5a? + 1004 + 100% + 50% + al? = =5(1 +a? + 2at + 20° + a), at? = 4, since Further a =2, af = 2a, af = 203, and, consequently, (A + @)® = 5 (1 + a? + 2at + dew -} 2a’), It only remains to prove that A+ a+ a)? = 1+ 4a + a? + 203 + 2at, The last equality is readily proved by removing the bra- ckets in the left member and performing simple transfor- mations. To prove the second equality we have to show that S/T, 8/8 _(8/,3/B, bf E_ sy ty V $+ V $=( pet V at V w-V as). or 5 (14974) =( 164/84 /2—1)°. Put VY2= a, a = 2, af = 2a, a? = 2a*%, a? = 2a’ Then we have to prove that (af +o +a — 1)? =5(t + 07). Expanding the left member, we find 1+ a? + a + a + 2a? + 205 — 2at + 2at — 203 — 2a. Making use of the equalities enabling us to replace high powers of « by lower ones, we find the required identity. 3. Put A B 7 Then A=a,, B=bs, C=cr, D=d. 12-1225 178 Solutions Consequently V Aa+VBb+V Co+VDd=Vi(a4-b+e+d). But A+B4C+4+D=A(a+b-+ce+4d) and A4+B4C4D a-~bje+d’ As ie. 5 VAFB+CHD ,= Vv Va+bte+d Replacing YA in the equality V Aa+V B+) Co+V Dd=V i (a+b+e+4) by the found value, we obtain the required identity. 4. Put for brevity VW ae + by ce = A. We have 3 Saas bys = S+%- since 4 1 4 3 =: bys 3 aa az? = by’ =cz> and gtytyeal Likewise we find A=yVb and A=itVe. A_y,p A_yz tab, = =Ve. Adding these equalities termwise, we get A(44444)=Va+Vi+Ve. Hence Solutions to Sec. 3 179 Hence, finally, A=Vaty b+ Ve. 5. Put 1+ wi Soo Then a, =a" +B", ba =a™—B*, where af = + Prove that Amn — "B® = myn We have dina — "BG" = (2 +B) (a +B") SEPT _ =a" 4 pment otpr (yn 4 pmn) amen Bm=n eee But oad, consequently, Gm—n men gmen Onn — FE = ant BM — Omens The second relation is proved in the same way. 6. Put 14V5 _ 1—V5 _ Pecaeeat Bese ed Then a+Pp=1, ap =—1. Furthermore w@—a—1=0, p?-p-—1=0 180 Solutions and = —B")- Proof. 1° We have _p qh pm) — ttm re M+R pr) “Vi {(a" +a) — (B" + B")}. Multiplying both members of the equality a2—a—1=0 by a”, we get ati=a, a®ta™tag™!, Analogously, it is easy to conclude that 4 Bt Bret, Therefore Unf Unt = Vi (a1 B1) = ting. 2° We have UpUn—k + Up—Un-h-t = =F ChB) (BF) + (Bh) (ar ht prey) — =f fort pr—atpr*— pla pat 2 4 pet pltgyn ht —Pthight} a ot p+ Bt p(t ra) — R-1 (S4+55 )}= ae hy RI Fe = a a2 PEE — on Barer y = {ohare pt pt prot a on ABE =F (era + B+ By), Solutions to Sec. 3 184 since aB+41=0. Then we perform the following trans- formations Leotpar spre pat { amt (a+ a )+8" (B45 )}= =F {2 (@—B)+ 8" (Bay) = 258 (orp = Gees ali aay cet “5 ( Br?) = un. 3° Obtained from 2° by putting n= 2k, and then repla- cing k by n. 4° We have to show that 5 (a3"— Bo") — (a — p")3— (a+ — prty3 4 (at — prs = 0. The left member is transformed in the following way 5 (a"— po) — a8" (a8 44—A,) + 302"8" (atB+4— Ze) — —80 (9P ae) (Pt ). It is easy to show that *B +4 — a =0, af?+4 — a =0. On the other hand, we can easily make sure that OHI P HI a +B = (a+) (of + B)+1=@ —af + B4+1=5. Hence follows the validity of our identity. 5° We have to prove that (a —B")s— (a? — By ("1 — p™) (1 — B™) x (a? — Bm?) = 25... First prove that (cx? m2) (42 Bm?) gt BEM — (— 1)" (at + BS), (1 Br (at — Br) 0). Multiplying the two obtained expres- sions, we find an ues oto] a a 9. Factor the expression rn — 3n — 2. We have n® — 38n —2 = n§ —n— 2n— 2 =n (n? — 1) — —2(n+ 1) = (n+ 1) (W? —n—2) = = (n + 1)? (n — 2). Likewise n® — 8n + 2 = (n — 1)? (n + 2). 184 Solutions Now we may write: n3—3n—2-+4 (n2@—1) Voi n3—3n+24(n2—1) Vn2— (844)? (n= 2) + (n2—1) Vr’ (n $1) Va—2 (n—1)2 (n-+-2)+ (n2@—1) Vr' (n—1) Vn-F2 yc (tt) Vn=2+ (m1) Vn $2 _ (n+) Vn=2 (n—1) Vato (nt) Va—2 (n—1) Vaz 10. Consider the second one of the fractions contained in the first brackets, namely: 1—a _ 1—a a Vi-@—-1ta VYi-@—(i—a Vita—Vi-a* And so, the transformed expression takes the form [ Vita ire ViFa-Vi ¢ (Vi=@—1) ae 2(YWi—a?—1) aay ~ ({-a+1—a—2Vi—a) 41. From the formula (+) it is easy to get: VALVB4V A_-VB=2 ee In our case A B=4r—4, A-B=27?—42+4, geese cee aac z—2 if c>2, 2 B= 2 — Ut ed Crem Ce if 7<2. In the first case we have Viq2V sai 4V 2—2V ent 2 f =x =2V2-1. Solutions to Sec. 3 185 The second case yields VeqaVani+V 2-2) 221-2 V =. It is easy to see that at z = 2 the expression under con- sideration is also equal to 2. 12. In this case A=a+b+e, B= 4ac + Abe, A*— B= (a+b +0)? — 4dc — 4be = = a* + b? + c? + 2ab — 2be — 2ac = =(a+b—0)%, If a+b—c>0, then V P—B=a+b-c. If a+b—c<0, then V &—B=c—a—b, Hence, we easily obtain that the given expression is equal to 2Ya+b ifa+b>e, and to 2Vc ifa+b, cos a sin =+ V See : But in our conditions Fame ye (0 =(—1)* The second assertion is proved analogously. 19. Let us prove the validity of some of the proposed for- mulas. Let us, for instance, prove that Aj, = 0 ifn =0 (mod 2). Put n = 21. Then FAs = 00s (2 +H) +eos (HE 4+a—Zn) + +008 (32452) +005 (TF +350) = = —00s (48 #2) —cos (In—B Sn) + +008 (“E+ in +m) +e0s (Qn 4 Bn) ~~ 115) (eo ($+ hs) + +(—1)! cos (B+ gx) +005 (E-$2) =0. Solutions to Sec. 3 489 Let us prove, for instance, that A;, =0 if n=1, 3, 4 (mod 7). We have: 3 3 FAu = cos (+ nn —F x) + cos (+ nt — a) + 5 3 + cos (Fan - a) . If we replace here n by a number, which is comparable with it by modulus 7, then all the cosines will acquire only a common factor equal to +1. Indeed, let us assume that n=a (mod 7), i.e. n =a + 7N, where N is an integer. Therefore cs (Sp) men (MEINE 4) =cos (2+ bv) =(—1)" cos (4-8) = =(—1)* cos (4-8), since in our case k = 4, 3, 5 and, consequently, is odd; (B is equal either to an or to Ba . Therefore, in order to prove that Ay, = 0 at n = 1, 3, 4 (mod 7), it is sufficient to prove that it will take place at n = 4, 3, 4. The validity of this is readily checked. First put n = 1. Then we prove that 4 43 3 3 5 3 cos (7a—-Gr) +00s (Fa-Hx) + 00s (F2-q") =0. After ST we get: 44 cos Ht x + c0s & m+ cos t= cos (2-2) + cos ant +cos $= — 008 7 : 7% + cos gx=0. Let now n=3. Then we have to prove | 3 3 cos ae ) +008 (Faas ) +c0s (2 ayn \= 27 = cos n+cos 4 +008 Hx =oos (n-++47) + + cos (2x—#) = —cos Fp t00s 7 =0. 190 Solutions Reasoning in the same way, we make sure that at n = 4 we also obtain zero. In conclusion, let us prove that Ag never becomes zero, i.e. at no whole values of n. We have i) = cos (Gr -% )+(=1)" cos (4ra +h). 4 Ag=cos (4rx—h2) + cos (—4na-+na— Consider the following cases: 1° Let n = 0 (mod 4), n = 4N. Then + Aa= cos (Nx-F x) +(—1)'% cos (Ma+% x) = 7 4 =(—1)" cos 7 n+(—1)" cos Fg = —1)" (cos 4-008 7 2). The bracketed expression is not equal to zero, since it represents a sum of cosines of two acute angles. 2° Let n= 1 (mod 4), ie. n=1+4 4N. 1 7 3: 4 x As = cos (FtNa— sex) + cos (F+3Wa—752) oe N nT 3a 4 =(—1)" {cos (—7_x) +008 (“E—7e) } = =(—1)" {cos #005 # a}. It is obvious that the braced sum is not equal to zero, and, consequently, in this case Ag is also not equal to zero. It only remains to consider the cases: n = 3 (mod 4) and n= 2 (mod 4), but we leave them to the reader. 20. It is required to prove that Dp) =0 ifk=n, n—1, n—2, n—4, n—5, n—6, and the sign before p(k) is chosen accordingly. It is evident that DP=4% (+32 +C Dy (—4)"+D 5 cos 22 | Solutions to Sec. 3 491 The first two sums on the right are equal to zero. It remains to prove that > cos 28 — 0. If k is a whole number, the following cases are possible: 1°k is exactly divisible by 3, k = 30; 2° k, when divided by 3, leaves the remainder 1, k = =314+ 1; 3° k, when divided by 3, leaves the remainder 2,k = = 3142. In case 1° 2nk cos = = 1. In cases 2° and 3° cos 2 = cos a, Let us first assume that n is divisible by 3. Then 2nk __ Cos 23 2n (n—1) 2x (n—2) cos 2H — SS FAM _ ogg BN) _ cos wen -+ cos ———— = tant) ait cos 2265) aed) 5) _ ggg 22-8) a6 2 But 2 = —1 (mod 3) and 2Ak og 20k" cos —~ = cos = if k =k’ (mod 3). Since by the assumption z = 0 (mod 3), we have =—1,n—2=1,n-—4=-41, =edi,n-—6= n— and our sum takes the form on an an 2n 1—cos—>- — cos —- + cos 3~ + cos 3-— 1 =0. It remains to prove that our sum is also equal to zero in the cases when n = +1 (mod 3). The proof is similar to the previous case. 192 Solutions 241. We have sin 15°= sin (45°—30°) =sin (7 Analogously we find cos 15°. We have ie ar an ET sin 18° = sin => =cos >. But 2 sin cos + = sin sin = cos =sin =, 2si 2h cos 2 — sin AB si > sin 5 Sian Ba iD 5° Multiplying these equalities termwise, we find cos cos 24 = 4 5 5 4" On the other hand u an . 8%. x an 4 cos = 00s = 2sin Go sin Gp = 2 cos pcos =>. Thus, if we put in 2x os = sin p= cos -=2, cosy, 5 we have 1 1 Y-t=5, WHz But += (c—yP + 4ey=Z+i= Fs. Consequently, 7 5 zty=4. ~ Using this relation and the relation y—t=7, we get feeeeeCae tities oo 14+ V5 z=sin75=sin 18° =——. Now cos 48° is readily found. Solutions to Sec. 3 22. Indeed 193 sin 6° = sin (60° — 54°) = sin 60° cos 54° — cos 60° sin 54°, But sin 54° cos 36°= 1 —2 sin? 18° = 1 —25=2V8 _ 14 V5 cos 54° =V Isnt =4 V 10-25, 4 To obtain the result we have to substitute these values into the first formula; cos 6° is found in the same way. 23. Bear in mind that (1) —$, —f0. Let us, for example, also prove that cos (arctan x) = yor Put arctanz=y, tany=2Z. 13-1225 194 Solutions We have to find cos y. We have a tira 2 ay = t+ tan y=142. Consequently and 1 Vi-2! where the radical is taken with the plus sign again, since cosy > 0. cos y= Cos (arctan 2) = The rest of the formulas are proved in the same way, 24. By definition, n a > O and +28, Therefore it is obliga- tory that arctan x-+ arccot © => Likewise, let us prove that aresin z + arccos z= > First of all we have - $< aresin x +-arccos rc, On the other hand, sin (arcsin x + arccos z) = = sin (arcsin x) cos (arccos z) + + cos (arcsin z) sin (arccos z) = =24/T-#yI—e wherefrom follows that aresin z + arccosz = + : 25. First of all it is easy to prove that the quantities arctan z + arctan y and at+y arctan Tay differ from each other only by ex, where e is an integer. Indeed, tan (arctan 7) afte tan (arctan z+ arctan y) = __ _tan (arctan z)+tan(arctany) _z+y ~~ {—tan (arctan z) tan (arctany) 1—ay ° 196 Solutions But if two quantities have equal tangents, then they differ from each other by a term divisible by x. Therefore, indeed, ety arctan 2 -- arctan y = arctan Ty + en. (*) Let us find out the exact value of e. Since —F 0, ie. if 2y< 1, and V G—ay = —(U—ay) if 1—2y <0, ie. if zy>1. Therefore, cosen =1 if xzy<1, and cos en = —1 if zy > 1. Since ex can attain only the values 0, x and —x, it follows that if zy <1, then e = 0, and if zy >1, then e = +1. What sign is to be taken is decided in the follo- wing way: if zy>1 and x>0, then also y>0, then arctan x>O and arctany > 0, and arctan ft 1 andzr<0,y <0, then e = —1. 26. We have 2 Aarctan 4 = 2arctan + + 2arctan t=2 arctan 2 a 15 25 = 2arctan a- = aertan +> at arctan a= 5 5 earn ee arctan = ieaeaicn: 119 * 144 Further 120 1 arctan 775 + arctan ( —a) = ieee =arctan 8 29 arctan t=. Hiaenis 4 4 119° 239 27. Using the formula of Problem 25, we easily obtain the result, 198 Solutions 28. First of all let us notice, that since arcsin x is con- tained between — t and + 5 , and 2 arctan z lies between —x and +7, we have 3x . 2x 3 -—zS< 2arctan x -+ arcsin Te <+7.- Let us now compute the sine of the required arc, i.e. find what the expression sin (2 arctan x + arcsin sie) is equal to. We have sin (2 arctan z-} arcsin ta ) = = sin (2 arctan z) cos (aresin ea} + cos (2 arctan z) sin (aresin ee ) 7 First compute sin (2 arctan z). Put arctanz = y, tany = 2. Then sin (2 arctan z) = sin 2y = tan 2y-cos 2y. But Qtany 4—tan? y — tan? y’ tan 2y = Tptanty cos 2y= Consequently, P : 2tany 2x sin (2 arctan 2) = Thimty Tp Further cos (aresin 725) = 1—( 1. Further, it is obvious that —22 cos (2 arctan x) = to 7 2 i aresin 77) = eS sin ( 1-2) TR? Solutions to Sec. 3 199 therefore . x Qe sin (2 arctan x +-aresin Tor) = ee Pt i tao de ~ Tye Te) Te Tae Thus, the sine of the required arc is equal to zero, consequ- ently, this arc can have one of the infinite number of values: sey —8m, —20, —n, 0, +0, 2n, 38m, 4n,.... But among these values there are only three (—zx, 0 and x) lying in the required interval between — = and + a . On the other hand, z > 1 and, consequently, 2 arctan z >0 and arcsin poe > 0, and therefore the required sum 9 fat 2 arctan x-+aresin Tht will also be greater than zero and, consequently, can be equal only to x. 29. It is evident that —a0 (since in this case Wz*=z). And if z<0, then V Hence follows that —z and we have sin (arctan x +aretan + ) =-1. arctan gz arctan 4 s=H=45 e+ 2kn, where plus is taken when x > 0, and minus when z < 0. But since, on the other hand, it must be —n~ cos and, consequently, either B= kn or £48 — (2x44) 2 (k and k’ integers), i.e. either a= f+ 2kn a = —B + (2k +41) 0. Therefore we may assert that aresin x + arcsin y = 1 arcsin (x V1—y? + y V1—2") + ex, or Solutions to Sec. 3 201 where y = +1 if ¢ is even, and » = —1 if e is odd. To de- termine e more accurately, let us take cosines of both members. We get cos (arcsin + aresin y) = =cos [yaresin (2V¥1—y?+y V1—2") + en]. Hence V 1-2. Vi1—y—2zy= =(—1)® cos [aresin (x V1—y® + y V1—2)]- Further VI=PVIap—y= =(-' V1-@V FV The radicand on the right can be transformed as {—(« Vi=e+y Via = =1—22(1—y) —y? (1— 2) — 2ay VI —@- VT = 4-2) (1—y*) —22y VW1—@- V1 + ty? = =(VI=#V1— yay)’. If it turns out that 7 V1i—2-VYi-y—azy>0, then Vi-@Viae VT = -V VIRB VF ay VV Ty. Therefore, in this case (t= +4, ive. e is even. And if i V1—2#V1—y?—2zy <0, then (—)e = 4, and, consequently, ¢ is odd. 202 Solutions Let us now consider the expression 1—2-—y’. We have 1-2-— {—-P yt ey ey =(1-2) (1-y)—2y =(VI=#.V 1-9 — zy) (VI-2- V1 + ay). The quantity 1 — x? — y® can be greater (smaller) than or equal to zero. Let us consider all the three cases. 1° Suppose 1— 2? — y>>0, ie. z?+ y2><1. If the product of two factors is positive, then these factors are either both positive simultaneously, or both negative simu|- taneously. And so, we have either Vi=eV1—P—a>0, VI-FV1-#F + zy >0 or VI=#V1H¥—azy <0, YI=#PV1-¥ + cy <0. But the second case is impossible, since, adding the last two inequalities, we get VI-eV1-¥ <0, which is impossible. If, however, the first two inequalities exist, then Vi-#V1—y—azy>0. Consequently, in this case e is even. Thus, if z* + y? <1, then in our formula ¢ is even. 2° Let now 1 — 2? — y? <0 and, consequently, either VI=8V1—P—a>0, Vi-e®Vi-—¥P+ ry <0 or VI-8V1-P—ay <0, VI-#VI-¥ + ay >0. But from the first two inequalities we easily obtain zy <0. If this inequality is fulfilled, then it will obligatory be VI-#V 1-#—wy>0, and, consequently, € is even, a Solutions to Sec. 3 203 From the second pair of inequalities we get zy >> 0, and e is odd. 3° Finally, suppose 1 — 2? — y? = 0. Then again two cases are possible: either zy <0 or zy >0. In the first case /1 — 2?-//1 — y®? — zy >0, and, hence, e is even. Likewise, the second case gives an even ¢ (€ = 0), since there exists the following relation: aresinz +-aresin V1—@ =F (x> 0). Thus, we can judge whether e is even or odd. Now let us consider the value of e. We have | arcsin x + aresiny |<. Consequently | naresin (« VI—y ty V1—2) 4+ en| 1 or if zy >0, then e = +1, n = —1. To find out when e = +1 and when e = —1, let us notice that at c>0, y>0 arcsin z + arcsin y >0 and, con- sequently, —aresin(z V1—y?+y V1-—2") + ex > 0, and therefore in this case e = +4. If, however, r< 0, y <0, then it is obvious that ¢ = —1. 31. We have (see Problem 24) arccos z+ arccos ($+ + V3— 32) = = —aresin z—aresin ($+4V 3-38); on the other hand (Problem 30), aresin x + aresin ($+ +V3—38) = marcsin + ex, 204 Solutions where gaa V'1— ($+ va) + +(44+Bvima)yviee But 1-(£4 Byt-ay-Lvire and since s>t, we have 42?>{4: 327>1— 2? and V32>V1—-2. Therefore 1— ($4 Byt-ay'=4 (Via Va) = =4(/32-VT=#) 2 2_Y32), and b=> V3 | | ecgEeEAgt aresin § = z: The only thing which is left is to find y and ¢ (see Prob- lem 30). Let us prove that a4 ($+ yima) a1. We ae e+e ao tia) 4+4)32Vin# 2S 3 4 4 5 STtzet+yzi-w= 7. Consequently, y=—i,e= +1. Therefore, arccos 4-+ arccos (44+4 V 3— 3a) =1— ( $+ n) = oe Solutions to Sec. 3 205 32. We have tan A=t, tan B= + Let us compute cos 2A. Since we have 4 cost A But 2 98 cos 2A =2cos*A—1=—5— Further sin 4B = 2 sin 2B cos 2B. But 2co* B—1|= 25, 1=4 cos 2B = 2 cos’ B-\\= aR =30 sin 2B —2sin Bcos B=2 tan B cos? B= 24 _. 3 Tptam®s~ 5° Consequently, singB=2.4.2=2 and sin4B=cos2A. 33. By hypothesis we have (a+b)?=9ab or (4 )*=a. The rest is obvious. 34. Put loggn=2, logman=y- Then a=n, Hence x a=m"-a!, a =ma. Taking logarithms of this last equality to the base a, we get the required result. 35. Put z(y+2—2) _ y(e+e—y) _ 2(e+y—2) 4 logz logy Jopz OE 206 Solutions Then logz = tx (y+2—2), logy =ty(2+2%—y), log z = iz (x + y — 2). Hence y log x + « log y = 2tayz, y logz + 2 log y = 2tzyz, zlogx + x logz = 2tzyz. Consequently ylogz+arlogy=ylogz+zlogy =zlogx+-< logz, log a¥y* = log 2”y* = log 272". Finally 36. 1° Put log, a = x. Then =a. Taking logarithms of this equality to the base a, we get «logs b = 1. But x = log, a. Consequently, indeed, log, a log, b = 1. 2° We have I aa” — p, Therefore logy (oR, @) 1 a THE (ql A) !B (108 2) _ (g} fg Py (oy 0) __ logy (log, a) =bo "= logya. 37. From the given relations it follows that yi-loze = 10, zi-logy = 40. Taking logarithms of these equalities to the base 10, we get (14 —logz) logy =1, (4 — logy) logz = 14. whence loge =1— pho =1 ae Te logz and, consequently, 4 x= 101-1082, Solutions to Sec. 3 207 38. The original equality yields a? = (c — b) (c + dD). Hence 2 logesn a = logesn (¢ — bY + 4, 2 loge-» a = loge» (c + b) + 4. Multiplying these equalities, we find 4 log.+a-loge-vd = logeyy (c — b) + log.-4 (c + b) + +1 + logesy (¢ — 6) loge» (c+ b). However, loge (¢ + b) logess (¢ — 6) = 4. Therefore 4 log.+, a log.-y a = 2 log.4, a — 1 + 2 log.» a —1+42. Finally loge+y @ + log,» a = 2 logesy a log ._» a. 39. Put log, N= 2, log.N=y, logyN =2 The last equality yields (ac) =N. Hence logs N=$(1+logac), log. N =+(1+log, a). Therefore 2, 2) Fa 1=logae, —1=log.a. Consequently (7-1) (4-1)=1 or ‘at yo aay _40. We have past = 1 eae logaras. ..an® = Tegra, ay ovay ~ fogear Plogsay eas = ae 4 4 Tog, 2 * Tok +e 208 Solutions AL. Let a, = aq", by, = b + nd. Then log a, = loga+n log gq, log a, — by = = loga+nlogq —b—nd = loga—b, Hence nlogqg—nd=0, logsq=d, Bi =. And so Bagi. SOLUTIONS TO SECTION 4 1. We have (Sh —e) + (SF —0) + (FF 4) 0. Hence xz—ab—ac—be , x—ac—ab—be | r—be—ab—ac _ a+b ate + bre =0 or 4 1 1 (e—ab — ac —be) (+> + pee) =O Assuming that ; 1 1 a+b . ate a be is not equal to zero, we obtain x = ab + ac + be. If, however, 4 a+b | 4 ti then the given equation turns into an identity which holds true for any value of z. 2. Rewrite the equation as follows Co . Solutions to Sec. 4 209 We have ab Hence 1 4 4 (e-a—b-0) (tartar and, consequently, z=atbt+e It is assumed, of course, that none of the quantities a, b 1 1 1; and c, as also feta gb is equal to zero. 3. If we put in our equation 6c + 2a = A, 3b +¢=B, 4+ 6a=C,b+3¢=D, then it is rewritten in the following way A+B C+D A—-B C—D° Adding unity to both members of the equation, we find 2A eee A—B*™ C—D Likewise, subtracting unity, we get OB aucudeaD. A—B~ C—D°* Dividing the last equalities termwise, we have ee BUD? ie. 62420 2x+6a Bbc o+3¢ ° Hence (aos-t¢x) e=(pe-pe)e Boe +30 b+3e Bbpe)™ Finally 14-1225 240 Solutions 4. Add 3 to both members of the equation and rewrite it in the following way (2 41)4+ (A= 41) + (4S*41)= a 4a mo ayoye * Hence 4,4, 4) _, atbpe—2 (a+b-+e—2) (454+ 5) =4 Pe Consequently deeded 4 (atb+e—2) (24+542-aq) =? and, finally, t=a+b+e 5. Taking 3/6 + = outside the brackets in the left mem- ber, we get —— Vote -tyi. Consequently, 1 +2)? _ be tet a° eae) Hence pHi Pp (4) P ==, bye (4% yn z Further 6. 1° Squaring both members of the given equation, we find at+14+2e—1+42Y2—T =1. Solutions to Sec. 4 244 Consequently, 2/f 2 —T = 1 — 22, 4a* —4 = 1 + 42% — 4c, aaa =f: Since squaring leads, generally speaking, to an equation not equivalent to the given one, or rather to such an equa- tion which in addition to the roots of the given equation may have other roots different from them (so-called extraneous roots), it is necessary to check, by substitution, whether dis really the root of the original equation. The check shows that t does not satisfy the original equation (here, as befo- re, we consider only principal values of the roots). 2° Carrying out all necessary transformations similar to the previous ones, we find that x = is the root of our equation. 7. Cube both members of the given equation, taking the formula for the cube of a sum in the following form (A + BY = A® + B® + 3AB (A +4 B). We have at+Vipa—Vet3Y/@—a(VatVe+Y a—V2) =b. Since VatVa+Ya—-V we have eee 243 24+3YF—z-Yb=s, cae—O We assume that a and b are such that 2 (b—2a)8 0 —

1. Solutions to Sec. 4 213, Indeed, we have Vare= Vor envs (OP va, Vana-Ya-2 va Ve =Va ‘it (if Vb—1>0). Substituting the obtained values for Va +z and Va — a into the original equation, we make sure that our assertion is true. 11. Adding all the given equations, we have _ atbtetd =o. t+ytz+v Consequently va(etytet)y—(ety pa) Seber ds _ _b4e+d—2a a Likewise, we obtain _ate+d—2% abb+d—2e | afb+en2d Sd Ot a: 12. Adding all the four equations, we get 4x, = 2a, + 2ag + 2a3 + 2a,, na iets pat Multiplying the last two equations by —1, and then adding all the four equations, we find 44 dg —03— 04 X= Similarly, we get mye BM EHN Hg BORO 244 Solutions 13. Put «+ y-+2-+v=s. Then the system is rewrit- ten as follows ax+m(s—z) =k by + m(s—y) =1 cz + m(s— 2) =p dv-+m(s—v)=q so that ms +a(a—m) =k, ms+y(b—m) =1, ms + 2 (c — m) = p, ms + v (d— m) = q. Hence : __k m auacee m _ Pp m eam am Y= Fam bam Bm q m 0= am d—m * (*) Adding these equalities termwise, we find saan trem tot —ms ( Consequently s[t+m(sR Wherefrom we find s, and then from the equalities (*) we obtain the required values of the unknowns z, y, 2 and v. 14, Put Hence Solutions to Sec. 4 215 Substituting these into the last one of the given equations, we get mt+a+... $ap=a= = (a + a, + 2... +p) + A(m + mg +... + mp) Consequently, a— a4 —a,— my ping pm? and then we readily get the values of A= Ty Lay... Lp. 15. If we put Saat = fi ee Eager ieee op ere then the solution of this system is reduced to that of Pro- blem 11. Using the result of Problem 11, we easily obtain 3 2 T= T}bpe—2d’ apby+d—k? 3 3 Vv 2 pepd—m’ ~~ bpepd—2a* 16. Dividing the first equation by ab, the second by ac and the third by be (assuming abe ~ 0), we get y,e_e eee be eciyuee ptotm stoma: etek: Adding all these equations termwise, we find x a y z_1/c¢ by a ete+eorlatate): Hence 2 (2, ¥ 2) (Zi. ¥\_te b a € ¢ (4+4 ; =) (34 sazle wt te) ab * 2 a+ be—ee _@4+R—2 Consequently, [=a be. 2=—Bg— and then analogously a4 2 be _ 4k a? Ya sees Dagerec secretes be 17. First of all we have an obvious solution z = y = =z = 0. Let us now look for nonzero solutions, i.e. for 246 Solutions such in which z, y, z are not equal to zero. Dividing the first of the given equations by yz, the second by zz and the third by zy, we obtain eb aaitee, eee Ott ae tte stan, Steaua’, gece: Hence a,bi,e " styptgadtd gee, Therefore a , " bi mgt pert "gn gadtd—d, Gadt+d'—d', S=d+d'—a. Finally oe a site b Hae e “Pree Yarrie’ * ape: 18. Rewrite the system in the following way ayt-be 1 azter 14 betey 1 fey aaecet eco ae toacea ea ae eee eta pate ge Hence sO rae Maan ge eae ee eeeaee ea erate a re gongs? pitts iat Consequently (see the preceding problem) ___ 2a%be 2ab ec ___ Dade? T= gepab—be’ Y=depab—ac’ 7 bepac—ab’ 19. The obvious solution is x = y =z =0. Dividing both members of each equation of our system by zyz, we get og tea eee deed ata ye? why ae 4 4 4 wile aw Adding pairwise, we find 2 fi 4 2 4 4 2 a if wiate: poate grata: Consequently 2a2p2 2b2e2 2a%ed wW=aTR y= per? w= fea ' (*) Solutions to Sec. 4 247 Multiplying the equalities, we obtain 22,0 Batbtct PUP ELD (Be (BFA Hence ae 2 V2 rer PSE TET TTA Using the equality 2a2b2 Y= BTR? we find for z two values which differ in the sign. By the obtained value of z we find the corresponding values of y and x from the equalities (*). Thus, we get two sets of values for x, y and z satisfying our equation. 20. Adding all the three equations, we find (e+y+2) (at+b+c) =0. Hence z+y+2=0, whence a—b a—-c b—a pe 21. Adding all the three equations termwise, we get (b + ¢) x + (c+ ay + (a+ b) zg = 2a® + 26% + 20°. Using the given equations in succession, we find 2(b +c) a = 263 + 2%, 2(c + a) y = 2a® + 2c*, 2 (a + b) 2 = 2a? + 203, whence c=b—be+c%, y=a?—ac+ec, 2 22. Consider the following equality vet rhe t rhe = — een ee era Let us transform the equality, by reducing its terms to a common denominator and then rejecting the latter. We get a second-degree polynomial in 0 with coefficients depending on z, y, 2,4, p, v, a, b, c, which is equal to zero. If now we 2? — ab + b?. 218 Solutions substitute successivly 4, » and v for @ into the original expression, then, by virtue of the given equations, this expression (and, consequently, the second-degree polyno- mial) vanishes. However, if a second-degree polynomial becomes zero at three different values of the variable, then it is identically equal to zero (see Sec. 2) and, consequently, the equality oy y + zy _ _ (8A) (0p) (0—v) a+ ' d+0 ' +8 0 Fa) 0+4)0Fe) (by virtue of existence of the three given equations) is an identity with respect to 0, i.e. it holds for any values of 0. Multiplying both members of this equality by a + 6, put 6 = —a. Then we find ga th C+ at) (a=) (ae) : Likewise we get aH OtA btu (b+¥) 2 (et) (e+H) (e+) iY eauate (boa ey baa) seauHeeiecaee eae ay emo) eas Of course, we assume here that the given quantities A, p, v, as also a, b and c, are not equal to one another. 23. The given equations show that the polynomial ao + za® + ya + 2 vanishes at three different values of a, namely at a = a, at a = band at @ = c (assuming that a, 6 and ¢ are not equal to one another). Set up a difference a + za* + ya + 2 — (a — a) (a — dD) (a — 0c). This difference also becomes zero at a equal to a, b, c. Expanding this expression in powers of a, we obtain (gc +a+b+c) a+ (y—ab—ac—be)a+ + 2+ abe. This second-degree trinomial in « vanishes at three different values of a, and therefore it equals zero identically and, consequently, all its coefficients are equal to zero, i.e. ata+tb+cec=0, y—ab—ac—be=0, z+ abe =0. Solutions to Sec. 4 219 Hence r= —(a+b+9, y =ab + ac + be, 2 = —abe is the solution of our system. 24. We find similarly t=—(a+b+e+d), x =ab+ac + ad + be + bd + cd, y = —(abe + abd + acd + bed), 2 = abed. 25. Multiplying the first equation by r, the second by p, the third by q and the fourth by 4 and adding, we get (@ + ag + ap +1) x + (0% + Bg + bp + ry + + (P+ eq teptryz+ (P+ Pq + dp+rnu= =mr-+np+kqg+l. Let us choose the quantities r, p and q so that the follo- wing equalities take place B+ bg + bp +r=0, + cq + ep +r=0, @+dq+dp+r=0. Hence, we obtain (see Problem 23) q=—(b+e4+d, p=be+bd+ed, r= —bed, and, consequently z= where N = —mbced + n (be + bd + cd) —k(b+e+ da) +1. As to the equality a& + aq + ap +r=(a—b) (a—c) (@—ad), it follows readily from the identity o@ + ga? + pa+r = (a — b) («a —c) (4 — 4d). 220 Solutions To find the variable y, the quantities q, p andr are so chosen that the following equalities take place a@+a’g+ap+r=0, O+cq+ept+r= @®+dq+dp+r The remaining variables are found analogously. 26. Put m+aat+... +e, =. Adding the equations term by term, we get s+2s+3s+... +ns=a,t+ag+... +a). But 44+24+34+... tn = 204) (an arithmetic pro- gression). Therefore 2 S=TaED (a, + ag eters a,) =A (for brevity). Subtracting now the second equation from the first one, we find zy + 22+ 23 eects Ly ng, =a ap. Hence nz =A+a,—aQ, and aya Subtracting the third equation from the second, we get A+a3—ay = 4tSo% and so on. 27. Put Mm+aet... $y =. Then we have —s + 2x, = 2a, —s + 42, = 4a, —s + 8x3 = 8a,..., —s+ 2"z, = 27a. Hence Hats, mats, Bats, mat. Solutions to Sec. 4 224 Adding these equalities, we get s=nats (S++ i +4). But 4 4 1 4 gtetetealg: Therefore s = 2"na. Consequently qy sa+5sa+2 na =a(1 + n-2n-), m=at+7=a+42"na=a(t $2.2") and so on. 28. Let H+, t+az3+...+ ey, =s=1. Thea s—2%=2, s—23=3,...,8—2,4~=n—1, sS—-X =n Consequently (since s = 4) t= —1, x3 = —2. +) Zn = —(n — 14). Hence ttayt... + =—-MA+2+...+@—tl= _ __ n(n—t) =- >. Finally a =1—(e tat... tm) =14+20-9, 29. Suppose the equations are compatible, i.e. there exists such a value of x at which both equations are satisfied. Substituting this value of x into the given equations, we get the following identities ax+b=0, ax+b' =0. Multiply the first of them by b’, and the second by b. Sub- tracting termwise the obtained equalities, we find (ab’ — a’b)x =0. 222 Solutions If the common solution for z is nonzero, then it actually follows from the last equality ab’ —a’'b =0. If the common solution is equal to zero, then from the ori- ginal equation it follows that b=0b' =0, and therefore in this case also ab’ —a’'b = 0. And so, in both cases, if the two given equations have a common solution, then ab’ — ab = Hence, conversely if the condition ab’ —a’b =0 is satisfied, the two given equations have a common root (the coefficients of the equations are proportional), and, consequently, they are compatible. 30. To prove that the given systems are equivalent it is necessary to prove that each solution of one of the systems is simultaneously a solution for the other system. Indeed, it is apparent, that each solution of the first system is at the same time a solution for the second system. It only remains to prove that each solution of the second system will also be a solution for the first system. Suppose a pair of numbers z and y is the solution of the second system, i.e. we have identically +8 =0, m§ + m’t' = 0, Esar+byt+e, B =ae+ byte’. Multiplying the first equality by m’ and the second by /’, and subtracting them termwise, we find (Im! — ml’) & = 0. Likewise, multiplying the first equality by m and the se- cond by 1, and subtracting, we get (im' — ml’) ’ = 0. where Solutions to Sec. 4 223 But since, by hypothesis, lm’ — ml' £0, it follows from the last two equalities that —=0 and b =0, i.e. az + by+ce=0 and az+ byte =0. Thus, the pair of numbers z and y, which is the solution of the second system, is simultaneously the solution of the first system. 31. Multiplying the first equation by b’ and the second by 6, and subtracting termwise, we find (ab’ — ab) x + cb’ —c'b =0. We get similarly (ab’ — a'b)y + c'a—a'e =0. These two equations are equivalent to the original ones. It is evident that if ab’ — a’b #0, then there exists one and only one pair of values of x and y satisfying the last two equalities, and, consequently, the original system as well. 32. Multiplying the first equality by b’ and the second by b, and subtracting, we find (ab! — a'b) x = 0. Since, by hypothesis, ab’ — a’b & 0, it follows that r = 0. In the same way we prove that y = 0. 33. From the first two equations we get c’b—cb! __ ae—e'a “aba? Yah If the three equations are compatible, then a pair of num- hers x and y being the solution of the system of the first two equations must also satisfy the third equation. Therefore, if the three given equations are compatible, then there 224 Solutions exists the following relation wn c'b—cb’ no ae ac—c'a ‘a waae th ae t eH 0 or a” (c'b — cb’) + b" (a’c — ca) +c” (ab’ —a'b) = 0. (s) Conversely, the existence of this relation means that a solution, which satisfies the first two equations, satisfies the third one as well. This relation may be rewritten in the following ways a’ (cb" — c"b) +b’ (ac” — ca”) +c’ (ba” — b"a) = 0, a (c"b’ — c'b") + b (a"c' —c"a') + ¢ (b"a’ — ab’) = 0. Hence it follows that the solution of each pair of the three equations is necessarily the solution of the third equation, ie. our system is compatible provided the condition (+) is observed. 34. Subtracting from the first equality the second, and then the third one, we find (a— b)y + (@® —b)z2=0, (@—c)y + (@—c%)z = 0. Since a — b #0 and a—c 0, we have the following equalities yt(at+b)z=0, y+ (at+ez=0. Subtracting them term by term, we have (b—c)2=0. But by hypothesis b — c 0, therefore z = 0. Substitu- ting this value into one of the last two equations, we find y = 0. Finally, making use of one of the original equations, we get = 0. 35. Multiplying the first equality by B, and the second one by B, and subtracting them termwise, we get (AB, — A,B) x + (CB, — C,B)z = 0. (4) We find analogously (AC, — A\C) x + (BC, — B,C) y = 0. ) Solutions to Sec. 4 225 Suppose none of the expressions AB, — A;B, CB,—C,B, AC, — A\C is equal to zero. Then we get z a C\B—CB, ABy—A\B [multiplying both members of the first equality by the product (AB, — A,B) (C,B — CB,)\ and : eee eee C;B—CB, ~~ CA,;—AC, * Thus, in this case the required proportion really takes place. Let now one and only one of the expressions AB,— A,B, CB,—C,B, AC,—A,C vanish. Put, for instance, CB; —C,B = 0. Then from equalities (1) and (2) we get 2 = 0. Further, suppose that two of the mentioned expressions, for instance, CyB — CB, and CA, — AC, are equal to zero, and the third one, i.e. AB, — A,B is nonzero. We then find = y = 0. In these cases our proportion, or, more precisely, three equalities, z= (CyB —CB)), y =4 (CA; — AC), z = 4 (AB, — A,B), will also take place. Thus, in these cases two given equations determine the variables z, y and z “accurate to the common factor of pro- portionality”. If all the three quantities AB, — A,B, CB,;—C,B and AC, — A,C are equal to zero, then there exists the following proportion eA eee cs Aji Byers 15~1225 226 Solutions In this case the two equations (forming a system) turn into one, and nothing definite can be said about the values of the variables z, y and z which satisfy this equation. 36. From the first two equations (see the preceding pro- blem) we get eee 2 oe be—a® abe * Hence c=A(ac— bd), y=A(be—a’), 2 =A (ab— cc’). Substituting these values into the third equation, we find b (ac — b*) + a (be — a?) +c (ab — c*) = 0 a + b? + 3 — 3abe = 0. 37. Multiplying the first two equations, we get or edie aR ae oe BR The same result is obtained by multiplying the third equa- tion by the fourth one, which shows that if there exist any three of the given equations, then there also exists a fourth one, i.e. the system is compatible. To determine the values of x, y and z satisfying the given system proceed in the following way: equating the right members of the first and the third equations, find y y a (144) =u (1-4). Solving this equation with respect to y, we have —ptos y=b wah Substituting this into the first two equations, we get zs Paced Hence Solutions: to Sec. 4 227 38. Rewrite the system in the following way a (c+ py) + b(@ + gy) = ap? + bg? ap (x + py) + bg (x + ay) = ap? + be ap** (a + py) + bg' (x + gy) = apt? + bgt. Now it is obvious that the system is equivalent to the follow- ing two equations rt py=P otgy=a, and, hence, the system is compatible. 39. We have Serr ted ete Ly = Az — Tp = Op — a + YH, Ty = A3 — X3 = Az — Ag + 4 — XY, Ty = An4~— Ano +... Ha, Farry. It should be noted that in the last equality the upper signs will occur when n is odd, and the lower signs when n is even. Consider the two cases separately. 4° Let 7 be odd. Then Ey = Ant — Ang +--+. + a2—-a +X. On the other hand, Zz, + 2, = ay. From these two equalities we get Gn —@n_4-}-Gn_g—..- —An +a pie nat tna ata and, hence, 2 = 2° Let now n be even. Then Ty = Aq — Ang b+. — Og bay — %. 228 i Solutions On the other hand, Ly = An — Ty. Consequently, for the given system of equations to be com- patible the following equality must be satisfied Gn — Ang +... — Og + ay = ay, i.e. Oy dyin +. $y = Ont tapas t+... tay (the sum of coefficients with even subscripts must equal the sum of coefficients with odd subscripts). It is apparent that in this case the system will be indeterminate, i.e. will allow an infinite number of solutions, namely: a=A, tT =a—A, Ig =ag—4 +h, t= a3—a,-+a,—h, Ty = An — Aj-g + +. +43 ~ a2 + a4,—A, where A is an arbitrary quantity. 40. From the first two equations we find ~a=d —d Substituting this into the third equation, we have a (hg (see soe) +e (SS -S)+ 7 (see) } = 4-9) 0-9) (e-0). After simplification we get a ve 2 b @ a goa (Goa 7 aa) +a (eee) + c a By d(a—b)(b—c) (a—c) too ( angaacbod )= (ad) (6a) (ed) * Therefore 4 = —(a—@ (b—@ (c— 4), Solutions to Sec. 4 229 and, consequently, x = (a — d) (b — c) (db + de — be), y = (b — d) (c — a) (de + da — ac), 2 = (c — d) (a — b) (ad + db — ab). A1. Solving the last two equations with respect to z and y, we find otnaomana ate —_ =} (m=o) yt b= Hence ofan 9 (nay (an EE, Analogously y+l=(I—b) oa . Substituting the found values of « + a and y + [ into the first equation, we see that it is a consequence of the two last equtions. Thus, the system is indeterminate, and all its solutions are given by the formulas __(c=m) (n—2) =) (me) Dea eae esearer att oe crac pear b for an arbitrary z. 42. From the second and the third equations we have (1 —k)a + hy = —[ +h) x + (12 —&) yl, hence, taking into account the first equation, (5 — k) y = = 0 wherefrom either k = 5 or y = 0 (hence x = 0), which yields (substituting into the second equation) k = —1. 43. We have sin 2a = 2 sin a cosa, sin 3a = sin a (4 cos? a — 1), sin 4a = 4 sin a (2 cos* a — cos a). 230 Solutions Therefore the first of the equations of our system is rewrit- ten in the following way a + 2y cos a + 2 (4 cos? a — 1) = 4 (2 cos® a — cos a). The remaining two are similar. Expand this equation in powers of cos a. We have 8 cos? a — 4z cos? a — (24y + 4) cosa +2—2=0. Putting cos a = ¢ and dividing both members by 8, we get esp th yy 2 =0. (*) 8 Our system of equations is equivalent to the statement that the equation (*) has three roots: ¢ = cosa, t = cos b and t = cosc, wherefrom follows (see Problem 23) = —cosa+cosb-+cose, © wl yt a m—2 andi cos acos bcos c. = —(cos acos b+ cosacesc-+cosbcosc), Therefore the solution of our system will be x = 2 (cosa + cos b + cose) + 8 cos a cos b cos ¢, y = —2 — 4 (cosa cos b + cos acosc + cos b Cos c), 2 = 2 (cosa + cos b + cosc). 44, Put a b Sind ~sinB sin Since A + B+ C =n, we have sin A = sin(B + C) = sin BcosC + cos B sin C. But from the given proportion we have ; a 7 6 ‘ es sinA= >, sinB= >, sinC=—7. Substituting this into the last equality, we find a=beosC + ccosB. The rest of the equalities are obtained similarly. Solutions to Sec. 4 234 45. Expressing a and b in terms of ¢ and trigonometric functions (from the first two of the given equalities), we get (cos A~ (1) 2) Substituting (1) and (2) into the third equality and accom- plishing all necessary transformations, we find 1 — cos? A — cos? B — cos* C — 2 cos A cos B cos C = 0. Let us now prove that A+B+C=n. Transform the obtained equality in the following way b= c (cos B+ cos A cos C) on sin? C ; cos? A + 2 cos A cos B cos C = = 1 — cos? B — cos? C — cos? B cos? C + cos? B cos? C, cos? A + 2 cos A cos B cos C + cos? B cos? C = = 1 — cos? B — cos® C (1 — cos? B), (cos A + cos B cos C)? = sin? B sin? C. But since we have obtained [see (1)] that bh sin2 cos A-+.cos Beos C=C > 9, we have cos A+cos BcosC =sin BsinC, cos A =sin B sin —cos B cos C = —cos(B +C), cas A +008 (B+C) =2 cos 4+24© cos ABE _ 9, wherefrom follows that either ALB+C _ x agree a eee eg or A-B— 4 Oy a sl ee, 232 Solutions where / and l’ are integers. Let us first show that the second case is impossible. In this case we would have A—B-—C=(2l +1), B=A—CH— (2l’4+1)a, cos B = cos (A — C — n) = —cos(A —C) = = —cos A cos C — sin A sin C. Consequently, cos B + cos A cos C = —sin A sinC <0 which is impossible, since we have obtained (2) in? cos B+-cos AcosC = 25186 5, 0. Thus, there remains only the case A+B+C=(21+1)a. However, by virtue of the inequalities, existing for A, B and C, we have 0<214+1<3, 2+1=1 ie. and A+tB+Ce=n. It only remains to show that pact dap ieLdinetgaaa sin A sinB sinc * We have shown that cos A + cos B cos C = sin B sin C. On the other hand, cos B + cos A cos C = cos (n — A — C) + cos A cosC = = —cos(A + C) + cos A cosC = = sin A sinC. Using this equality and also equalities (1) and (2), we easily obtain the required proportion. 46. Let us first show that equation (2) follows from equa- tions (1). Multiplying the first of equations (1) by a, the second by 6 and the third by —c and adding them term- Solutions to Sec. 4 233 wise we get a + b? — c? = 2ab cos C, i.e. the third of equations (2). Likewise we obtain the re- maining two of equations (2). To obtain equations (1) from equations (2) add the first two of (2). Collecting like terms, we find 2c?—2 be cos A — 2 ac cos B = 0. Hence c= bcos A +acosB, i.e. we get the third of equations (1). The rest of them are obtained similarly. 47. From the first equality we get cos a—cosb cos c¢ A= ee cos sind sine Hence sin? A= 1—cos’ A= ___sin®b sin? e— (cos a—cos b cos c)? a sin? b sin2c fe __ (1—cos? b) (4 — cos? ¢) — (cos a—cos 6 cos c) ma sin? b sin? ¢ it 1 —cos? a—cos? b—cos? c-}-2 cos a cos b cosc oe sin?d sin2c¢ 7 Consequently sin? A 4—cos? a—cos? b—cos* c+ 2 cos acos bcos c sinta sin? a sin? b sin? c = Since the given formulas turn one into another by means of a circular permutation of the letters a, b, c, A, B, C, and as a result of this transformation the right member of the last equality remains unchanged, we actually have sin?A _ sin?B in? C in? ¢ sinta ~sin2b But the quantities a, b,c and A, B, C are contained between 0 and x, therefore sind sinB sine sina ginb sine 234 Solutions ~ 48. 1° Let us take the last two of the equalities (+) from the preceding problem. We have cos b — cosc cosa = sina sinc cos B, —cos a cos b + cose = sina sin b cos C. Multiplying the first of them by cos a and the second by 1 and then adding, we find —cos c cos? a + cosc = sina sinc cos B cosa + + sina sin b cos C. Hence cosc sina = sinc cosa cos B + sin b cos C. But since it was shown in the preceding problem that from the equalities (*) follows the proportion b nB sina _ sind sine sin * in the last equality we can replace the quantities sin a, sin 6 and sinc by ones proportional to them. We get cosc sin A == sin C cosa cos B +- sin B cos C. It is apparent, that there exist six similar equalities. Let us take one more of them, namely, the one which also contains cos¢ and cosa. It will have the form cos asin C = sin A cose cos B + sin B cos A. (This equality can be obtained in the following way: mul- tiply the second of the equalities (*) by cos c¢ and the first one by unity, add them, and in the obtained equality repla- ce sine by sin C and so on.) Thus, we have cose sin A = sin C cosa cos B 4+ sin B cos C, cos asin C = sin A cosccos B + sin B cos A. Eliminating cos c, we find cos A = —cos B cos C + sin B sin C cos a. The rest of the equalities are obtained from this one using a circular permutation. 2° The formulas (+) of Problem 47 make it possible to express cos A, cos B and cosC in terms of sina, sin 6, Salutions to Sec. 4 235 sin c and cos a, cos b, cos c. Let us find the expressions for sin $ and cos a We have cos a—cos b cos ¢ sindbsine 2sint 4 —1—cos A= A aia 2cos* 5 sin bsine Hence sarge Oi ean sin >= , b a = foul Hees sin sine cae . + . B Similar expressions are obtained for sin. cos cata Cc . AB , sin-z , cos. Now compute sin z+ We have See Ae Bese EEA B, A. B sin—J— = sin = cos > + cos sin = = — afbpe. aph—e sin SE sin SS sinasinb sin 2te=2 sin BE=* a—b ‘ fee ome eee | pe sine tine irehedy yeu ee cos Thus, we have obtained the following formula 236 Solutions Likewise we find cos AEE Therefore and, consequently, Hence cos cos and, consequently, & tan tan ( (4) Using the formula a+b s cos 4t2 — Seti sin & ‘ cos < faa 2 we find analogously e Gite vai eps tan = cot (+--+) =tan-> tan (2) Multiplying the equalities (f) and (2) termwise and extracting the square root, we get ae P ta 7 tanteny/ tant tan z— tan Solutions to Sec. 4 237 49. We have a {tan (x+ y)—tan (2+ B)] +6 [tan (xz + «) — tan (x+ y)] + +e [tan (w+) —tan (x--a)] =0. Hence asin(y—B) + bsin(a—y) : “cos (ef) cos (2p) | COs(e+a)cos(zy) csin (B—a) +s (e+) cos (z7-F a) asin (y —f) cos (x + @) -++- b sin (@ — y) cus (x + B) -- -|-¢ sin (B—@) cos (x+y) =. Finally tan 2 q £510.28) cos ab sin (ay) cos B-Le sin (Ba) cosy An T= —Tsin (Y—p) sin a--bsin(a—y)sinP-pesin(p—a)siny * 50. We have cos? 3 = 1 : 4+-tan? Therefore 4—tan? 5 cos x= 2cos*— 1 =——. 22 tp tan? > - —tan2—- ode : 2 tan Z 4—tan' 2 2 tan sin «= tan x cos ¢==———=_—. —____— _ ___-_, —tan? -tan? = ace 4A —tan’ z 1 -{- tan z 44 tan Z It is obvious that if tan + is rational, then sin z and cos x 2 are also rational. Show that if sin x and cos x are rational, then tan = is rational too. From the first relationship we have (14 tan?) cos x= 4—tan? >. Hence tan? > = 238, Solutions Consequently, if cos x is rational, then tan? + is rational as well. But from the second equality it follows that meee z 2tan $= sin x (1+ tan’). Hence, it is clear that if sin x and cos z are rational, then tan ¥ is also rational. 54. Since sin? 2 + cos? = 1, we have sint z+ costa + 2 sin? 2 cos? x = 4, sint xc + cost = 1—2 sin? x cos® z, Therefore the equation is rewritten as 1 —2sin? xz cos? z =a, 2 sin? x cos*z = 1 —a, sin? 22x = 2 (4 — a), sin2z2 =~ + Y20 —a). For the solutions to be real it is necessary and sufficient that +N 0 for any A. Since D (A) is a second-degree trinomial in 4 and D (0) = (a — c)? > 0, it is sufficient to prove that the roots of this trinomial are imaginary. And for the roots of our trinomial to be ima- ginary, it is necessary and sufficient that the expression 4 (ab + ad + be + de — 2bd — 2ac)? — 4 (a — c)* (b — d)? Solutions to Sec, 5 251 be less than zero. We have 4 (ab + ad + be + de — 2bd — 2ac)? — ~4(a—o(b— a? = = 4 (ab + ad + be + de — 2bd — 2ac — — ab + cb + ad — cd) x X (ab + ad + be + de — 2bd — 2ac + ab — —cb —ad+cd) = = —16 (b — a) (d —c) (c — b) (d — a). The last expression is really less than zero by virtue of the given conditions a 0. We have 4 (a+b +0)? —12 (ab + ac + be) = =4 (a? + b? + c® — ab — ac — be) = = 2 (2a? + 2b? + 2c? — 2ab — 2ac — 2bc) = = 2 {(a? — 2ab + b*) + (a? — 2ae + c%) + + (b8 — 2be + c4)} = = 2{(a — b) + (a— 0) + (b — 9) BO. 7. Suppose the roots of both equations are imaginary. Then pe —4q<0, pi— 4m <0. Consequently p+ pi — 4g — 4g, <0, p? + pi — 2pp, <0, (p — ps)? <0, which is impossible. 252 Solutions 8. Let us rewrite the given equation as (a+ b+) 2? — 2 (ab 4+ ac + be) x + 3abe = 0. Prove that its discriminant is greater than or equal to zero. We have 4 (ab + ac + be)? — 12abe (a +b +c) = = 2 {(ab — ac)* + (ab — be)* + (ac — be)*} > 0. 9.-By properties of the quadratic equation we have the following system P+q=—P, PY=4 From the second equation we get a(p ~1) =0. Hence, either g = 0 or p = 1. From the first one we find if ¢g=0, then p=0; if p=1, then q = —2. Thus, we have two quadratic equations satisfying the set requirements z=0 and 2 +2—2=0. 10. We have e+ y+ 2 — zy —az— yz = =F (Qa® + 2y? + 22% — Qay — 2x2 — 2yz) = 4 = ze —yP t+ (@— 7? + y— 2} >0 (see Problems 6 and 8). But we can reason in a different way. Rearranging our expression in powers of z, we get 2?—(y+2)r+y?+ + 2? — yz. To prove that this expression is greater than, or equal to, zero for all values of z, it is sufficient to prove that: firstly yt+e—yz>0 and, secondly, (y + 2)? — 4 (y? + 2? — yz) <0. Solutions to Sec. 5 253, It is evident that there exist the following identities yp+e—y= (y-$2) +38, (y + 2)? —4(y? + 2 — yz) = —3 (y — 2)? and, consequently, our assertion is proved. 41. We have a 2 2 P+ype—fae+ y+ (a—2—yk—s. It is necessary to show that the last expression is greater than, or equal to, zero for all values of x and y. Rearranging this polynomial in powers of y, we get P+ (e—ay+o—ar+—. It remains only to prove that for all values of x P—ar+F50, (@—a)'—4(228-ar +2) <0. We have 2 eaar+S=(2—$) + p00, (e~a)*—4(2*~ar+ 4) as (2-52)"<0, which is the desired result. However, the proof can be car- ried out in a somewhat different way. Indeed, it is required to prove that 3x? + 3y? + 327 > a? zt y® + 22 + Qry + Qaz + 2yz = a?. Consequently, it suffices to prove that 3a3 + By? + 32? D> a? 4+ y? + 2? + Qey + ez + Qyz or if Qa? + Qy? + 22? — Ary ~ 2az — 2yz > 0. And this last inequality is already known to us (see, for instance, Problem 6). 12. See the preceding problem. 13. By the properties of quadratic equation we may write a+Bp=-—p, aB=¢. 254 Solutions Therefore Ss = —p. Since a and f are roots of the equation e+ pet ge =0, we have w+ patq=0, B+ pB+q=0. Adding these equalities term by term, we find So + psi + 2¢ = 0. Hence 8. = —ps, — 2g = p® — 2g. Multiplying both members of our equation by 2", we get oh? 4 prkt 4 gok — 0, Substituting @ and B and adding, we find Sate + PSati + qsx = 0. Putting here k = 1, we have $3 = —pS2 — 954. Further 83 = —p (p? — 24) + ap = 3pq — p® Likewise we find 5, = pt — 4p?q + 29°, 85 = —p® + 5p*q — 5pq’. To obtain s_s, let us put in our formula k = —1. We have 5, + pS + gs, = 0. But S$ = 2, 4 = —p. Therefore 1=+P—2p=—p, s4=—2. Likewise we get s_,, $_3, s_, and s_s;. However, we may pro- ceed as follows spe tp be tH ak pk («)* qt wherefrom all the desired values of s_, are readily found. Solutions to Sec. 5 255 14. Let Vaty p=o. Then otaat 4 GB+6 Y OR +A Y of +B. But a+B=—p, ab =¢. Consequently of = —p4+6VG+4 7 of (Va+VB). But (Va+VB)'=a+8+2Vab=—p+2Va, therefore & _ oo o=V —pt6Va44/ o-V —p+2V a 15. Let x be the common root of the given equations. Multiplying the first equation by A’, and the second by A and subtracting them termwise, we get (AB! — A’B) x + AC’ — A'C =0. Likewise, multiplying the first one by B’ and the second by B and subtracting, we find (AB’ — A'B) 2? + BC' — B'C = 0. Take the value of z from the first obtained equality and substitute it into the second one. Thus, we obtain the, re-. quired result. 46. Adding all the three equations termwise, we find (@tyt+ pat hfe, Hence ttytz=tVEPE TE. Consequently a2 Be eee epee’ my 256 Solutions 17. It is obvious that the system can be rewritten in the following way (z+ 2) (e+ y) =a, Y+2y¥ta=b, @+aGty =e. Multiplying these equations and extracting a square root from both members of the obtained equality, we have (© +2) (@4-y) y+2) = + V abe. Hence Vabe + Yabe oe ete= dV Vam =t yiz=a z 1 ety=t Adding these equalities termwise, we find a+yf+2= But since y+e= we have Analogously eV abe pA dated Vaeyi,1 1 yet (etecg)) m4 (F+e-s) simultaneously taking either pluses or minuses everywhere 18. Put yta=y, tc+z2=8, y+z=a. Then our equations take the form y+ B = ayb a+ y= bay B+ a = cap. Solutions to Sec. 5 257 Solving this system (see Sec. 4, Problem 17), we find the solutions of the original system z=y=z=0 4 4 1 ay t=> (= az moor) , 4 —s) : 1 ae 2p=a+b4+e 19. Adding unity to both members of the equations, we get where A4tyt2z+y=a+t, 4ta2+2+a=6+1, 44+-a+ytay=c+i1 (Q+ydt)=a+4, “Q+2)4+2) =b+414, Q+yd+2 =e+4. Multiplying these equations, we get (1+ a) (4 + yP? (A + 2)? = (t+ a) (1 + 0) (1 + 0) or or (+2)44+y (4+2)=+V +a (1+) (+0). Consequently, (4+ 6) (1+ e) (+4) (+e) i+e= +f Ge) Tansee ee yf eats ret 142= af A. 20. Multiplying the given equations, we obtain (xyz)? = ab cx yz. First of all we have an obvious solution rs = y = 2 = 0. Then xyz = abe. 17=1225, 258 Solutions From tke original equations we find xyz = ax*, xyz = by®, xyz = c2* Hence az? = abe, by? = abe, cz* = abe, e=be, y® =ae, 2 = ab. Thus, we have the following solution set z=Vbe, — y=Vac, t=—Voe, y=—Vae, r=Voe, y= —Vae, z=—Vbo, y=Vac, z= —Vab. 21. Adding the first two equations and subtracting the third one, we get 2a = (c + b — a) xyz. Likewise we find 2y? = (c + a — b) xyz, 227 = (a + b —c) yz. Singling out the solution z=y=2z2=0, we have Qn = (c+ b—a) yz, 2y = (c +a — Dd) az, 22 = (a + b —c) zy. Then proceed as in the preceding problem. 22. The system is reduced to the form ay + 2z =a’, yz + yx = 0%, wet zy = c%, Adding these equations term by term, we find ay +22 yas (e+e te). Taking into consideration the first three equations, we get _ Bera? e+e Re y=—y > waw=—z ye caro recat Solutions to Sec. 5 259 Multiplying them, we have ee apa / CES, Now we easily find = (ac? — B) (a? + b2@~c?) a 8 (62+ c2—a2) , a (@2 +b — c2) (2 + 2 — a?) Yat Vera ) af CLARE +f, 23. Adding and subtracting the given equations term- wise, we find +P =aliety+bety=@+o)e+y, Pe —y=a(c—y)—b(@—y) = (a— Bb) (e@ —y). Hence (2 + y) (2? —ay + yy? —a—b) =0, (@—y)(@+ay+y—atd)=0. Thus, we have to consider the following systems Va+y=0, r—y=0; 2xety=0, 2+ay+y?—-a+b=0; S2—y=0, 2-—zy+y*—a—b=0),; xr —cy+y>—a—b=0, 224+ 2y+ y2?-a+ +b=0. The first three systems yield the following solutions 1 c=y=0; 2 2r=tVa—b, y= FVa—b; 3° +Vayo. 3° gs 260 Solutions The last system is reduced to the following one e’+y=a, z= —bd Solving it, we get z=4(eV¥a—B4+nV a2), y= (eVa—B—nVay2), where ¢ and take on the values +1 independently of each other. Thus, we get four more solutions. 24. Reduce the system to the following form (@+y—2)(@+2—y) =a, Y+z—)yYtz—) =), @+z2-wWe@+y—2)=c Multiplying and taking a square root, we get (@+y—2) @+2—y Y+2—2) =4Vabe. Further yt2—2a f=, stay. t+y-2S4+ y=. Consequently on (V E+ 2), nx (V EHV), roe (V EHV 9). 25. Put zty _ ee eye ryyrey prepay ™ = y2-Pbaa =P Then the system takes the form by +ch =a, ca+ay=b, aB+ba=c Solutions to Sec. 5 264 or pga tng rae gear gee a daa teu crc a elite atl llicet tee aptih depnitetat tape Therefore 4 a+ b2+ 02 S4$tta5 oe and, consequently, B24 2a? a2 B2 a? 2 —c2 ee Doe eS Peeeeee Dae) ra ab Further Sci ato ect aE ery Pa eee vay ae Finally 4 4 ¢ tyr Analogously, we find 1 y Hee. aa, $tt-op ou sieaaa? wherefrom we find z, y and z. 26. Multiplying the first, second and third equations respectively by y, z and z, we get cx + ay + bz = 0. Likewise, multiplying these equations by z, z and y, we find bx + cy + az = 0. From these two equations (see Problem 35, Sec. 4) we obtain Sea ray Hee eeeee Eee Boe Rae = Ba ie. z= (a?— be), y = (b® —ac)h, 2 = (c? —ab) dh. Substituting these expressions into the third equation, we find Re c 1 “~ (2 — ab)? — (a2 —be) (h2— ac) a3-+ 3-3 — Bade Now it is easy to find z, y and 2. 262 Solutions 27. Rewrite the system as follows (y? — 22) + (# — ay) =a (2? — yz) + (# — ay) = 6 (a? — zy) + (y? — 22) =e. Hence 2 7 _b4+e—a ye 7 ypu ry = i.e. we have obtained a system as in the preceding problem 28. Subtracting the equations term by term, we have (@—y@+yt2) =a, (@@—a(etyt+a=2—a. Putz +y+z2=4, then (c—y)t=B—a@, («—-a2t=2—a Adding these two equations termwise, we have [8x2 —(cx+y+2)]t=b? 4 c?— 2a. ate—b og pF : a+b—e z Hence _ P44 2202 eee 3 Analogously Pp arper@— 22 _ Pp a2 42202 y= 3 ee 3 : Substituting these values of x, y and z in one of the equations, we find HAR +A A+ att + ct — — ab? — ch — Be? = 0. Hence pa PPL e VSP PEAK erbs AGH PECTIN I), ; Knowing ¢, we obtain the values of z, y and z. 29. We have the following identities (ety + 2)? — (+ y+ 8) = 2 (ay + az + y2), @+yt7—@+yt2)= = 3(@+y +4 2) (cy + a2 + y2) — 3zyz. Solutions to Sec. 5 263 Taking into account the second and third equations of our system, we get from the first identity ay + xz + yz = 0, From the second identity we have xyz = 0. Thus, we obtain the following solutions of our system r=0, y=0, 2=4; z=0, y=a, 2=0; ody Ogle lO: 30. Let x, y, z and u be the roots of the following fourth- degree equation at — pa? + qa? —ra + t= 0. (s) Put ak + yk + ch + uh = sy. Then Sy — PS3 qS2 TS, t= But by hypothesis Ss =a, ss=a, 8 =a, 4 =a. Therefore, the following identity must take place at — pa? + ga*—ra+t=0, i.e. the equation (*) has the root @ = a, and therefore one of the unknowns, say 2, is equal to a.. Then there must take place the equalities utytz=0, Wt+y+2=0, B+y¥+2=0, and, consequently, (by virtue of the results of the last problem) u=y=2z=0. Thus, the given system has the following solutions gz=a, uy y=a, r=u Z=a, r=y = ll = 8 264 Solutions 31. Equivalence of these systems follows from the iden- tity (a? + bP + c? — 1)? + (a’? + bY + c'? — 1)? + + (a + pe ht (aa! + bb! + cc’)? + +2 (aa” + bb” + cc”)? + 2 (a’a” + bb" 4 c'c")? = = (@ + a + a — 1)? + (0? + b'? + 6 — 1)? + + (c? +c’? — 1)? + 2 (ab + a’d’ + a"b")? + +2 (ac + a'c’ +a")? + + 2 (be + be’ + bc"). It should be noted that nine coefficients: a, a’, a", b, b', b", c, c’ and c” can be (as it was established by Euler) expressed in terms of. three independent quantities p, q and r in the following way 2 gp? eat awit? = at p= 2etee | ex 2h Eee et —p2 2 2 a a2 yt ge of = Reta ap _pogyn a? = 2h Pe | pr — 20 petra) eat pod +r WW=1+P+¢+r°). 32. Multiplying the first three equalities, we get wy's? (y + 2) (e + 2) (z + y) = ab. Using the fourth equality, we have YY +2) @+ 2 (e+ y) = abe a(y + 2) + y? (2 + 2) + 2? (@ + y) + Qayz = abe. But adding the first three equalities, we find PYFTDEPC+)D+F E+ MPTP +e. Thus, finally or @ + +c + abe = 0. 33. Adding the three given equalities, we get Lo = W—2) 2-2) (z—9) a+b oon aa arapaicanns Solutions to Sec. 5 265: Similarly, we have ye ey pee eet od aye boc — Faw +2) ryZ if c—a—b a SMTA E+ 2) aye Hence (a+b+c) (b+ce—a) (a+c—b) (a+b—c)= == (2G) (ES) Gd) = a. Hence, we finally get the result of the elimination 2b%c? + 2b%a? + 2a%c? — at — b* — c4 + a®b*c? = 0. 34. We have pia ae eee 21,4 _ ztypa%s a typ=2b, mae = 2c. Squaring these equalities and adding them, we get 2 2 2 2 2 2 : o+ gtatatpet athe 4a® + 4b? + 4c2. On the other hand, multiplying these equalities, we find ye ®t ® o g yt tpt ata tpt ate Babe. Consequently, the result of eliminating z, y and z from the given system is a + b+ c? — 2abe = 1. 35. We have an identity (a+b+e)(b+e—a)(a+e—b)(a+b—c) = = 4b?c® — (b? + c? — a®)*, Replacing in the right member a?, b? and c? by their expres- sions in terms of z, y and z, and using the relationship ay + xz + yz = 0, we get Ab*c? — (b? + c? — a)? = 0. 266 Solutions Thus, the actual result of eliminating z, y and z from the given system is (atb+ob+e—a)(atce—b)(atb—c) =O. 36. We have (@@t+yP=2e+y + 3ry (ety) = =O +P thet yet yr —@+ yy And so oa (x + y)® = 3(@ + y) (@ + y*) — 2 (8 + 9’). r+y=a, ®+y=b, + ye Consequently, the result of the elimination is a = 3ab — 2c. 37. Put Mey eet: asacb fier tana Then a=ah, b=yh, ¢ == 2h, (+) On the other hand, we have (a + b +c)? = a? + Bb + c? + 2ab + 2ac + 2be. Since a+ b+c=1, a? +3? +c? = 1, we obtain from the last equality ab + ac + be = 0. Taking into consideration the equalities (+), we find ay + 22 + yz = 0. 38. We have (= E) (0G) (#-B)-1 Solutions to Sec. 5 267 39. From the first two equalities we find aan w(d—c) +2 (a—c) + y (b—0c) = 0. Multiplying the first equality by y, and the second by z, and adding them, we get (gy + we) (d — ce) = 2° (¢ — a) + y* (b—d) + + ay (at+te—b—d). We find in the same way that (ex + wy) d — ¢) = a (a—d) + y? (c — d) + + ay (b+e—a—d), aw (d — c)? = x? (a — d) (ce — a) + + yt (b— a (e— 6) + + ay [(a— @) (¢ — b) + (6-4) (¢ —a)]. Substituting the found expressions for zy + wx, 2c + wy and zw into the third equality, we get Ax? + 2Bry + Cy? = 0, where A =(c — a (a—d)? (6 —c)? + (€—d@) x x (b — d)? (ec — a)? + + (a —4@ (¢ — a) (d—¢) (a — bY, C = (b — @) (a — d)? (b — ce? + + (¢ — d) (b—d (¢— af + + (b —d) (¢ — b) (d—¢) (a — bY, 2B = (a+ ¢—b—d (a—d)?(b—c) + + (b+ ¢-a—d) (b—d? (¢— ay + + (d@— of (a—tP + a—@ (¢—d) + + (b — d) (ce — a)] (d — ¢) (a — B)*. Performing all the necessary transformations (the work can be simplified by making use of the result of Problem 8, 268 Solutions Sec. 2), we find A =(a—4*(c — ay (ec —d), B= (d—¢) (a—4) (b—¢) (a—¢) (d— b), C = (c — b)* (b — d? (¢ — d). Therefore we have Ax? + 2Bry + Cy? = (ec — @ [(a — d@) (a — ce) c — —(b—0c) (d— db) yP =0. Hence z fe y =e d—b) (@—d) (@—0)* Substituting these values into the equality (+), we get the required proportion. 40. 1° We have 2cos 2+8 cos 254 — (2008 $8 1) — 7 2 z or 4cos* ote — 4008 8 cos ath +1=0. Hence cos ate Since the radicand is equal to — 16 sin? ae and cos +6 2 2 ae “2 or equal to, zero. But this expression cannot exceed zero. Therefore we have is real, the expression — 16 sin? 6 must be greater than, sin arf =0. But sincee0O. 2° Analogous to 1°. Solutions to Sec. 5 269 41. By ieh as 0+@ 2cos > cos 2= 2 + =a, b. Hence tan ore = 2 . But f 2tan > , sinz= = 4+ tan? z Therefore we b i-z 22 a2 —p2 - a 2ab cos (0+ 9) = = aye sin(6+9)=—Gr-=ay 1a 4 ta 42. By hypothesis we have acosa-+bsina=c, acosB + bsinfB=c. Adding these equalities termwise we find 2acos a cos aot + 26 sin a c Hence a—B ¢ 2 a ath bein 28 c 2 cos 2B SE (ato tan =fB)* Subtracting now the given equalities termwise, we obtain —2asin 2¢? sin 25h + 2b sin 28 cos +6 =0. Since @ and 6 are different solutions of the equation, then sin a =£ 0. Consequently, the last equality yields tan 248 2 7’ 270 Solutions Let us return to computing cos? art « We have a—B cos? ce ape G4) 43. Rewrite the given equalities in the following way sin 6 (b cos a — acos Bp) = cos 0 (b sin a — asin B), sin 6 (d sin a — c sin ) = cos @ (c cos B — dcos a). Eliminating 0, we find (b cos @ — a cos B) (c cos B — dos a) = = (b sin a — asin f) (d sin a — c sin 8). Hence be cos @ cos B — ac cos*B — bd cos? a + ad cos a cos B = = bdsin?a — adsina sin B — be sina sin B + ac sin? B or (be + ad) cos a cos B + (be + ad) sina sin B = bd + ac. Finally cos («— fp) -as a 44. 1° We have et __Af2ecosBHe® _ Tpiecsape ef 2e24+20cosB __ e-+cos B “Beeb Recosa e-+cosa (by the property of proportions, from the equality = =4 ate =4) b+a~ ob)" Similarly, we have et _ 14 26¢cosB+e2 _ 126 cos afer et __—2—25ecosB _ _—A-fecosB “272ecosa ~~ T+ecosa * follows Solutions to Sec. 5 274 Then e-+-c0s B js ee __ 400s? B—1—e2cos?B _sin2B (tore (I-Fe 00s a)? ~ e2 feos? a—1—e cos? Sinz a * Consequently, __ Adecosp | _sinB 1--2ecosate? ~~ “I+ecsa — sina’ 2° From the given equality follows (see the result of 1°) e+eosB ___ 1-+ecosB e+cosa i1+ecosa~ Consequently, e+cosB—1—ecosB _ e+cosB+1-+ecosp e-+cosa+i+ecosa e+cosa—1—ecosa ( from the equality $= follows as = = ) fi Further (=e) (10s 8) (t+) (t-b008 8)_ (ie) (1 cosa) (fe) (1 cos a) or (1— cvs 8) (1—cos a) = i (1 +-c08 B) (1+ cos). Finally a B ite tan > tan ae egeng ie 45. Solving the given equation with respect to cos z, we find cos x (sin? B cos « — sin® & cos B) = = cos’ a sin® B — sin? « cos? f=cos? a — cos? Bp. But sin? 6 cos a — sin? a cos B = cos a (1 — cos? p) — — cos § (1 — cos? a) = cosa — cos B + + cos @ cos B (cos & — cos f) = = (cos a — cos f) (4 + cos a cos f) therefore cos a -+ cos B C8 F= FF cos a 008° 272 Solutions Further tant 2 10082 __1-4+605 4 cos B—cosa—cosB 2° t+cosz 1-+cosacosB+cosa+cosp _ (L~cosa)(1—cosB) _,, 9 a, 0B = "CF eos a) fF enspy — ban" ~y tant and consequently z ee % tan & y= tan > tansy. tan 46. We have sinta =4sin? £ sin? 2 = (1—cos@) (1 —cos 6) = 2 = (1-38) (1- cos & cos B cos p Hence eee ae cosB-+cosy , costa 1—cos?a = 1— cose cos B cos y az cos B cosy ’ i.e. 4 cos B +cos y 2 = SSE Oy a a (4+ G05 f COs p ) = cosa cos B cosy * Assuming that cosa is nonzero, we find cos y+cos B {+ cos ycosB * Now it is easy to check that cos a= tan} = tan? 7 tan® t : 47, Put tan 2 =a, tan Me B. Then the first two equa- lities take the form za? — Qya + 2a—x=0, xP? — 2yB + 22-2 =0. Consequently « and f are the roots of the quadratic equation az? — Qyz + 2a —xz = 0. Therefore 2 2a— a+p=, of=-=— Furthermore a—f=2i. | | | | | Solutions to Sec. 5 273 Let us now eliminate a and 6 from the last three equalities. We have identically (a + B)® = (@ — B)? + 4aB. Consequently, 2y \* 2 42a (4) 1844 After simplification we actually get y = 2az — (1 — P) a 48. From the first two equalities it is obvious that 6 and @ are the roots of the equation zcosa + ysina@ — 2a = 0 (unknown a). It is clear that 0 and @ are also the roots of the equation (2a — x cos a)® = y* sin? a. Transform the last equation in the following way x? cos? a — tax cos & + 4a? = y? (1 — cos* a), (x? + y?) cos? a — 4axr cos a + 4a? — y? = 0. Therefore the quantities cos @ and cos @ are the roots of the following equation (2? + y*) 22 — 4arz + 4a? — y? = 0, and therefore i had yt ___ Aaz cosdcosq “FF, C080 -+c0s p= We then have 4sin? 2 sin? Angst tcos@ 4 or 1— (cos 0+ cos ~) + cos 0 cos p= 1. Hence, y? = 4a (a—z). 49. We have tan? > tant & 2 2 0 a” —tan? = tan? = 4—tan’ z tan’ 3 18-1225 274 Solutions z- But 8 1—cos® __1—cos a cos B tan? = = = — 2 —TFeos0 ~ tpeosacosp ’ an? & tan? Ty eos * Consequently 1—cosacosB — 1—cosa O+a 03 & cos B t+cosa Van _ acosp 1—caa — 1-+ cos a cosp ' 1+-cosa __1-cosB 4 9 B = TH cos 6 = tan? = 50. We have ate ___cosz-cos(z-+20) ___cos(z-+0)cos0 __b bd ~~ cos(a+0)+c05(z +30) cos(z+20)cos0 ce * Hence afte b4+d 6 «° 51. We have 29 cosB an?) cosh 1+ tan?6= cosa 1+tar?g= cosa: Hence tan20 __ cos B—cos a cos tan2@ cosa cos p— On the other hand, it is given that tan?@ tan? tant@ tan? * “Therefore we have cosB—cosa cosy ___tan?a cosB—cosy cosa taney * From this equality we get cos? asin? y—cos? ysin?a _ sin? y—sin?.a cos B= = = - 7 B= cosa sind y—sin@a cosy ~~ eos asin? y—sin® a COS p Solutions to Sec. 5 275 But tan? — 1—c0sB _ cosa sin y—sin® « cos y—sin® y-+sin?a _ 2 ~ Tos B ~ cos a sin? y— sin? « cosy} sin? y—sin? @ ___Sin? a (1—cos )—sin? y(t—cos a) sin? y (1 cos a) — sin? @ (1--c08 y) __ 8sint & cost $ sin? 1 —8sin® 3 cost $ sine S Bsin? 2 cost 1 cos? F-—8sin® F- cos? cos? J sin? 5 sin? 3 (cos? $-—cos? 7) tan? & +? =a fant ay > OM Eta as cos? z cos? > n? > —sin? since cos? 5 — cos’ t= sin? }—sin?}. 52. Put tanker tant 2 . 2 Then cos 0 = 4-27 = cos a.c08 8, cos p= oe = = COS Cos f. Further gta tnensacosB yn te0say cos T+ cosa cosp ’ T-f cos ary cos B ” therefore tan? 8 aye = (1—cos @ cos B) (1 —cos a cos B) 2 (1--c0s a cos B) (1-4 C05 xy cos B) Add unity to both members of the equality. We find 2 2 (1 cos a cos cx cos? B) T+cosp — (1-}cos a cos p) (1 -+ cos a cos f) Assuming cos B ¥ 0, we obtain cos @ + cos a = 1 + cos a cos a cos? B, 276 Solutions ie. cos @ + cosa, = 1 + cos a cos a (1 — sin? B), cos @ cos a sin? B = 1 + cos @ cos a — cos @ — cos & = = (1 — cos a) (1 — cos a), and, consequently, indeed sin B= (sez!) (saa 1): 53. We have cos (B—y)—cos (a—f) _ _cos(y—a)—cos(B—y)_ _ ‘eos (a + B) — cos (B+) cos (B+ y) —cos (y+ a) cos (a. — B) — cos (y~a) = "cos (ya) —cos (ap) Hence sin (44 = 2 2 mo(Efiva) ve(EEan) (Ess) or tan B—tan 2% tan y—tan PF tan a —tan Bry tan B+ tan at tan y+tan bre tana+tan ary But from the equalities follows ‘an a 7 6 tany ton (B+y) tn taty) tan $ (a+) 54. From the first equality we have (tan 0 cos B—sinB) cosa (tan @ cos @—Sin a) 60s B (cos a—tan@ sina) sinB _ “(cos B+ tan p sin) sina Solutions to Sec. 5 277 Hence sin a cos B cos (a — f) tan 0 4- sin B cos @ cos (« +B) tan p= =2sinBcosBsina cosa. (+) From the second equality we get tan0 _ ___ cos (a—B) tan B tangy ~ cos(aTf) tana" Therefore we may put tan 0 = 4 cos (a — f) tan B, tan @ == —A cos (a + B) tan a. Substituting the expressions for tan @ and tan @ into the equality (*), we find 4 = Tone sinp © Thus 1 tan = SEB = 3 ota + tan Bh _ __ _cos(a+B) _ 14 tan @ =—Toosasinf 2 (tan a — cot B). 55. We have sin? @ + sin? B — 2 sin @ sin B cos (a — 8) = = sin? a + sin? B — 2 sina sin B cos a cos B ~ — 2 sin? sin? B = sin?a — sin? @ sin? B + + sin? B — sin? a sin? B ~— 2 sina sin B cosacos f = = sin? a cos? B + sin? B cos? a — — 2sin asin B cosa cos fp = = (sin « cos B — cos asin B)? = sin? (a — 6). Therefore sin (2 — ) =-:n sin (a + 8), sin a cos B — cos a@ sin B= + n (sin a cos B + cos @ sin B), tan a — tan B =-tn (tan a + tan B). Finally dan tana = 7=; tan B. 278 Solutions 56. Expanding the given equalities, we get cos @ cos 30 + sin a sin 30 = m cos? 6, sin @ cos 30 — cos @ sin 36 = m sin? 0. Multiplying the first equality by cos 30, the second by — sin 30 and adding them term by term, we find cos @ = m {cos? 0 cos 30 — sin’ 6 sin 30}. But it is known that cos 30 = 4 cos? 0 — 3cos 0, sin 30 = 3 sin 6 — 4 sin’ 0. Consequently cos’ 6 cos 36 — sin? 0 sin 30 = 4 (cos® 6 + sin® 0) — — 3 (sin 6 + cos‘ 6). But squaring the original equality and adding, we get cos’ 8 + sin? @ = a. Compute cost 6.+ sint 6. We have cos® 8 + sin? @ = = (cos? 6 -+ sin? 0) (cost @ + sin* 0 — cos? 0 sin? 6) = = cost @ + sin* 6 — cos? @ sin? 6. Therefore Ay = (cost 0 + sin® 0)* — 3 sin® 0 cos? 0, ' 3 sin? 0 cos? 6 = 1——, sin* 0 + cost 0 = 1 — 2 sin? 6 cos? 0 = =1 31-3) =$(1 43). Thus cos a = m {4 (cos® 0 + sin® 6) — 3 (sin* © + cos* 0)} = 44 _ 2) 2=m =m {a Seti J eeeece ie cen ie. m? + mcos@ = 2. 57. From the first equality we obtain a [sin (0 + g) — sin (8 — g)] = = b [sin (8 — g) + sin (0 + @)]. Solutions to Sec. 5 279 Hence atan » = b tan 0. Consequently aun? ene Fe But from the second equality we have pian Lhe 0 2 tan > =———_ @ therefore 2tan-e _ 2(vtan tS +e) by). 3? LT (pf) \7° —tan? 2 & 1—tant .[ (etan $ ‘yy 1—i_ a Putting for brevity tan $ = and transforming the last equality, we find be (A + 2c?) = — (B® + ® — a) ze. But ar Tr =sing. Finally Qbe sin @ == Gop * 58. From the third equality we obtain sin? @ sin? @ = (cos 6 cos p — sin f sin y)*. Using the first two equalities, we find (1 sin? B (1— int) wy —sinBsin vy). ~ sin? a, a After some transformations this equality yields tan? @ = tan? y + tan? B. 59. We have asin? @ + bcos? 8 = 1, acos’@ + bsin?@ = 1. 280 Solutions Hence atan?6+b=1+tan?6, btan?g+a=1-+ tan’Q. Consequently (a—1)tanP®?6=1—b, (b— 1)tan?g = 1—a, tan? 6 4 tantg =): On the other hand, tan?0 _b2 tantg a2” From the last two equalities we get (assuming that a is not equal to b) a+ b—2ab=0. 60. Rewrite the first two equalities in the following way cos@ cosa + sin 0 sina =a, sin @ cos B—cos 0 sin B=b. Multiplying first the former by sin f and the latter by cos a, and then the former by cos B and the latter by —sin « and adding them, we find sin 6 cos (« — f) = asin B + bcosa, cos @ cos (2 — B) = acos B — bsina.- Squaring the last two equalities and adding them, we get cos? (a — B) = a? — 2ab sin (a — fp) + b% 61. Since cos 3x = cos* x — 3 sin? x cos 2, sin 3x = —sin® x + 3 sin x cos’ z, the equation takes the form (cos? z — 3 sin? x cos z) cos® x + + (—sin’ x + 3 sin x cos? z) sin? zx = 0 cos® x — 3 cos‘ x sin? x + 3 sin‘ x cos? z — sin? z = or (cos? — sin? z)? = 0, cos 2z = 0. 62. Since sin 2x + 1 = (sin z + cos z)*, we have (sin z + cos x)? + (sin t + cos z) + cos? x — sin? z = 0. Solutions to Sec. 5 284 Hence (sin z + cos z) (4 + 2cosz) =0 or cos x (1 + tan z) (1 + 2cosz) = 0. And so tan z = —1 and cost = — 4 are the required solutions of our equation. 63, We have sinkz _ 1—cosz _ coz I—sinz Hence (cos? z—sin’ z)— (cos? z—sin? 2) _¢ aaa cos? x (1—sin z) or (4 — tan z) (1 — cos z) = 0. Hence tanz = 1 and cosz = 1. 64. We have cos 3a = 4 cos? a — 3 cos a. Therefore cos 62 = 4 cos* 2z — 3 cos 2z. On the other hand, 14008 22 \3 cos’ t= (-= ) . The equation takes the following form 4 (1 + cos 2x)’ — (4 cos 2x — 3 cos 2x) = 1 or 4 cos? 2x + 5 cos 2x +1 = 0. Thus cos 2x = —1, cos 2x = a 65. We have sin 2z cos x 4- cos 2x sin x + sin 2x — msinz = 0. Hence sin x [2 cos?.x + cos 2x + 2cosx — m] = 0, sin x [4 cos? x + 2 cos x — (m + 1)] = 0, 282 Solutions And so, one solution is sinz =Q, The other is obtained by the formula =1 + Vint SS. cos x= Hence, first of all, it follows that there must be 4m +5 >0. Further, for one of the roots to exist it is required that |—1+ V4m+5|<4, ie. that -4< —-14+V4im F< <+4o0r—-3 5 it has no real values. 66. Rewrite the equation as ay M(t +4) €08 208 (22 — a) — cos = —(1+k cos 2x) cos (x—x)}=0. But cos x cOs (20-0) = 4 cos (3e—a) + yeos (x—a), cos 2x cos (x —a) =F 00s (32 —a) + $ cos (eta). Solutions to Sec. 5 283 Therefore =e {(1 +k) [cos (8a —a) + cos («& —a)] — —2cos (t— a) —k [cos (3a — a) + cos (x-+-a)]}=0 or 4 cos (=) {cos (32 —a) — cos (x —a) +- +k [cos (t—a) —cos (x -+ a) |} =0, Gas {ksina—sin (2r—a)} +0. Hence sinz = 0 and sin (22 — a) = ksin a, 67. Since sin? z + cos? x = 1, we have sin’ z + cost z + + 2 sin? x cos*?z = tandsintz+-costz = 1 — Aisin 22). The equation takes the following form sin? 2x —8sin2e +4=0. Hence sin 22 =4+ V16—4, sin 2x =4 + 2V3. Rejecting one of the solutions, we get finally sin 2z = 4 — 273. 68. We have log: ee log. a=—t_, logua= 1 7 Toggz ? 08029 Tovar? (O8utx = Toate The equation takes the form 2 1 3 is Toga = + log, 2-1 a loga +2 fay Put logg z = z. Finally, we have to solve the following equation 2 1 3 stay taro Hence O24 11244 FETE 284 Solutions The required roots are 4 eget y=—z B= —Z- Thus bake oe m=a 3, m=a 2, 69. We have rays, Hence ha? yvayey, Consequently, either y = 1 ora +y = ~. But aty =1 a*¢ = 1 and, consequently, c = 1. Thus, we get one solu- tion z=1, y=. Let us now find a second solution. We have (x + y)? = 4a’, ie. x+y =2a. Therefore 2a yo ze ya way, (+)'=1, and consequently Pay, ie. a = 2a — 2. From this quadratic equation we find 1 1 z=—zyt Vt 42a. The positive solution is 4 eae t=——st+ V 1420. The corresponding value of y is found by the formula yar’. Solutions to Sec. 6 285 70. Raising the first equation to the power q and the second to p, we obtain uPiy? = a*t, Papp? = a¥?, Dividing one of these equalities by the other termwise, we find pe-P? ag, and consequently pu-g v=aP-?, Analogously, we find =p—v9 u=aP-@, (*) Substituting these expressions for u and v into the third and fourth equations, we have aP(=-+u2)— 2x09 — ppt, q2xyp—ax2-+y2) — cP?—92, Hence P (x? + y?) — 2xyq = (p® — q") loge b, 2ayp — q (x? + y*) = (p? — ¢) loge c. Consequently z+ y? =plog,b+qlog,c, 2cy =q log, b + p log, ¢; wherefrom we find x and y, and then wu and v using the for- mulas (+). SOLUTIONS TO SECTION 6 1. Let rc =a+ Bi, y = y + 6i. Then rty=atyt (B+ 4)i, z—-y=a—yt (B— di, lz+yP + le—yP = (a + yy? + (B+ 6)? + +(a— yi + (B68)! = = 2 (a? + B) + 2 (y? + 6) = 2 {laP +1yP}. 286 Solutions 2. Let x =a-+ Bi, hence x = a — Bi. 1° By hypothesis, a — pi = a — Bp? + 2ai. a=a— Bp, —f = 2ap. Ba +1)=0, a=a?— pr Assume first B = 0, @ = @ or a (a —1) =0. And so, first of all we have the following solutions a=0; B=0, <=0; a=1, p=0, c=1. Let us now pass over to the case when 2a + 1 = 0, i.e. a=—5, —$=5-P Pag, pas, Hence Therefore fe ae ee a four complex values of z sa- tisfying the condition c= 2 1g 44:4, en —4-1 3. namely z=0, c= 2° Let us solve the a system a (a? — 3p? — 1) = 0, B (3a — 2 + 1) =0. We find the following solutions a=0, B = 0; a =0, B= +1; a= +1, pB=0. And so a=0, <=+1, t= +i. 3. Put a + bi = 2, ap + bot =Y, ~~ +) Ana + Opi = u, a, + bi = w. Solutions to Sec. 6 287 Then the inequality to be proved may be rewritten as latyt...tutwl[< asin, 2 lel-+lul+--.+101=3) om n 2,2 ie+yt +s wP= (Si pa cos ga) +(3 pa sings). It is required to prove that A= (Son)? (3 pros gr)” -(3 | basin gs)” 20. we have aug (D1 px) = > ph +2 D par, k=1 kat s¢t a 7 (% pn cos px) =D pk cos* gn +2 >) pape Cos ps cos gr, k=1 kai a¢t : ete (2 pa sin on) = & pi sin’ ge +2 Dy PsPr Sin gs sin @r, consequently A=2 ¥ papi —2 Dd) psp C08 (Ps — 1), spt set A=2 5} paps {1 —c0s (9. —G)} =4 Dy paps sin? SSS. apt spt 4. Proved by a direct check, taking into consideration that ee? = —e—1, ®& = 14. Solutions to Sec. 6 289 5. It is obvious that a + UD? + ce — ab — ac — be = =(a + eb + ec) (a -+ &%b + ec), ey + 2 — cy — 22 — ys = = (x + ey + e%2) (x -+ ey + £2). Therefore (a? + b? + c? — ab — ae — be) (2? + 2 4+ FF — — zy — xz — yz) = [(ax + cy + bz) + + (cx + by + az) e + (bt + ay + cz) &*] x x [(ax + cy + b2) + (cx + by + az) & + + (bx + ay + cz) e] = = X?+ Y?+ 27? — XY — XZ— YZ, where X=ar+cy+bz, Y =cxr+ by + az, Z= br +ay+ez. 6. 1° Solving the given system with respect to z, y and z, we get _ Af Be+ Ce? A4LB4C pee eee 2° We have |AP + IBP + |CP = 44 + BB + CC. But AA =(@+y +9 @+y +9 =_ =lcP + lyP +2? toy +2+ i +¥@+)+7e+y), BB = (x + ye + ze) (& + ye? + ze) = =l2P+ ly P+ 12h +2 ye + ze) + + y (ae? + ze) + 2 (xe + ye), CC = (x + ye? + ze) (z@ + ye + ze’) = = | + [yl + [zl + 2 (ye? + ze) + + y (we + ze%) + 2 (ze® + ye). 19-1225 230 Solutions Adding the three equalities term by term, we find |AP-+ [BP +|CP = AA + BB+ CC = =3ljzP+iyP+2A+ et + e+ e%) + +2Qt+etel+yle(t+ e+e) + +2+e+ el tele +eteX+ tyd@tetel. But since 1+ e+ e? =0, the last three expressions in square brackets are equal to zero and [AP + [BP + CP = 3 ljzP + |yP + [zP) 7. On the basis of the result obtained in 1° of Problem 6, we have » _ AA'4.BBI+CC' » _ AA’ 4 BB'e2 4 CC'e fas, ofr ve AAT BOs + CC's? 4 7 Further AA’ + BB+ CC = (ety +a ty +2)+ + (© + ye + ze’) (x’ + y'e + 2’e*) + + (@ + ye? + ze) (x! + y’e? + 2/2) = = 3 (ex! + ay + y’). And so 2" = xa’ + zy’ + yz’. Analogously y” = yy’ + + a2’ + 22’, 2” = 22’ + yx’ + zy’ (the last two expres- sions emerge from the first one as a result of a circular per- mutation). 8. Though this formula was already proved (see Prob- lem 2, Sec. 1), we are going to demonstrate here another proof, using this time complex numbers. We have the identity (a5 — By) (a'6" — Bry’) = (aa + By’) (yB’ + 88°) — — (aB" + BS’) (ya" + by’), let us put here a@=a+yi, P=24+li, y= —@—t), 6=a4— yi. a’ =a+ bi, B =c+4di, ¥ = —(c di), & =a— bi. Solutions to Sec. 6 294 Then ad — py by? + ote, avi — Py H=?4+ P+) +a, aa’ + By’ = (ax — by — cz — dt) + + i(bx + ay + dz — cl), vB + 88) = FY + a = a Fy). Therefore (aa’ + By’) (yB’ + 88’) = (ax — by — ez — dt)? + + (bz + ay + dz — ct). Further ap’ + BS" = (cx — dy + az + bi) + + i (dx + cy — bz + al), ya’ + by) = —(ca — dy + az + bt) + +} i (dx + cy — bz + at), ie. —(ap’ + BS’) (ya’ + Sy’) = (cx — dy + az + bt)? + + (dx + cy — bz + at)*. Substituting the obtained expressions into the original identity, we find @+ P+ i+A@+y+2+A= = (ax — by — cz — di)? + (bx + ay + dz — ct)? + + (ce — dy + az + bi)? + (dx + cy — bz + al). Replacing in it d by —d and t by —t#, we get the required identity. 9. Expand the expression (cos g + isin @)", by the binomial formula. We have (cos @ + isin g)" = cos" @ + n cos"! pi sing + n(n—1) cos" @ (i sin @)? + cos"? p Xx n(n—1) (n—2) + 72:3 X (isin g)’ + ...+-ncos p (i sin gp)" + (i sing)”. 292 Solutions Separating the real part from the imaginary one in this expansion, and using de Moivre’s formula, we find cos np-+i sin np = (cos" p— 2S" cost sintp + ale )+ +i (ncos"* @ sin g— EFI?) cosr-2p sintp+... } : Hence cos np = cos" p— ae) cos" @ sin? @ +. ' —1)(a— E sin np =ncos"@ sing — 227 a=?) cos’* @ singg+.... Taking into account the parity of m and dividing both members of these equalities by cos" @, we get the required formulas. 10. First prove case 1°. We have cos = eset isin 2) (eos o—isin g) Putcosg+ising=s. Thencosp—ising =e+ 2m costmgm (SEE) at. St hye tet * R=0 Further mot am 2R™ cos p= Dy Che +. CIm+ Come. n=0 m1 In the second sum put m—k=—(m—k’). Then this sum is rewritten in the following manner. 0 m1 2m—k’ ,—~2(m—k’} hk —2(m—k) Dy BME = Che, him-1 And so 2" cost p -"3 CEm (27 4 22) 4 CB However, e2(m—h) + g-2(m—k) — 2 cos2(m—k), Solutions to Sec. 6 293 Therefore, m-1 2?"cos*" p= 3} 2Cimcos2(m—k) p+ Con. k=0 Replacing in this formula 9 by->-9 we get formula 2°. Formulas 3° and 4° are deduced as 41° and 2°. 41. Form the expression Un + iv, = (cosa + isina) + + rlicos(a@ + 6) + isin(a + OJ + ...4 + r® [cos (a + n@) + isin (a + n6)] = = (cosa + isin a) {1 +r (cos @ + isin 6) + ...+ + r™ (cos nO + isin n6)}. Put cos @ + isin® =e. Then u, + iv, = (cosa + isina) {1 +re+ ...+ (re)"} = (cos a@ + isin a) ent (rey? —14 Let us transform the fraction » separating the re—1 real part from the imaginary one. We have : (rertt—1 __ [(reyr*— 1] fre—1] re—1 (re—1)(re—1) __r™2 [cos nO-+i sin nO|—r [cos 0—isin 0] ia T= cos Op re ac “Lr [eos (1) @-Fisin (n+4) O)-+4 t T—2r cos 0+ r2 a Multiplying the last fraction by cos @ + isin @ and sepa- rating the real and imaginary parts, we get the required result r+? [cos (n0-+ a) +i sin (n0-+a)] t= 2reesOfr2 —r[cos (a--6)-isin(2—O)] , te T=-dreos0y 72 rt {08 [(n +4) 0+ a)+i sin [(n-+1) 0 +a]} +008 a-fisin a + 1—2r cos Or? : Un+ivna= + 294 Solutions Hence __ 608 & —r cos (4 —0)—r"*1 cos [(n-+1) 8-a] + r"*? cos (n0-+ a) ee 1—2r cos0--r2 : __ sin a—r sin (@—0)—r™1 sin [(n+-4) 0-+.a]-+r™? sin (nO +a) n T—2r cos 0+ r2 Putting in these formulas a=0, r=41, we find sin att Bcos +0 1+ .cos0+ cos 20+... ++ cos n§ = ———_,——, sin > 2 sin 2+ 0 sin m sin 0+ sin 20+ ...+sinn0= sing 12. We have S+4S'i =>) Ch (cos 0 +i sin 0) = 5} Ch (cos 0+: sin 0) = h=0 k=O =(1+cos 0+ isin 0)"= [2 cost 24 24 sin $c0s$ ]"= =2" cos" F (cos Be isin $)"= = 2" cos" 2 (cos 4G sin +) : Hence 8 a) 0. nO = 2% cos" © eos 2% 9 e958 sin 2, S = 2" cos’ 7 OOS, S’ = 2" cos’ 7 sin-3 13. Put n S =sin??a+sin?? 2a+...+sin?? na = >} sin? la. 4 But (see Problem 40) p-1 4 4 = gpa (— 1)? D)(—1)"C3, 008 2(p—k) la +e Cho, k=0 therefore i (

y, it is sufficient to prove that (Va—Vo)? 1 (a—b2 2 poet Consequently, it is necessary to prove the following (Va+Viy veri St iver vor 1 1 =3( (14/3) in since => ae The second inequality is proved in a similar way. 24. Put a =2°, b =y®, c= 2. The only thing to be proved is that We have B+ y+ 2 — 3ryz >0 for any non-negative z, y and z. But we have (see Problem 20, Sec. 1) a+ y+ 2 — 3ryz =(@ +y+2) x x (e? + y® + 2% — zy — zz — yz). And so, it only remains to prove that e+ yt 2 — cy — a — yz >0. 406 Solutions But we have (see Problem 10, Sec. 5) Qa® + 2y? + 22? — Qey — 2e2 — 2yz = =(¢—yP + @—y+(y—y>0. 25. We have ay +49 VamVag. a, And so, indeed, (1 + ay) (1 + a2)... 1 + an) S 2 27. 1° Make use of the following identity (a+b) (a+c)(b+c) = +an= 1. = (ab + ac + be) (a+ b+ c) —abe. But etbte > y/ ae, abt ach be >y/ ae". Therefore (a +b +c) (ab + ac + be) > Yabe, and consequently (a + b) (a +c) (b +0) > Babe. 2° We have a b c¢ __ atbte bate Bpeateere (arb useericemasre any etatb 4 4 1 : +SEE 1 (04040 (pee tah tas) 3: But (+e) +(at+e)+(c+)>3/ O40 a+ (a+), apote>s/ O+q ara (ate). Solutions to Sec. 8- 407 Further 4 ee 4 ao ope tape + ape Pro ero ary OT) (ate) + +(O+0) (+B) + (+d) e+e} > Soran” OFF OTH ATH Therefore b weet ate tate eV CF HEF HTH x x apaeaern/ OF c}* (a+e)* (a+b — Thus a b be at ate oe a zt: 28. It is sufficient to prove that (a+) (b+) (c+ m)>(V ade + /kim)*. We have (@+k) (b+) (¢+m)= = abe + klm + (ale + kbe +- abm) + (kle + alm + kbm), (/ abe + j/ km)? = abe + klm +. +3 kim +3 / FP mabe. But i a SY avCkim, Hecpelm-t Hom >j/ PP mPabe. Hence follows the validity of our inequality. 29. We have 3 — > +See = —— But rttte : oe Vabe U VY abe< itt: . ie. 4 3 re 408 Solutions Therefore 4 j4het tag? a ° TF: 30. It is necessary to prove that the arithmetic mean of n positive numbers is not less (>) than the geometric mean of these numbers. We are going through several proofs of this proposition. Let us begin with the most elegant one which belongs to Cauchy. Thus, we have to prove that nteete tins Fay En At n =1 the validity of this inequality is obvious. At n=2 and n =3 the proposition was proved in Problems 23 and 24. Let us first show how to prove the validity of our assertion at n = 4. We have zi+z2 zitte , tote str tteetegtmy 2 > 2 ae sate z > 7 : But ate sya, 28>), Therefore abet at WV aim V tary = WV tsi. Let us now prove that, in general, if the theorem holds at n=m, then it is valid at n=2m too. Indeed, arte bast e-beam + 2am 2m atre , tat Zam-1 + tem Sec tam TATE ee ee Ee ™m > V ate Eamtb eam nt 2 2 Hee 2 (since we assume that the theorem is valid at n=m). Solutions to Sec. 8 409 Further tteebaat ttm am ee SV Vite V at, V tamaitom = 29/ And so, assuming that the theorem is valid at n =m, we have proved that it is true at m = 2m as well. And since we proved the validity of the theorem for n = 2, it is valid for n = 4, 8, 16,..., i.e. for n equal to any power of two. However, we have to prove that the theorem is true for any whole n. Let us take some value of n. If n is a power of two, then for such a value of n the theorem is valid, if not, then it is always possible to add a certain q to n such that n + q will yield some power of two. Put eT gl, ... Lom. n+q=2". We then have sebeetrat tan tenit eo tenes SSeS Oo gE SW tits vee UnEnyy +++ Engg for any positive x; (i=1,2, ...,+4). Put paeneey ean nq = abet. ten We get abate pen pet tee g wa = neq f{——— = V wees, (EE Eee), Hence + n subarhaten 5 "/ z 7 Uyly Lp a, \' or (zctexbe- ten) a2, a ay (Bobet en) n (ecbeebben) Say oan 410 Solutions and finally nteat ee Fin o/s on rae And so, the theorem is valid for any whole x. It is obvious that if 2, =z, =... =2,, then the sign of equality takes place in our theorem. Let us prove that the sign of equality occurs only when all the quantities 2,, %2,..., Zp are equal to one another. Suppose at least two of them, say x, and 22, are not equal to each other. Let us prove that in this case only the sign of inequality is possible, i.e. it will be mba tiny amie Indeed ate, t+z0 tteetee fen 2 ee = n n >V (Ea te But if x; is not equal to z,, then b> am, consequently In > VW teats and therefore abe SY aime if at least two of the quantities 2,, z2,...,Z, are not equal to one another. Given befow are some more proofs of this theorem. Let us pass over to the second one. Let n be a positive number greater than or equal to unity (n >1). We assume here that a and & are two real positive numbers. Then the follow- ing inequality takes place (a™4—b") (a—b) 20, Solutions to Sec. 8 4at Hence a" +b" >a™'1b 4 bra, Consider 7 positive numbers a, b,c, . . ., k, 1. Let us apply this inequality to all possible pairs of numbers made up of the given n numbers. Adding the inequalities thus obtained, we find (a +b") 4 (a Fe") +. (aH) + LOPE EOE. LEMS >(a"b + b"a) + (a™“Fe + c?4a) -f (aD Va) + (RU PR). Hence we have (@—N@+OL. YS Sab +o™t +... PY fo He IF EP YT fe(att4+o4 4) + FU HOE ERO). (s) Using this inequality, it is possible to prove our theorem on the relation between the arithmetic and geometric means of nm numbers by the method of induction. We have to prove that Seb sen bn amy aurea Put aya", =O", t=O, ey na =k, tal". Then it is sufficient to prove that epi PES ab td Let us assume that this inequality is valid at the exponent equal to n — 1, ie. PIE ERIE DS (n—1)b-k.. PALME EM DS(n—A)arc. PEE ERIS (n—A)a-d.. 442 Solutions Using the inequality (+), we find (nA) (QF O"4 ER ETS Da(n--1)bk...14b(n—Nac... IH... 4 +1(n—‘) ab... k. Hence (n—1) (a" +0" 4... 4k" +1") >(n—1)-n-abe ... kl, i.e. anon tT abe... hd. Thus, our theorem is proved for the second time. Let us pass over to the third proof of this theorem. It will be carried out using the method of mathematical induction once again. Let there be n positive numbers a, b, . k, l. It is required to prove that a+b} ...$k4+]>ni/ab... kl. Assuming that the theorem holds true for n — 1 numbers, we have atb+...+k+1>(n—1)"/ab.. FLL And so, the theorem will be proved if we prove the inequality (n—1)” ab. k++ ln ab... kel. Thus, we have to prove the inequality 1 ab. ke "fab... nat)” VF ton VS. ab... kl Tt Put = Eno Therefore, it is required to prove that (n—A) B+ 12>nk™, And so, to prove our theorem means to prove the inequa- lity ng" (E—1) >8"—4, Solutions to Sec. 8 413 where § is any real positive number and n is a positive integer. Let us prove this inequality. At § = 1 we obviously have the equality. Suppose now § > 1. It is required to prove that 4 ne This result is obtained as in the previous case, and, thus, the theorem is proved. All the considered proofs were carried out using the method of mathematical induction. Therefore, it is desi- rable to get such a proof which would establish immediately that abet ten Ga, dn if a, @2,...,@, are any positive quantities not equal to one another simultaneously. Put a; = 2z?. Then we have to prove that atta t. tan n 242... 2n >0, i.e. the problem is reduced to finding out that a certain function (form) of n variables x, z2,..., 2, is positive. As is known, n letters 2, 22,...,2, can be permutated At4 Solutions by n! methods. If f (1, 22,...,2,) is a function of n variables xj, 22, . . ., 2,, then the symbol >}f(z;, 22, . . ., Zn) will denote the sum of n! quantities obtained from f (t1, Za, +. +, 2p), using, all possible permutations. For example, Dd x22 vee En =! Lrg... Lny Dapa—~Al (pads bay). Introduce the notation n State tet yey see Tn =P (Ly Lyy +005 Tn)o It is easily seen, that whatever permutation is used, the function @ (x, 22, -.-,%n) remains unchanged. Therefore we have nlp (4) B25 ++ +s Zn) = HAD batt ba Saute sete But Dattapt ...papanl (p+ apt ... +27). On the other hand, taht oo. pahe s! rth therefore E = Dat — Daye... ta. (*) T1@ (4s Lay + +r Let us consider the following functions =D (ey — a) (G1), 2 = Di (20 — 23) (1 — 22) 25, G3 = Di (ar — a3) (41 — 2) ar, conn = 3 (4 — 22) (04-4) Hoty 6 te Solutions to Sec. & 445, We have O1=2 Dap—2 Dara, G.=2 atx, —2 Datars, (3 = 2D) 22-2232 >) x? *zyr9x,, at D tires... ty-— 2 Dp ayrety ... Ln, Adding these expressions termwise, we find D+ Gott + On1=2 Yat—2 Dd) tyr, ... a Comparing this with the equality (+), we get nl @ (zy tay ees ta) =F (G+ 2+ H+ tena) And so Poet... an 4 ee ee ee But it is evident that @,, G2, .- +, @n-1 vanish if and only ifz, =2, =. = Zp. If not ‘all of the variables are simultaneously equal to one another, then all ; > 0. Indeed, we have =X (aim) (r+... +a) S0, G2= Di (x1 — 22)? (zPP + «.. +23") 230, Prt = Di (e122)? 250% +++ On. Therefore n Sitsbd tat iy we. tn 0, the equality being possible only if a =a, =... =2,. And so, the theorem is proved. This proof belongs to A. Gur- witz. 34. We have (using the preceding problem) °/Gidy Oy eet tn ten py ten To prove the second inequality consider the product (a2. » + Gn)? = (Aran) (424n-1) « » « (Gnas). 416 Solutions But we can prove that GyGn-n+1 = 24a, (see Problem 19, Sec. 7) Therefore (ajay. « + Gy)? > (aya,)” v Ady 32. Consider a quantities equal to = b quantities equal to and c quantities equal to = The arithmetic mean of these quantities will be and a>V aan. 4 1 4 Sratenbarerer a cig a+b+e “atbte * The geometric mean is equal to atb+e T 1 T ae" bb oo * Consequently 3 yates a-bye= aa Bb ee? ie. a b c 4 qa Goro Gare SZ (At b+). 33. Put Ger peepee tan ™ m ™ where a, B, y and m are positive integers. Consider the product oe ee @ oe b—c \@ -V0rSVl Since a, B and y are whole positive integers, the radicand may be considered as a product of @ factors equal to 44 poe each, f factors equal to 1+ = a—b each, and » Solutions to See. & 447 factors equal to 140% each. Then we have oy (Se) (Sy - a (14754) 48 (14-55%) +9 (14 ay / ( Raising both members of this inequality to the power a+b-tc, we get the required result. 34. We have fa fect en EE segeee Vs 68)... 6)” Weoaye—b).. 6-1 But VY 696-5) C-D< ebro te—+. +6—)) ni, S n noieainten Therefore 4 n V (s—a) (s—b) ... (6—)) AUDs The further proof is obvious. 35. First of all this inequality can be obtained from Lag- range’s identity (see Problem 5, Sec. 4). But we shall pro- ceed in a somewhat different way. Let us set up the following expression (Aa, + pbs)? + (Ade + po)? +... + (Ada + Won)? = = AM + 2Bap + Cy, where A=ai+az+...+a4, C=vi+b}+...4+05, B=ayby+agb,+ ..-+anbn. Since the left member of this inequality represents the sum of squares, we have AM + 2BaAy + Cy? > 0. for all values of A and p. 27-1225 A18 Solutions Consequently, the trinomial Az* + 2Br +C is greater than or equal to zero for all real values of x. The- refore, the roots of this trinomial are either real and equal or imaginary, and its discriminant is less than or equal to zero, i.e. BP—AC<0. Thus (ayby + agby + 6 $ dnby)?— (Gi abt 5. +45) (bi +... +05) <0, wherefrom also follows that the equality sign is possible only if free Ga tear By 36. Put b, =bp =... =b, =1 in the inequality of the preceding problem. We then have (aaa... + an)’0. But Gd =9—%; (ty +a. t... + tia + tit. .. +e,) = =4—% (p — 2). Therefore (p— 2)? 222 (q— pai +250. Consequently ni —2pz; +2 (n—1)q—(n—2) p?<0. Consider the trinomial of the second degree na* — 2px + 2(n — 1) q — (n — 2) p® and denote its roots by a and B 2 420 Solutions Solving the quadratic equation, we find 2 an P74 p=24+7 VY pha, (>. We then have an identity nat — 2px, + 2(n — 1) q—(n— 2) p? = =n (x, — a) (x; — B) <0, wherefrom follows that x; lies between @ and 8, i.e. a 0, then a? — b? > 0 for a > b; and if p <0, then a? — P< <0 for a> b. Therefore we may assert the following: (a? — bP) (at — b*) >0 if p and q are of the same sign; (a? — bP) (a? — b’) <0 if p and q are of different signs and for any real a and b. Let us first consider the case when pand q are of the same sign. We have aP*1 4 BPS Pht + alb?, aPH1 4 PH ght aP*t 4 [P+ SP ltt ail”, BPH 4 PHT HP Ch 1 HMe?, Adding these inequalities termwise, we get (n—1) (a?*4 4 bP*T 2 4 1PM) > Ya", where a and b (in the last sum) attain all the values from the series a, b, c,..., 1. Adding S'a*? to both members of this inequality, we get n(aPt4 bP 4 IP) Sah 4 bP 4 +1) (at + bt... 1%). The second inequality is obtained just in the same way. From these inequalities we can easily get the results of Prob- lems 36 and 37, Solutions to Sec. 8 424 40. 1° Let A=", m>n. We have a ta) + lied) ( the factor fam of the radicand is taken n times, the factor 1 is taken m—n times) . Hence n (1402) 44+a%. 2° Put aeat and first assume that m>n, ie. X> 14. We have 1< 1—a)ntm—n (inet etmn The factor tam of the radicand is taken n times, and m the factor 1 is taken m—n times. Hence n Cj 1 1 "a ta)" f-at< 7 amt 1 (Q+ay™ <7 m ae n Let us assume now that m (145) i.@. Uns, > Un. Here is one more proof. Without using the theorem on the arithmetic mean, let us prove that nti a\n (tte) > (148) if a>0 and n is a positive integer. Hence Consider the identity tnea tnt A+ te tt8e 12ers T+(n—1)c T+(n—2)¢ 9° 1-22 14+2 1 (x > 0). But tet) a 2 ze t4(n+dye epee Tee one eee (k=0, 4, 2,...,n—1). Therefore fine >[AEGEPe PY Utne > tint tar. ; a Putting here == Taph? We get = — >(1+¢) Solutions to Sec. 8 423 In particular, at a=41, we find >(t+5)" (14+ ex)™ 2° We have vo (14144) Hence for any whole positive k. If k=6, we find (1+ 3)"<(s)'<3 42. We have Vaal nee Je , oe pe nin+t) Tet ety “Yue Vi (see Problem 41). But the fraction Therefore nt —ttet if nds. Yu 43. It is required to prove that ar aoe 7m iT We have 424 Solutions 44, Let us prove that log yi > ay log xy -+ aig log 22+... + ain log tn (i=1, 2, ...,n). To this end it suffices to prove that log (ax + by-+ez+...+lu)>aloga+blogy+...+ + llog u. (*) if a+b+...+1=1 and a, b,...,l are rational positive numbers. Put — Wo Then a+tP+...-A=N. To prove the inequality (+), it is sufficient to prove that ax+ by+cz+...4+-lu>arty?...ul, But we have zty? ul = atye uk Dee LY Yee Ue US Set Bu EM grt by +... lu. Thus, it is proved that log yi> ai; log 2; + aig log z+ ..-+ ain log zn (i=1, 2, ...,n). Hence n 7 2 log ys (log x1) 2 ain + (log tz) 3 diet. + é é { + (log tn) 3) din. or n D> log ys > log 2; + log 2, -+ ... + log rq == log 2422... an- im Solutions to Sec. & 425 Finally YaYo «++ YnDeUyL ».. Lp. 45. Put gaa (i=1, 2, ...,n). Then we have to prove the inequality Y Oa) OF a). Ta) S14 tite The theorem is valid at n=1, 2, 3 (see Problems 21 and 28). Suppose it is true at n=m and let us prove that it also holds at n=2m. We have =V V+ 2) 1+ ay) V0 +s) (1+) 2 WV EF tomes) LP om) > SV (14 V aaa) (1+ V tat) «(LFV tame) > St4$V Va Visti... Vimitm = = 147 tite am. Thus, the theorem is valid for all indices equal to any power of two. Let us now prove that it is true for any whole n. Let n+q=2”. Then WTF a) +m). 1+ an) (19) dm) EFS D1 4+ airy... enya --- Yq Put 1ty=Itm=...=1+y= = (Fa) i-Fm) -- (iFa,)=¥. We have Sit am. a, (Y—1P But (A+ ay) (1+ a)... (1 4+-¢n) =” 426 Solutions Therefore MYYYIS14"Ys,...a, 0-1, ie. Yoit aa... an(¥—1)! or (Y —1)™4 Saya... tn (Y —1)%. Hence (Y 1)" DS ayty ... try Y-137 aa,.0:a- Finally Y= Oa) Om) Fan) St tite en and the theorem is proved. The equality sign is possible only if 4] =z, =... = =”, =1. 46. This theorem, as the previous one, is proved using Cauchy’s method. The proposition is valid at n = 4; let us first prove that it holds true at n = 2, i.e. prove that Lay \k_ ata (4z*) <<“ (*) for any whole positive k. At k = 14 the last inequality really takes place. Assuming the validity of this inequality at k =1, let us prove its validity at k =1 + 1. And so, we have (by supposition) “(xpfa9)t — at tah vee 0. Therefore (zepayit caltnalt 2 = 2 and the inequality (s) is proved for any whole k. And so, our basic proposition is valid at n = 2. Let us now prove that if it is true at m =m, then it is also true at n = 2m. Indeed ( ayttetag ryt. + tam1tzem y‘= 2m Feet ayt+zy , r34+24 Zom—1-+ tam \k es Saeonedtet tal carmen SSeS ec EE arte \R, (asta \h (45) (FA) m < k Je ata aptat 23m—1 + 22m 2 S m thebtabeshttebn tim = om : Thus, we have established that the theorem is valid at n equal to some power of two. It remains to prove its validity for any whole n. Put n+ p = 2”. Then ty +t. tnt yrtyet: +s typ i (ea) ah abt ee taht tubte typ . ae Put W=w= ee We have Lyf aye. barby t+ yet... +yp= ie (ert s+- ben) (n+) n 428 Solutions Hence aptief... tan \h ee ee (aa) a Finally n n (ee aetna and the proposition is completely proved. It is easy to establish that the equality sign is possible only if Uy tg =... =X. 47. This proposition is the generalization of the previous theorems (see Problems 30, 45, 46). The proof is carried out in the same way as in the mentioned theorems. Namely, assuming the validity of the theorem at n =m, let us prove its validity at n = 2m. We have tte tam—1t tom pie yg fama tem t+. 2 2 o(4 Bn mJ —log (1-+-E 2). Thus, the function @ () = —log (1 +1) really possesses the following property o( nth lc @ (ts) + & (ta) and therefore it must be o( tytig tee ttn )< (41) +9 (t2) +. +. 4 n n ie. —log (1 + blah bin )< e_t t)-+ log (1-++ to)-+.. -tlog (ttn) n log f/ (1-4) 2+ 4). +hh) < 1). h T en t+ tite op (“F*) =log(t+e ? ), 2) Fe () _ Jog V Oe) (1 Fe). Since 2 trite VEFM AF > 1467, fulfilled for the function @(¢) is the inequality o( hte )< ete) (tt). Therefore #( ty-pigb eee ttn \< Pt. Folin) ‘ n ie. tttate. tty, ei i th) log(t+e7 < eee log (he ; tettet tty 146" log A1+...-Hlog A, >i+e 7 432, Solutions Finally VUFR) IER). Aan > 14+ I 48. Let t;, to, ...,tn be contained in the interval bet- ween 0 and x. (O0 en ttte «9 ty—te =sin—y+ -2 sin?@ (in our case g(t) = —sin 2). Thus sin tsinat.. tsintn ain tte} tin (if O0); and this greatest value is attained at Y=a=.. a a,=—. n 49. Let us prove that the difference zP—1 at—1 P q (if c ~ 1 and p > q) exceeds zero. To this end it is suffici- ent to prove that A= q(2?—1)—p(at—1) >0. First let us assume that «> 1. We have : Aesg (2? =A) — p(t?) = (2-1) (gay + +24+1)—p(xtt att... 4241} =(e—1) {g (2? t+ +aP 24... 42%) —(p—q) (at 4+ at? 4 ...4044)}. If > 1, then xP Tt oP? +... + 21> (p—qg) at. Therefore A= q(x? —1)— p(2*—1) > (e@—1) {9 (p—g) 21 — —(P~9) qa} = qt (p—q) (t&— 1)? > 0. Thus, if z> 1, the theorem is proved. Now let us assume that <1. In this case we have Prd eh 24 tat < (p—q) x4, etal? |. fat 1 > get, g(x? tt... +24) —(p—g)(2tt+... F424 < <(P—49) qe" —q (pg) 21 =q(p—q) at (z—1). 28-1225 434 Solutions Consequently A>4a(p—q) 24 (z—1P > 0. However, this proposition can be proved proceeding from the theorem on the arithmetic mean. We have the following inequality (see Problem 40) ({ta)*>414+02 (4 > 4, rational, a > 0, real). Likewise we can deduce the following inequality (1—a)* > 1-02 if O1, rational. Using these inequalities, we shall prove that aP—41 raI—1 P q if p>q («#)). Put «t= £, ah Then we have to prove Be—1>2(E-1) B41 -A(E—1>0. First suppose z > 1, § >1. Put & =1 -++ a. We then have Bo41_aA(E—1) =U +ay-—1—da >. If <1, then § <1. In this case we put E=1—a (0 0. 50. Let us first assume that m >> 1. Put m = fp >a or positive integer). We then have (see Problem 49) teeta 7 (E41). 1 Putting E’=2, E=27, we get a™—1>m(x—1). Solutions to Sec. 8 435 Replacing in this inequality x by 4, we find 4 4 aol >m (=-1)- Multiplying both members of this inequality by — 2", we get 2™—1< max"! (x—1). Thus, if m> 1, then m2”! (x—1) >a™-—-1 > m(x—1). (1) Let us assume now that O--(@—-1). Replacing here x by 2”, we find a™—1 0, rational. Let us first prove that if m is nega- tive, then z™—1>m(x—1). Since n> 0, it follows ‘that n+1> 1 and we may make use of inequalities (1). Namely, we have a™t—_4 < (n+ 1) 2"(x—1). Hence nz" (x—1)>a"—1. Replacing here n by—m, we find —mx"(«—-1)>2°"—1. Multiplying both members of this inequality by —2”, we get z™—1>m(x—1). 28" 436 Solutions And if we replace here x by 4, then we find a™—1< m2" (x—1). Thus, indeed mz™1 (x—1) << 2™—1< m(x—1), if0 way, fam. Then the inequality is rewritten as follows vit yet yr \"— wtyP+... ee + where m > 1, rational. Using the results of Problem 47, it is sufficient to prove that ( atts y"< een for any rational m > 1 and for any real positive t, and fp. In other words, it is sufficient to prove that Qty ym Qty ym (ae) + (ae) S2 (4) Let us make use of the results of Problem 541 (A42)">14 mz if m > is rational and 1 + 2 > 0. We have two inequa- lities (aaty "St +m (a54-—-1), ( ea "St+m( Tat -1). Solutions to Sec. 8 437 Adding them, we get inequality (1) which is the required result. The solution to our problem can be obtained imme- diately from the inequalities of Problem 51. Let us show that, using this method, we can deduce even a more general inequality. So let us prove that ( yityot +n \\< uetyht... tuk aa n if A is a rational number not lying in the interval between zero and unity and ywityot... tun \h eth t.. tuk ( aye C n ) Soe = if 0<2<1. To prove the first inequality it is sufficient to prove that (Gato) + (Gao) tt But we have (see Problem 51) nye a nyt HE ( vit yet Fun ) aita( vit yet. Fyn ‘). Putting here i =1,2,..., and adding the inequalities thus obtained, we actually get inequality (2). We proceed quite analogously for the case 0) (ai —22))*, Bi C?=S2+ 2g, wherefrom we find nS, =C?+ > (x: —2;)*- j>t The last equality shows that S, takes the least value when the least value is attained by >) (t; — 2;)*. The least value of this sum is equal to zero and is attained at a a a But since at+a,+... +4, =C, it follows that at. bak Solutions to Sec. 8 439 takes on the least value at _ Cc ==... ==, 55. First let us assume that 4 does not lie in the interval between 0 and 1. Then the following inequality takes place taht... 2h t+aet... +n \h ae 2> (4 et a), the equality sign (as it is easy to find out) occurring only if eg ee tae If it is given that ttt... +t =C, then at all values of 2, %,..., 2), we have ata... taron(S), wherefrom it is seen that the least value of the expression ahah... bah is n (£y which is reached at 2;=2)=...=2,= ae But if O...= = -VC. 58. First let us assume that p; (i =1,2,...,m) are whole numbers. We have Bithethst.. eee rm He Bn att aa me V Bows 7 Bs We in Ma m7 +He et +pn Tat Cc E pitbet--.-+pn Mit... pln Consequently wituat. - wighe m ppt pee. apap... aie< (Ta) Ht ce and the equality sign is obtained only if zy Ht Ha Solutions to Sec. 8 444 Let now p; be fractions. Reducing them to a common denominator, we put where A; and p are positive integers. Since xpicve ... cn = aoe. an, the greatest value is reached by the product xix? ... un simultaneously with the product xxl... 2%, where Aj are integers. As follows from the above-proved, it hap- pens if and only if ped eee tee =an MWe Dividing the denominators by w, we get eae ee zn Bro He Ba” Thus, if 2/0 and 2;+-a,4+...+2,=C, then the pro- duct atizt?...ciin(u;>0, rational) attains the greatest value if and only if Mm tn wo Ba Bn” 59. We have aya ane GX {+ Ogly ... nin ar. 65. Since sin? « + cos? z = 1, i.e. the sum of the two quantities sin?z and cos*z is constant, their product sin? x- cos’ x reaches the greatest value when these quan- tities are equal to each other. It happens at z = +. However, = the same is easily seen from the identity Hy ae sin x-cos z= > sin 2x. 66. It is known that if rye then tanztany + tanztanz-+ tanytanz =1 (see Problem 40, 4°, Sec. 2). Thus, the sum of the three quantities tanztany, tanztanz, tan y tanz is constant. Therefore, the product of these quantities tan® z tan? y tan®z reaches the greatest value if tanztany = tanztanz = tan y tan z, i.e. if tanz = tany = tanz and consequently at. a=y Solutions to Sec. 8 49 67. We have 4 4 4 4 4 4 wat tapets teeters (arte) 4 (4 1 1 an 2 + ( wpD +3)+ + tRET = GED Gneh + dn 42 bn4+2 tape t+ ep > n 1 > (in+2) { Gat + Baap } 68. Put a=, It is required to prove that a" {>n(a"!—a™), Or, which is the same, 2n oP —15>na" (81), ob nant, But een 4 20n=-1? /2iN—-2 2 gat = +a f+... ¢ 41> Sn ea... a (using the theorem on the arithmetic and the geometric mean of several numbers). Since 24+44+...4 (2n—2) =n(n—1), we have indeed o2n—1 n-1 wat nant, 69. Rewrite the sum in the following way t4$+(fte)+(F+Gtptw) tet +(gegrt ter) taegat tet Each of the bracketed expressions exceeds + and, consequent- ly, the total sum is more than t On the other hand, the 29-1225 450 Solutions sum may be rewritten as 1 4 4 1 1 4 t4(gtag)t(ststetz)t et 4 4 4 +(gertaegrt + +a): But each of the bracketed expressions is less than unity, and, consequently, the total sum is less than n. 70. On transformation we get the inequality (a +6) (a + ¥) (b +4) (¢ +4) — —(@+b+e+4+d (+a ab— —(a+b+e+dcd(a+b) >0, or the following one (ad ~ bc)? > 0. SOLUTIONS TO SECTION 9 1. Putting in the basic formula n = 1, we find vg = 8y, — 2v9 = 3-3 — 2-2 =5 = 2? +14. Suppose that vp =2* +4 (eh =1,2,...,7), and let us prove that nes = 24-4. Indeed Vays = 8Vq— QWy-y = 3 (2" + 1) — 2 (2"1 4 1) = =3-2"4+3—2"—2=2" (3—1)+1=2"1+1. 2. Solved as the preceding problem. 3. As is easily seen, the required relation is indeed valid atn = 1. Assuming its validity at the subscript equal to n, let us prove that it is also valid at the subscript equal to n + 4. Solutions to Sec. 9 454 Indeed @nss— VA ant VA =) a3—2VAa,+A =( a,—VA )? ~@42VAa,+A \a,tVA But by supposition a,—VA =( a—VA jen an+VA \at+VA i Therefore fu VA _(to=VA )?_ anus} VA on +VA =( a—VA yr ( a—VA Na a+ VA a+VA) 4. We have m= tH, ga tte, g— wie, 4, ste Hence =a, = Ft, as— a=, a —ag= BE, Consequently It is easy to see that there exists the following general formula yraita oeT: (-4 An — An-4 29" 452 Solutions Adding term by term all the last formulas, we have an— ay = oo. z)= ay — a a4—4 a4— 49 -1 4 — 49 soy $b (Aye Wrz ea {(- yrs Henee, finally, dy EY (AE 5. Consider the relationship Qy, = 3a,44+ 1. Putting here k equal to 2, 3, 4, ..., n, we get a = Yan =3 >) aaa tn—t. k=2 h=2 Put Q4+ta,t+...+a,=S,. We then have S—a =3(S —a,)+n—1. Consequently 4 S => (3a, — a —n+ 4}. It remains to express a, in terms of a,. We have Gy = 38Gn-1 +1, Gp = 3a,-2 + 1. Hence Gy — Any = 3 (An-1 — Gy-2)- Therefore Gy — On-y = 3 (dq — Gq-2) = 3° (an-p — Anz) = = 3° (@n-3 — dn) =... = 3? (a, — ay). Solutions to Sec. 9 453 But a, = 3a, +14 =7. And so Q, — Ay4 = 5-3", Putting here n equal to 2, 3, 4, ..., n, we have a, —a, = 5-4, a3 — a, = 5°3, a, — a3 =5°3*, G, — An-1 = 5-3", Adding these equalities termwise, we find a —ay =5U44+343874...4 37%) = 5 gn-1 = 3 @"t— 1). Rewrite the expression for S in the following way S=443 (a,—a)) + 24-2 + 1} = at {F (711) 4-4—n+1} =4(5("—1)—2n). 6. We have ay = kan +l, On = kn + Ll. Consequently Op — Any = (An — Ang) = Ke? (dn_g—ap-3) = --- = =k" (a, — a)). Hence — 04 = (0, — a4), @3— a2 =k (a, —a), a — 0, =k? (a, — ay), Adding these equalities, we find Oy = kta, + 454 Solutions 7. Rewrite the given relationship in the following manner Gn+4 — Gy — (dn — Ons) = 1. Put Gy, — On4 =2, (n =2,3,4,...). We then have Trti—~ Zt, = 1. Putting here n equal to 2, 3,..., n — 4 and adding, we find Ty — tz =n —2. Putting then in the equality Qn — An = Ly n =3,4,...,n and adding, we get a, — a2 =e ++... +a. And so a, =A, +23 +at+...+2y. But Dd a= D) (@t+h—2)=(n—2)+(n—2)4+ k=3 k=3 +(n—3)+ cept (n= 2) e+ CONOR? | Hence dy = ag + (2) 29 + PDO) = a+ (n= 2)(@q— a4) + PVE HSNO (n—-1) —(n—2) a. 8. Put Gnt2 — Anti = The Then the following relationship will take place Sati — 22, + ory = 4. Solutions to Sec. 9 455 Using the result of the preceding problem we have (n—1) (n— reece t= 24 (n—1) 2)—(n—2) 24. But it is obvious that no2 -—Q=4+H+. tina mp. Consequently n-2 mam 3 (= 1) (e—2)4 n-2 n-2 tH Db D— 3 2) k= kai Finally ay = CaO a, (n—3) (n—1) a9 + Oem gy (n=) (n= 2) (23) : 9. The required formulas can be deduced by the method of mathematical induction. It is evident that they take place at n = 1. Since Gn-1tbn-1 cS ott ont | assuming that the formulas are valid at n—14, let us prove their we at x. By supposition, we have ani=o+-4(0—a) (1—az), 2 1 besmats(b—a) (1497) - Then 7 a, = Set tbat g +4 -9 (1-4) and, consequently, this formula takes place for any whole positive n. It only remains to prove that the formula for b, is true for any whole positive n as well. 456 Solutions We have One 2 4 by ent a +5 (ba) (1+ 59x) and the proof is completed. However, this problem can be solved in quite a different way. It is obvious that ages Pat | py a Sect tBnct Multiplying both members of these equalities by some factor A, we get 4 On f Mn = (ZA) tret(F4+$A) dns Let us choose A so that edie (beba)a There will be two required values of 4, and they will be the roots of the equation M—A—2 =0, ice. will be equal to Ay =2 and A, = —4. And so, at these values of 4 there exists the equality dn +n = (+E A) (nat t Mins), which holds true for all whole positive values of n. Put- ting here n consecutively equal to 1, 2, 3, ..., n, we get a+ y= (F+-$A) (a+ 28), t+ Ma= (F474) (a+ Mi), THEA) (nat Mand). On + Abn = ( Multiplying these equalities termwise, we find Onda = (Z4-P 4)" (a+ 20), Solutions to Sec. 9 457 for any whole positive n and at \=2 and —1. Substitut- ing these values of 4, we find Gn + 2bp = a + 2b, 4 Qn — bn = (a—9), wherefrom we have indeed a= a+ (b—a)( 1=a), bn=a+ > 2 (ba) ( 14455 = ) : 10. We have Lp =Ip»~+2 sin? a Yrs, Yn = 2 COS? & Zp_4+ Yn—1- Multiplying the second equality by 4 and adding the first one, we get Xn Ayn = (1+ 20 cos? a) ay_y+ (2 sin? a + A) Yn-t- Let us choose 4 so that the following equality takes place (2 sin? a-+-A) =A (1+ 22. cos? a). Hence A= tana. We then obtain (2n + Ayn) = (1 + 24 cos? a) (tn-1-+ AYn-1) or (2n + Ayn) = (1 + 22 cos? x)” (29 + Ayo). Substituting the values of zo and y and putting in succes- sion 4 =tana and 4 = —tana, we find the following two equalities at, + ya-tanuw = (1 + sin 2a)" sin a, 2, — Ynstana = — (4 — sin 2a)" sin a. Hence sin a {(4 + sin 2a)" — (4 —sin 2a)"}, cos a {(1 + sin 2a)"+ ({ — sin 2a)"}. we 4 458 Solutions 11. As in the two previous problems, we get Tn t+ ayn = BE (Zo Ayo), Tat ayn = Uz (20+ ayo), where py =@ + My, Po =a + Agy, Ay and Ag being the Toots of the quadratic equation (B + 48) =A (a + Ay). If 44 A, then we have two equations for determining two unknowns z, and y,, and the problem is solved. Let us now assume that A, = A,. Then pw; = pz and the two equations coincide. To determine z, and y, proceed as follows. We have En = — Man + wt (20+ AaYo)- (+) Substituting the value of z, into the second of the original equalities, we find Yn = VL —IaYnaa + WE (Zot Mayo] + SYn—1- Hence Yn (Vay 8) Yas = PHT (o-+ Mayo). Put yn=ptzn. Then for z, we obtain the following relation Hien + (pay —$) 2n-1 = V (20+ AnYo) or — b=vM a= nt Ble + Anyo), wherefrom we find z, (see Problem 6) and then yp; 2, is found by the formula («). 12. Rewrite the given relationship in the following way Ln — Op — BEp_g = 0. Put a=a+b, = —ab (i.e. a@ and b are the roots of the quadratic equation s? — Solutions to Sec. 9 459 Solutions to Seo. 9 — as — B=0). Then we have Bq — Aly» — bEn-4 + abty_2 = 0, Tq — AL_-4 — D (Zy-4 — OE q-2) = 0. Put Ty — OEn-1 = Yrs The given relationship takes the form Yn — bYn-+ = 0. Hence Yn = bYn-t» Yn-1 = OYn-2 Y2 = by: Consequently Yn =O" ty. For finding z, we now have Tp — Ln =O" 'Y,. Put z,=b"z,, then ban — O2n-1= Yi or a Bn ant. Using the result of Problem 6, we find a\n-1 a) 1 ey, (F ™m=(+) pe ‘ $71 Performing simple transformations, we finally obtain an_pn ant pnt n= Gap 0b ee However, this problem can be solved by the method used in the previous problem, if we consider two sequences 2, and y, defined by the relationships By = OE_t + BY ats Yn = 1°ta-t + O- Una | I 460 Solutions 13. Solved as the preceding problem. In this case oo Bead a=1, b= page 14. Considering the two variables y, and z,, determined by the relationships Yn = Ona + Boats Zn = Wnt + ben-sy we put Then the variable z, will satisfy the given relationship etna tb yenatd ’ and the solution of our problem will be reduced to that of Problem 11. For instance, in the given particular case Iq = zn Ht an at 8 we have Yn = Yn-t + It» Zn = Yn-t + Ben4 and so on. The second particular case tnt on Bena FE is readily considered in the following way. Rewrite this relationship as follows A Bengt 1 Zn nd a+ Trt” Then 4 4 Tr tnt Putting here n=1, 2, 3, ..., m and adding, we get 4 1 20 ZowT em Tn =a * 15. It is easily seen that an4t0nt1 = Anbns Solutions to Sec. 9 461 and, consequently, Anby = Andy at any whole n. But Va— Van—Vin _ Vint Vin @nttbnet _4/——— Smt Pat Varin Sat PPh 4 Vania Van—Vin Put Van Von n- Then we have 2 Un-1 = Un—2» w=ui, 2 Uy =Up- Raising consecutively these equalities to the powers 1, 2, 2, , 2", we find But tn Vaahy Vanat Vin ena+ Vagbo Vame—Vbq _ 0—Vaghy Vaot Vb a+ Vagby * Uy = Therefore we have tot Vid _ (%9— Vay?" ant Vagbo \ao-+Vaobo/ 462 Solutions 16. We have 1 1 1 1 (es — 2k ~ 2k (Akz—1 -Glaa ~2eF1 |= 1 f2k—(2k—1) (2k +1)—2k eo a 2 \ 2k@k—1) 2k) ws ~ g i ie x os | x | ~ 2 morc {(14+$+ ‘ +39) + Hatt tama) tea? (ate + -- +a) }= = {2 (145+ +a i) toon ee a 4 Hence 1 4 4 in apitapat tie (see Problem 33, Sec. 1). 17. Let us denote our expression by g, (x). We have q (7) =(4—2) +2 =1, (7) =A —ad—2)+e(t—o) +o =1, wherefrom we can assume that 9, (z) = 1 for any n. It is easily seen that the following relation takes place Qnss (Z) = (1-29) Gn (2) +2. Assuming 9, (z) =1, from the last relation we obtain Qn+1 (z) = 1. But since q, (z) = 1, it follows that @, (z) = = 1 for any whole positive n. Solutions to Sec. 9 463 18. Put gn-t ai = n (2). z 2 2 tea tqoa te +7 Then 2 Qn+1 (2) = Pn (2) ig Now it is easy to prove the sequired formula using the induction method. 19. Pat (+2) (442%) (42%) 0.0" )=X, Multiplying both members by 1—2z, we find X(A—2)=[4—a) 14-2142) (442%) 20. 142) = =[(1—2%) (4-2) 142%) 2.0. 4 22™ 4) = =[(1—24) A +24)] (442% 2. (44224) = 0. = 1-2". Hence SH=14+2e4P434...422"-1, 1 14tactt, ott attt ott @ +141) 145 trate ab Let us assume that 1 a44 (a4) (O44)... +4) Ste + +} as ($A) O44). 41) E44) abe... sk = Adding @+1O+4 -.. (644) e441) co... skl to both members, we 464 Solutions get (oA EA) (641) EAN) | ON) OH Ns 1) eA) abe... sk abe ss ak ee ee abe ... ski and the formula is proved by the induction method. 21. We have pee Diesen (GO) lee ath) a+b) a e — @+b+0)—(a+b) @FHCPO+e) @ Fh) @Fo+e) t 4 4 ~G--o+...Pe+E * Adding these equalities term by term, we find b c zetht @PHhe@tepay tt g + @FEE FH OTOP TRED 4 4 bet... +k+1 =a @pot...+k+T a@pb+e}...+k+) and the identity is proved. 22. We have Fi@)=75, 4), Fs(@) = 441-2). Hence 1+ Fi(2)— Faq) =1+75 (1-2) — 74 (1 — 92) = 1, i.e. the identity is true 7 n=1. But Fr(2)= =Fra()t poy ge tae) - - (gz), Fp (92) = Fn (02) + qe (1—92) (1 g"2) «-. (1—-9"2).

) cos 2(p—k) la =cos&+ ... +008 n§ = ———— tot sin > (see the solution of Problem 11). Let us denote ne ne n sin — cos 2 1 2 SEE ace = Oe sin 5 Then we can prove that 6, = 0 if k is of the same parity as p {k =p (mod 2)} and o, = —1 if k and p are of different parity {k =p + 1 (mod 2)}, and we get p= —1)p 4 sa (HAF CE, +0 aap Cho Pt k=0 k=p-+1 (mod 2) Hence p-t 1 1 Ss! Re P S=357 >) Cp +n ay Cho- k=0 k=p+i (mod 2) * JR 2p—2 But we can prove that x Cop =2°P-* (see Pro- hep+t (mod 2) blem 58 of this section) and our formula is deduced. 14. 1° Rewrite the polynomial as 2” —a"—nza""! + na" =(«” — a") —na™ (4-0) = =:(c—a)(x"* az"? 4 ...+a™'— na"), At x = a the second factor of the last product vanishes and, consequently, is divisible by x — a; therefore the given polynomial is divisible by (x — a)?. 2° Let us denote the polynomial by P, and set up the difference P, — P,-. Transforming this difference, we easily prove that it is divisible by (1 — z)®. Since it is true 296 Solutions for any positive n, we obtain a number of equalities Pa — Pra = (t — 2) On (2), Pyar — Pa-2 = (1 — 2)? Ona @), Ps — Pz = (1 — 2)° @2 (2), P,—P,= (1 —2z) 1 @), where q; (z) are polynomials with respect to z. Hence Pn — Py = (1 — 2)* (2). Py = (1 — 2), it follows that P, is divisible by (1 — z)® and our proposi- tion is proved. : 15. 1° Considering the given expression as a polynomial in y, let us put y = 0. We see that at y = 0 the polynomial vanishes (for any z). Therefore our polynomial is divisible by y. Since it is symmetrical both with respect to z and y (remains unchanged on permutation of these letters), it is divisible by z as well. Thus, the polynomial is divisible by zy. To prove that it is divisible by x + y, let us put in it y = —z. It is evident that for odd n we have (c@ — 2)" — a" — (—2) = 0. Consequently, our polynomial is divisible by z + y. It only temains to prove the divisibility of the polynomial by a + ay + y? = (y — ze) (y — ze*), But since where e+et1i=0. For this purpose it only remains to replace y first by ze and then by ze? and to make sure that with these substitu- tions the polynomial vanishes. Since, by hypothesis, n is not divisible by three, it follows that n = 31+ 1 or 31+ 2. At y = ze the polynomial attains the following value (+ xe)" — 2" —(xe)" = a" {2 + 4+ e"} = 2" (14+ e+4e%)=0. Likewise we prove that at y = ze” the polynomial vanishes as well, and, consequently, its divisibility by zy (x + y) x xX (x? + zy + y?) is proved. Solutions to Sec. 6 297 2° To prove this statement let us proceed as follows. Let the quantities —z, —y and z+ y be the roots of a cubic equation @ — ra? — pa —gq=29. Then, by virtue of the known relations between the roots of an equation and its coefficients (see the beginning of this section), we have r= —x—yt+(e¢+y)=0, —p=ay—a2(z+y)— —y@+y), q = zy (x + y). Thus,—z,—y and z+-y are the roots of the following ‘equation 0, a — pa —q where : p=2+ ay +y, g = ay (x + y). (2) + (y+ (e ty)" =Sn- Between successive values of S, there exist the following relationships Put Snia= PSnit FOS ns S; being equal to zero. Let us prove that S, is divisible by p® if n=1(mod 6) using the method of mathematical induction. Suppose S, is divisible by p? and prove that then S,4¢ is also divisible by p?. We have : Snte=PSntg t+ QSnis, Snig= PSnsz+ 9S atte Therefore Sno = P (PSnt2 + GSn4s) + 9 (PSuts + 9Sn) = = D*Snze + 2pqgSnti + PSp Since, by supposition, S,, is divisible by p?, it suffices to prove that S,41 is divisible by p. Thus, we only have to prove that (e+ y)"+(—2)"+(—y)" is divisible by x? + zy + y® if n = 2 (mod 6). Proceeding in the same way as in 1°, we easily prove our assertion. And so, assuming that S,, is divisible by p?, we have proved that S,4, is also divisible by p?. But S, = 0 is divisible 298 Solutions by p®. Consequently, =(t+y)"—2"—y" is divisible by (x? + zy + y*) at any n= 1 (mod 6). It only remains to prove its divisibility by « + y and by czy. 16. Equality 1° is obvious. From Problem 15 it follows that (x + y)® — 2° — y® is divisible by zy (x + y) (2? + + zy + y’). Since both the polynomials (x + y)> — 2° — y5 and zy (x + y) (2? + zy + y*) are homogeneous with respect to x and y of one and the same power, the quotient of division (x + y)® — 2° — y® by zy (x + y)(x?, + zy + y?) will be a certain quantity independent of x and y. Let us denote it by A. We then have (+ y)® — a — y = Ay (w+ y) @? + ay + y?). Since this equality represents an identity and, hence, holds for all values of x and y, let us put here, for instance, x=1,y =1. We get 2—1-—1=A-2.3, Hence A = 5, and we finally get (z + y)> — a — y® = Bay (@ + y) @? + ay + y?). Using the result of Problem 15 (2°), we can write similarly (@ + yi — a? — y? = Aay (x + y) (+ zy + y')?. Putting here z = y -=1, we find A =7. 17. It is known that @+y+7—P— xP —P =3e+pY4+D ty. Let us prove that (# + y + 2)™ — 2™ — y™ — 2™ is di- visible by « + y. Considering our polynomial rearranged in powers of z, we put in it z = —y. We have (—y +y +2)" — (—y)™ — ym — 2 = 0, since m is odd. Consequently, our polynomial is divisible by (x + y). Likewise we make sure that it is divisible by (x + 2) and by (y + 2). 18. The condition necessary and sufficient for a polyno- mial f (z) to be divisible by x — a consists in that f(a) = Solutions to See. 6 299 = 0. Put f (a) = @ + kyzr = > + 2. For this polynomial to be divisible by z+ y+ it is necessary and sufficient that f(-y—2) =0. However f(i-y-) =—-Yt+7-—kyeyt+yat+yte= =-& +3) yxy +2), wherefrom follows k = —3. Thus, fer 23+ y® + 23 + + kxyz to be divisible by z + y +z it is necessary and sufficient that k = —3. 19. Divide n by p. We get n = lp + r, where 1 is a posi- tive integer and 0 — a" by 2? — a? it is necessary and sufficient that 2” — a is divisible by x? — ap, But it is possible only when r = 0, and, consequently, n = lp. Finally, for z* — a” to be divisible by x? — a? it is necessary and sufficient that n is divisible by p. 20. Put f (x) = 2 4 ato*t 4 gtet®? + 4d+3, On the other hand, PHo@tetia(@+)(@+1)= =(@+N) (e+) (e~d. It only remains to show that f(-1) =f =f (—-) = 0. 21. We have 1+at+att fata = an zt" Lpetet... parts an It is required to find out at what n ——- Bo be a polynomial in x. We find went md ape er ae 300 Solutions For 2" + 1 to be divisible by z + 1 it is necessary and sufficient that (—1)" + 1 =9, ie. that n is odd. Thus, 1+ 27+ ... + 2%? is divisible by 1+ 2+ +tat*+ ... +271 if n is odd. 22. 1° Put f (t) = (cos p + z sin g)" — cos ng — z sin ng. But. e@+it=(¢ + )(¢ — i and f(i) = = (cos'@ + isin g)" — (cos np + i sin ng) = 0 (by de Mo- ivre’s formula). Likewise we make sure that f (—i) = 0, and our supposition is proved. 2° Resolve the polynomial x? — 2px cos g + p? into factors linear in z. For this purpose find the roots of the quadratic equation aw? — 29zcosp +p? =0. We get z=pcosp +V p®cos? p—p*=p(cosp + ising). Let us denote x" sin p—p" “a sin np + p” sin (n — 1) p= f (z). We have to prove that f [p(cos p + isin g) = 0. 23. Suppose +1 = (e+ pr + Q) (P+ pe+q')= =et+(p+p)®+q+q + pret + (pal + gp’) & + aq’. For determining p, q, p’ and q’ we have four equations p+p'=0, (1) pp’ +q+q'=0, (2) py +qp' =0, (3) qq =1. (4) From (4) and (3) we find p’ = —p, p (q’ — 9) = 0. 1° Assume p = 0, p’ = 0,¢ +9 =0,q7 =1,7 = qz= ti, gd = Fi. —1 Solutions to Sec. 6 304 The corresponding factorization has the form xt 1 = (2? + i) (v? — i). 2g =qge=1,¢= 41. Suppose first q’ = q = 1. Then pp’ = —2, p+ p’ = 0, 2,p=4V2, p= V2. The corresponding facto- ization is att 4 = (2? — V2x + 1) (2? + V2e + 1). Assume then q=q =—1, pt+p' =0, pp =2, p= +V2i, p= ¥V2i. The factorization will be at441 = (2? + V2ix — 1) (x? — Vix — 1). 24. Put. Vatybi=r+-yi, whence a+ bi = 2? — y? + 2Qzyi; consequently, e—yaa, Ay =b. To find z and y it only remains to solve this system of two equations in two unknowns. We have Cee pe a Se ed A therefore P=atVe+b, y= —at+ Vere, r= tVatVeGPe, y=4+V —at+VEPe, the signs of the roots being related as 2zy=b. And so, the following formula takes place Vapbi= + (Vat Veth+i Vf —atVeqe) if b>O (since then the signs of x and y must be the same), and Vartan t (Vor Veru—i fat Vere) ifb<0. 302 Solutions 25. The roots of the given equation are determined by the formula 2kn + ct 2k tk cos [—-+isin =~ = = (cos +isin =)" (k=0, 1, ..., n—1). 26. We have not n=1 where 2m ge Ot e€= cos —- +isin =. Thus not . Pace D4 gap =1 dwatte fet... pem—ip, But 2p + of. mt e? = cos "+i sin a It is obvious that e? =1 if and only if p is divisible by n. In this case s=n. e?A~1, then s=ite?ter?4...4e-Ire =0, since e"?= 1. nat > ap=n if p is divisible by n, =O and not } ap=0 if p is not divisible by n. 50 27. We have not nt tH Dd AcP= > Andre neo 50 iets sinensis Solutions to Sec. 6 303 But AnAn= (z+ ye" + ze +... 4 wer Sk) x x (et yem* 4 ze- 2h 4. fem Dk) — =ar+-yy +... ww+a (yeh 4 ze fp we- 1h) + ye (GE FE-PE. f wer HR) + + ze? (x yeh... 4 wet) 4 + welt-0k (z+ yeh... + we-in-2h), Therefore ete ay Aede=n(lePtlyPt + ]eP+ n-1 ta >) (yeh ze-2* 4. f wen) k=0 nt ae ty >) (wep ze-kE ... f we BA) mo: n-1 +w > (Gelr—Dh 4 yelr—2)k 4 we. fue"), k=1t not But > e'*=0 if J is not divisible by n (see Problem 26). x0 Therefore all the sums in the right member vanish and we get [4oP+]ArP+ .-.+[AnaP=n{lePt+lyP+ +] wep}. 28. 1° Denote the roots of index 2n from unity by x, so that 2sn 7h 2, = cos ™ +i sin (s=41, 2, ..., 2n). Therefore 2n n-1 2n-1 a — t= TI (e— 2) =] (2-2) TT (2@-2)-(e*—1), s=t s=1 sant 304 Solutions since 2= —1, Jen=1. But Tons = Es, consequently, n-1 2" { = (2*— 1) T] (x2) (e@— 2) = sot n-t =(2*—1) TI (#22008 a4 1). sat The rest of the cases are proved similarly. 29. 1° Rewrite the equality 1° of the preceding problem in the following way n-1 gen-2 4 gmk ttt = TT (2*—2zc0s a+ 1) . sot Put in this identity. z=1. We have n-1 n-1 ae us sm) ear an ae n=] (2 2 cos —— )= [] 4 sin’ -- = sot s=t ye cae 7 -1 = 22) gine gine 28 gine (HAA n nm n Hence singin 2... sin 2o® Ve, n n n eee 2° Solved analogously to 1°. 30. We have a1 =(2—1) (2—a) (2—B)(2—y) ... (@—2). Hence atta... +24+1=(4—a)(x—B) ...(a—A). Consequently (1—a)(1—B) ...(1—A) =n. 34. Set up an equation whose roots are a—1t, a—1, 1.2, tat This equation has the form (e+ 1 +(e Att ...+(e+1)+41=0, Solutions to Sec. 6 305 ie. (e4ipi—i _ (epiint _ 9 eee olde eter vd Pte SECT Then set up an equation with the roots pale Sea moi? wi? > 1° It has the form — (ipaymtsentt 9 an Expanding the last expression in powers of z, we find (mpfr + CEO" ory 20 or 2 pay ee The sum of the roots of this equation is equal to -+: Consequently 4 4 1 n Z—1 asa Het en AoE 32. Consider the equation (with ¢ as an unknown) 22 y? Ttet maa By virtue of the given equations this equation has three Toots: y?, v?, 07. Expanding the last equation in powers of ¢, we get t(t — bY (t — 2%) — 2 (t — BY) (t — 2) — —y(t—c)i— 2 (t— byt =0, B+taf®+t...=0, where @ = —b? — c? — x? — y? — 2. But as we know, the roots of this equation are p?, v?, p?. Therefore, it must be wept pha Dt tt att yt tat 20-1225 306 Solutions Hence Ppt eowP+ve+p—P—c 33. Since cosa + isina is the root of the given equa- tion, we have D> pr(cosatisina)**=0 (py=1) k=0 or a (cosa+-isin a)” os Px (cos a+ isin a)* =0. But (cosa +isina)+=cosa—isina, therefore n n > pr(cosa—isina)*=0, >) px (cosak—isinak)=0. k=0 k=0 Hence, indeed, n > P,Sinka = pysina+ p.sin 24+ ...-+prsinna=0. k=0 34. On the basis of the given data we have identically 2 pyx™ 9 pow™ +... + Pat + Pa= =(x—a)(z—b) ... (z—h). Substituting for z first i and then —i and multiplying term- wise, we get the required result. 35. Extracting the two given equations termwise, we find (p — p') a+ (q—q') = 0. (1) Multiplying the first equation by q’ and the second by g and subtracting term by term, we have 2 (q — 9) + 2 (pq' — gp’) = 0 (2) 2 (q' — q) + pa’ — ap’ = 0. Eliminating then x from equations (1) and (2), we obtain the required result. 36. The roots of the equation a= Solutions to Sec. 6 307 are cos #8 4 isin Mt (b=0, 1, 2, ..., 6). Therefore, the roots of the equation i+2+24+ 42274+211=0 (*) will be a= cos 4 jsin Bt (k= 4, 2, 3, 4, 5, 6). Put aa t+>=y, then ephay2, e+ 5ay—3y. Equation (*) may be rewritten in the following way (@+4)+(#+4)4(2+4)41=0. It is evident that t= 7, L_ = Zp, t=, atient T= 2 cos, Hence, we may conclude that the quantities 2a 4n 8n 2cos—-, 2cos—-, 2cos—— are the roots of the following equation y+y—2y—1=0. Let us set = an equation with the = roots V 200s = cos = , VV 200s cos =~ , VV 2008 2 cos = Let the roots of a certain cubic equation —az*+bxr—c=0 be a, B, vy. We then have a+B+y=a, abt+ayt+By=b, afy=c. 308 Solutions Let the equation, whose roots are the quantities */y, be vy a — Ax®?+ Br —C=0. Then a ae i: a+VB+7=A4, VAY BV aV It VY BY P=B, Yapy=c. Let us make use of the following identity (mt p+agy= m+ p+ a+ +3 (m+ p +4) (mp + mg + pg) —3mpq. Putting here instead of m, p and q first ~/a, /B, / 9, and then 3/af, }/ay, ¥/ By, we find A® =a+ 3AB—3C, B®’ = b+ 3BCA — 3C?. In our case we have a = —1, b= —2, c=1, C=1. Hence A® = 3AB —4, B®’ =34B—5. Multiplying these equations and putting AB =z, we nd 8 — 922 + 272 —20=0, (2 — 3)? +7=0, 2=8-—/7, Ab=32—4=5-3YT, A=W 53/7. Therefore, indeed, But Varyiryi= == / 2003 = + V2 cos 4% 4. V/ 200s =V5—377. The second identity is proved in the same way. 37. Since by hypothesis a + b + c =0, we may consi- der that a, b and ¢ are the roots of the following equation + pxr+q=0, where p=ab+ac+be, ¢g = —abe. Solutions to Sec. 6 309 We have (a+ b+)? =a? 4+ b? +c? + 2 (ab + ac.+ be), Le. 8. = —2p. Putting in our equation in turn x =a, z = b, =, we get the following equalities @+pat+q=0, & + pb+q=0, &+pe+q=0. Adding them term by term, we find 83 + ps, + 39g = 0, But since 5 = a + b +c =0, we have s; = —3yq. Multiplying both members of the original equation by a, putting then x = a, b and c, and adding, we find Sht3 = —PSkt1 — Whe Putting here k = 1, 2, 3, 4, we find 8. == 2p, 85 = Spg, 8% = —2p? + 3g, s, = —7pg. Taking advantage of these relationships, we easily prove the first six formulas. The last one is also obtained readily. 38. We have g-usv—y, ®-wv=_y. The second equality may be rewritten as follows (e@—u)(@@+u)——yvVt+y =0. Since cr — u =v — y, the last equality is rewritten as (c« — u) [x +u—(+ yl] =0, wherefrom follows Pc—-u=0,v—y=0,2=4, y =p; 2 (e+ u)— Wty) =0, (2 -—w) —v—y) =00= =v,y=u. Consequently, indeed, zy = u® + vr, 310 Solutions Let us go over to the second case. Suppose z, y, 2 are the roots of a cubic equation a + pat+qatr=0. Prove that u, v and ¢ are the roots of the same equation. We have atytt=—p, yt+u+yz=q, zyz = —r. Hence, to prove that wu, v, and ¢ are the roots of the same equation (whose roots are z, y and z) it is sufficient to prove that utv+t=e2+y+2, wtut+vt= = ay + az + yz, uvt = zyz. The first of these equalities is true by hypothesis. The se- cond one follows immediately from the identity 2(ay tazt+y) =(e+y+27?—(@+ yt 2) and from the condition Y+y+2—w+y+ eB Likewise, the third equality follows from the identity sayz =o +yt2+3(et+y+2x x (ay + a2 + yz) —(@ + y + 2/9 and from the condition P+y+3=uU+U4 6. Thus, u, v, t as well as z, y, z are the roots of the same third- degree equation. Therefore, one of the six possibilities takes place z y y x x 2 y z z z z y gees Solutions to Sec. 6 311 It is obvious that in all cases we have apy tamu P+ 39, Squaring the first trinomial, we get AP = (x} + Qtyrs) + (5+ 2xyap) © + (2+ 2aryrs) €°. Then AB = (x24 23+ 2346212205) + (Bait, + 3a3a, + 3z}x5) e+ + (3xjx3+ 3232; + Baie) &. Put = Xihy + 73g Tyr, P= 2yxZ+ ay_x}+ zya}. Now at at a= —(pay-+ q)— (pa, +9) —(pts+9) = —39, since a +2,+ 273; =0. Furthermore XyXQl3 = —G, A®’ = —9q + 3ae + 3fe?. Substituting x2 and 23, we also find B® = —99 + 3ae? + 3Be. therefore Hence A® + B® = —18q — 3a — 3B = —279, since + B= ayy (ay + 2) + Tots (wp + ws) + + aay (tg + 2) = —Bayx—Qr3 = 3g. A®.BS = —27p3. It should be taken into consideration that oP 3atetet + (ata + 2h -+ he) + ater, + abet + ae eet + aean = 39+ 2th} (+a t+a)t+ + tyes (t{+23+23), eee ee eee eect Pe atatat Te: Likewise we get and 312 Solutions 40. Put a+b=c+d=p. We have (22+ px+ ab) (22+ px+ed)=m or [(2+$)+ 0-2] [(2+4)'4 ed— Let : (+4) Then the equation takes the form (y+ob—) (y-+ed—#) =m, ie. #t (ab+ed—) y+ (ab—#) (cd) —m=0. It only remains to solve this quadratic equation. 41. Make the following substitution a r=y—-—— 5 then a—b tha=y+——, t+b=y——Z The equation takes the form Therefore the equation takes the form —b \2 —b\4 +6 (>) r+(4s-)=4- Thus, the problem is reduced to solving a biquadratic equation. Solutions to Sec. 6 313 42, Put for brevity a+b+c=p and make the substitution z+p=y. We have (y — 4) Y — b) (y —¢) p — abe (y — p) = 0. Hence Pty — (a+b +0) y+ (ab + ac + be) y) — abcy = 0 or y{atb+ocy—(a+b4+o%y+ + (ab + ac + be) (a + b + c) — abc} = 0. And so, we find three values for y: one of them is zero, the other two are obtained as the roots of a quadratic equation. Then it is easy to find the corresponding values of z. 43. Rewrite the equation in the following way (x + a)? — 3be (e# + a) +B + = 0. Put z+ a= y. The equation takes the form y — 38by +B +0=0. But it is known (Problem 20, Sec. 1) that y+ B+ 8 — 3bey = = (y+ b+) (y? + B+ c — yb — ye — be) Consequently, one of the roots of the last equation will be —b —c, the other two are found by solving the quadratic equation. Then we find the corresponding values of z. 44. The equation contains five coefficients: a, b, c, d and e, and there exist two relationships among them. Thus, three coefficients remain arbitrary. Let us express all the coefficients in terms of any three. We have a=c+d, e=b+ec. The' equation takes the form (c + d) xt + ba? + cx? + dx + (b+ 0c) = 0, ce (z* + 2? + 1) + dx (a3 + 1) + b(2@° + 1) = 0. 344 Solutions But e+i=(¢+1)(’—2+1), a+ 1 = (ct + 22741) —2 = (2? + 1 a = =(?+2+1)(—-—2+ 1), The equation is now rewritten as (ato +4) {o(et e+ 1) +de(e +1) + + b(@ + 4)} =0. Equating the first factor to zero, we find aes rapti % . The remaining two roots are found by solving the second quadratic equation. 45. We have the following formula (a+b+2P 840423 4302 (b +2) + + 3b? (a + x) + 32? (a + b) + 6 abe. Using this formula, reduce our equation to the form x — (a + b) a — (a — b)? x + (a — Bb)? (a + b) = 0. Hence x (x — a — b) — (a— b)? (x —a — b) = 0,7 (x — a — b) [x* — (a — b)*] =0, (cx —a — b) (x +a — b) (tx —a +b) =0. Thus, the given equation has three roots: z=a+b, r=a—b, t=b—a. 46. Rewrite the equation as follows a, a2 art? gaa? et (a+2)2 a+r mts a+z* Consequently az y= 2 _2az? (z- aya) ~™ GFE Hence af 2 2ar? ray ater Solutions to Sec. 6 315 2 Put ae =y. Then the equation takes the form y+2ay—m?=0, wherefrom we find y and then z, For y we find the following values y=—atVGTm, (4) The corresponding values of z are determined by the formula 2 wat y/ Eta. (2) Let us take the plus sign in formula (1). In this case the value of y will exceed zero. Computing, by formula (2), the corresponding values of z, we make sure that z has two values: one positive, the other negative. And so, our equa- tion always has at least two real roots, positive and nega- tive. Consider the case when the minus sign is taken in formu- la (4). Now the value of y is negative, and for z to be real it is necessary and sufficient that y? + 4ay >0. And, con- sequently, it must be yt+4a<0, ie. V@+ m+ 4a<0, m > 8a. With this condition satisfied, all the four roots will be real. Since ay < 0, we have + \Vz+|<|4| and, consequently, both real roots, found from formula (1) taken with the minus sign, will be negative. Thus, if all the four roots are real, then one of them is positive, the remaining being negative. 47. Put for brevity Barb t1022+4 pees —!(2)- Then the equation takes the form F (2) +f (@) = az. 316 Solutions Further, we have (r—1)6 48 tf ()= pees t+1)= aE. Dividing the first equation by the second one, we find z—f(z) _/a—1\5 Z4f (a) =( zHt ) : ) Put z—1 a-t_, zt GET From the equation (+) we get z—f(z)=yz+y f(z), z—y)=F(2)(1+y), f(z) _ 1~yd zo t+ ye * Likewise we have f(a) _ 1-06 a 1+ * Now our equation can be rewritten in the following way t—ys 1408 ys 1? whence yo= —B. The last equation has five roots, namely Yr=—be® (k=0, 4, 2, 3, 4); Qn gt 20 e=cos—-+isin=-. But _ i+y ees consequently ap tte Abe et tom eh tye 1fbe® (a+ 1)-+(a—1)e* Solutions to Sec. 6 317 Further 2 T= r= (atije 2 4(a—1)e? See eee eae ak asin _ ae Fete Foot MS ee wk isin HE a(e *+e7)+e eee epi aaen ee In particular, at =O the solution is 48. Transform the left member of the equation. Denote the sum on the left by Sp. Then ay az Ee 2 Salt at Goa eaay ~ Goa eaay Prove that 2m Sm= Gray ena) eam Suppose this equality is true at m=n, and prove that it will be true also at m=n-+1. We have xin + Menor (ea)... (@—4an) © (#44) -.- (Gan) (© aens) ate Agnygr2ntt G—a 7 Reducing the right member to a common denominator and accomplishing all the necessary transformations, we get Sn = =4ans2) s g2nt2 me (eas) + tana)” Now our equation takes the form 22m —2prm4 p2 SEs clan ia ets) (ray (@—4am) or (2™— p) (2"— p) =0. The equation has m double roots. 318 Solutions 49. 1° We have 21+ 2+ 2%3= —p, 24%_+ 2403+ 2y%3= =, T,X2%3= —T. From the second equality we get Eykq + Lyky+ Hf = 1 (2+ F2+43) =], whence aaa. Using the first equality, we find ntoae i From the third equality we have opty . It only remains to set up a quadratic equation satisfied by 22 and 23. 2° Solved analogously to the preceding one. 50. 1° Using the identity of Problem 4 of this section, we can rewrite our system in the following way (y +2 +a) (y + ze + ae’) (y + ze? + ae) =0 (2 + & + b) (2 + xe + be*) (2 + re* + be) =0 (a +y +0) (v& + ye + ce?) (c« + ye? + ce) = 0. To find all the solutions of the given system it is neces- sary to consider all possible (27) combinations. Thus, we get 27 systems, each containing three equations linear in the unknowns 2, y, and z. If each of these systems is designated by a three-digit number in which the place occupied by a certain digit corresponds to the number of the equation and the digit it- self to the number of the factor in this equation, then the 27 systems will be written as 411, 112, 113, 124, 122, 123, 131, 132, 133, 211, 212, 213, 221, 222, 223, 231, 232, 233, 344, 312, 3413, 321, 322, 323, 334, 332, 333. Let us explain, for example, system 243: taken from the first equation is the second factor, from the second— Solutions to Sec. 6 319 the first factor and from the third—the third factor. Thus, system 213 will have the following form yt+zetae® =0, z+2x2+b=0, r+ye®+ce =0. Let us decipher some more systems y+z+a=0, z+x2+b=0, t+yte=0;, (114) y+ze+ae?=0, 2+2e%+be=-0, x+ye+ce’?=0; (232) y+ze?+ae=0, 2-+2e?+be=0, z+ye?+ce=0; (333) y+zta=0, z+2e+be?=0, xtyetce?=0 (122) and so on. 2° We have zw =2xyu+a, yt =zyzu+b, 24 =zyzu+e, ut = zyzu + d. Multiplying these equations and putting zyzu = t, we find A=(¢+a (t+ d(t+o(¢+a. Thus, for determining ¢, we have the following equation (atb+et+@B + (ab+act+...) 0+ + (abe + acd + ...)¢ + abed =0. a+tb+c+d=0, therefore, for finding ¢ we get a quadratic equation. Know- ing t, we easily obtain z, y, z and u. 51. We have 2 — Uta) = 1+ +a)+(t+ay+ + +2)"= “aaa = n+ nti at {> Chae} = >) Chaat h=0 h=t Wherefrom follows that the term containing x* will be A+ ke chtia", However, 52. We have (2A =14Che Chat +. FOr a 4h + 2, 320 Solutions Since this polynomial is multiplied by the second-degree trinomial (s — 2) 2? + nx —s, it is clear that the coefficient of x* in the product will be equal to (s—2) Ce? +nC3'! —sCh. Carrying out all the necessary transformations, we see that the last expression is equal to nes, 53. Put z =1-+ a, where a > 0 (since z > 1). Then we have pet — ga? — p--g= p(-payt—g (1a)? — pt q= (q—1) =P {ttqa+ 2959 att...) — 1 ma {t+pa4+20oP ary...) —pta= =(pC}—4C}) a+ (pCI— aC) a? +... Since g > p, we can prove that all the terms of the above expansion are positive [the coefficient of a* (if k > p) will be equal to pC*]. Thus, to prove the validity of our asser- tion, it is sufficient to prove that A= pCz—9Cp>0 ifg>pandk 0, qg—1>p—1, q—-2>p—2 54. Let the greatest term be Ty=Chr"ak, since Solutions to Sec. 6 324 This term must not be less than the two neighbouring terms T,,-, aud T,,,. Thus, there exist the following inequalities Ty >Tra,y Tr > Tre Whence k = n—k a noRET a Sh ET ‘2s. The first of them yields (n+4)a kS aa From the second one we get (n+41)a ke (n+1)a First assume the “are ‘8 a8 whole number. Then erie 4 is also a whole number, and since k is a whole number satisfying the inequalities @Hi)e 4epe (atthe : +a it can attain two values _ (n44)a _ (eta k ere re-set 1. In this case there are two adjacent terms which are equal to each other but exceed all the rest of the terms. Now consider the case when (ok Ne j is not a whole number. We then have ye Pmt Se -[Set +o where 0<0<1 (for the symbol [ ] see Problem 35, Sec. 1). In this case the inequalities take the form (n+4)a (n+4)a ks [Set +e e[ SE ]-c-9. It is apparent that in this case there exists only one value of & at which our inequalities are satisfied, namely v=[ 2424] 21-1225 322 Solutions And so, when ete is not a whole number, there exists only one greatest term Ty. 55. Let i and nm: be positive integers. We have (e412 = mat 4 AED gen see ma+4. Replacing here z by x-++1, we get (2+ 2)"— (2-1) = am (rp typ EMEb es ayy tm (ett) +4. Subtracting the preceding equality from the last one, we find (a+ 2)"—2(x41)"+ 2" =m (m—1) 274 p+... Analogously we obtain (e+3)"—3 (2 42)" +3 (e+ 1)"—2" = = m(m—1)(m—2) 2° +4 pyx™* 40... Using the method of mathematical induction, we can prove the following general identity (em — (ep ity” $ES2 ep iy et +(—A)iz™=m(m—1) ... Gye tt petit, wherefrom it is easy to obtain that at i=m (2+-my"—~F (e+-m—A)" +... +(— 1)" 2" =m. If i>m, we get (f+ y™— Et i— 1+ 4 HD ei $e (— tie" =0. Putting in the last ee x =0, we find the required identities. 56. We have (e+ ai)" = 2"4+-Cha® ai+Chz"™ a 4-Cic" 038+... = == {x"—Ciz"*a? 4+ Cha" 4a4— ...} Fi (Cha ta— Chea + ..). Solutions to Sec. 6 323 Going over to the conjugate quantities, we get (@—ai)" = {a —C2x”*a? + Cha tat — —i{Ch2"a—Cix™ a+ ...}. Multiplying these equalities term by term, we find the required result. 57. 1° We can write our product in the following way n on an de Dat= Dac, 120 s=0 150 wherefrom it follows that A= > 14. ffl Osssn O Are! #=0 ‘=o fa) 324 Solutions Hence ~ 5, (-1. lastt On, we arrive at the following conclusion if Ln, then 4,=0 when / is odd and A,;=(—1)" when J is even. Thus, A; =0 for any odd l, i.e. the product contains only even powers of zx, and if nm is even, then all the coeffi- cients (of even powers) are equal to +1; if is odd, then half of them is equal to +4, the other half to —4 Ap =42 =... =Ana =H, Anti = Ants = +++ = Aim = —4. 3° We have n n Qn Dy (+4) x* 2 (s+4)2°= 3 Ayt'. Hence Ay= 3 (+1) (s+ 1) = es (ks--1 +14). bike bohen 0Sacn OSsen Let us first assume that 1 UA 414 — k=0 ike v =l Sk > B+ (4 OED) +9) ‘ h=0 h=0 : taking as known that eae... pea Lee pelys 6 (see Problem 25, Sec. 1. Solutions to Sec. 6 325 Then assume n == > = {te=—¢# cos—z ising cos 3 +isin 3 on «Qn u stitae! 2— em =i —— =i ss t+te®=—e cos -—isin-> =cos —isin=> . Therefore 274 ({e)"+ (14 2%)" = 2" 42cos7F Hence, we obtain CLO toet sh (24200822), the other two equalities are obtained similarly by consi- dering the sums ate (1fey-+er (t-te, ar+er({teP+e (ite) 60. The solution is. analogous to that of the preceding problem. Consider (4+ i)”. 61. Since Ch= a we get. 20% =K? —k. Consequently, 3 2 a Ch= > Bee wherefrom our ae is Tease 62. Let a,=Ck, a,=ChtH, as=Cht?, a, =Cht. Then Be nok a4 n—k—-2 83. B—k-A a KEL’ ag kS ' Gg kp2 ~ It only remains to prove that See Eee eee ees aa cy fonts 14 tt t+ Solutions to Sec. 6 327 63. If we rewrite the equality in the form ea 7 nl a nl Tas 1 Bea + Bae ts tyr = then the problem is reduced to proving the following relationship (see Problem 58) ChAC + ...+071= 274, 64. Consider the equality 4 .V3\" 2: sein 2% \" (-44:¥) = (cos + isin) = = cos 2 4 § sin 2 (*) Further 1... V3\"_ (—4n 4 ayn —$+i By -GP -vay"= = 1404 (-iV3) + +0 (—i V3)? +03 (—iV3)?+...J= an =!GP 1—serg =i V3 (CL—3C, + 3°C3 —3°C}, + ...)}- Equating the coefficients of i in both members of the equality («*), we get —V3 (CL —3C3 + 3°Cs — 39°C) 4+...) = (—1)"2"sin ane Hence ie 3 =CL—3C% + 3°C3 — 3907 4. see (— 1 Te sin ae wherefrom we easily obtain s=0 if n=0 (mod 3), s=2"? if n=1 or 2 (mod 6), s=—2"" if n=4 or 5 (mod 6). 65. Consider the expression (fa. 328 Solutions We have (A+y?=14C + Ce4+ 8+ .... Hence +)" =(1-C4C4—Ch 4...) Fi(CL-@ 4+. 8—...). But 14+i=V2 (cost +isin=). Therefore n 2 o=1—C2+4+C1—Ch+...=2° cos nm To? o =C,—C,+C,—-C.t+.. a2 sin Hence, if n=0 (med 4), ic. n=4m, then o=(—1)"™2", 0 =0. If n=1 (mod 4), ie. n=4m-+1, then o=0'=(—1)"2", If n=3 (mod 4), ie. n=4m+3, then 7 4ym game,’ (1) ™ 2241, 2 (mod 4), i.e. n=4m-+2, then o=0, of =(—1)™22", 66. 1° Let us write our sum in the following way Finally, if n kon s=1-C2 42014 3024+... 4 (n-+4) C2 = 3 (&+4) C8, and introduce a new summation variable. Put k=n—k’. Then the sum is rewritten as k’=0 E k=n s= DB (n—k' +1) Cn-* =5 (n~k+1)Ck= hon = Bm +2—-F+ M1 GR= k=n kon = (+2) y oS (k+ 1) Ch = (n+ 2) 2"—s, Solutions to Sec. 6 329 Consequently, 2s=(n+2)2", s=(n+2) 21. This sum can be computed in a somewhat different way. Rewrite it as follows S= (CLE CLE FCN + (CLF 2CR +... + Ch) = 2+ nt 2 Bed, gRO En + tnin—t) tnd =rtn {itt +e y+ m—n4it}= HP pn (CX yb yb oo POM = 2" nd 2" (n-2). 2° We have Ch-2C8 4.3034... + (— DP nChan—22Go) $3 ROTNO— 4AM n= =n {1-4 n—1 fOePe ean eee art} ant" =0. 67. Rewrite the sum in the following manner (=1)rt 4 FO-FGtLQ—... +S ore mi MRE 4 wt {See OHNE am = mt (sap eee — tong 4 +(— 4] 14 F*} = te 68. 1° Consider the following polynomial (Af ay t= 14+ Chet Catt... + Cnya™. Hence (+a)nt—1 n+ Putting x=1, we get the required identity. oe ae Bate Baty. 4 am, 330 Solutions 2° Obtained from the preceding identity at x=2. 69. Put Ch Zt EOF $Y Che tn Then we have 4 —1)(n— ty ting = {nF AGSY 4 FORD) 4b 4 (n—1)(n—2) | 4 (n—1) (n—2) (n—3) = -{ Pee eee see ged oe vs} 4) —1) (n— 2) = toe 7 {7$5t 9} + n(n—1)(n—2)__ (n=) (n=2) (n—3) +z a 12-3 12-3 tees n—1 | (n—1)(n—2) a od eae oa eee 4 n(n—1) , n(n—1)(n—2) aa (es See ee 4 a 4 =F }=—- And so, 1 Un—Una=s + Therefore we may write a number of equalities Adding them term by term, we find 4 4 4 hs Oe 70. 1° We may proceed as follows. The expression on the left is the coefficient of 2” in the following polynomial s=(L Fay" 4(1- 2+ (140) 4... (fz), Solutions to Sec. 6 334 Transforming this polynomial, we have sa(LFa2y"A+(1ta)+(1 tart... +(1+2)}= = Etta 49%), =(1+ 2)" Gant The coefficient of x"* in the braced polynomial is equal to othe. Thus, our proposition is proved. 2° The expression on the left is the coefficient of 2” in the following polynomial a (f+ a)"—2" (14 2)" + 2214 a)" $ 000+ +(—A)Pa™* (14 2)" = (f+) {a"— a +(—4)* a4) = (1p) {2 4 (—1)" ay, It is obvious that the coefficient of 2" in the last expression is equal to (—1)' Ch. 71. 1° Consider the following polynomials (1+2)"= 3 Crt, Aa)" = Sena! We have (1pay" 1-2)" = 2 Cra" > Ch mn = (L$ 2" = Y Chyat?, p=0 wherefrom follows the required equality. 2° Follows from 1°. 72. 1° Consider the product (4-2) (1+2)"= (1-42). We have n 1 Cae” Dy Chat -3 Hence : 332 Solutions Consequently Cha 3B CCh= Sorc = 3 cy" stian 2° In this case we consider the following product (A+2)"(4—2)"=(1—2)", Consequently m m m Xi (—1)° Che? Dy Chat = Si (—1)' Cha", 3=0 t=0 t=0 therefore seg 7D! OmCn= (— 1)! Coe Let us assume first that m is even and put m=2n. l=n. Then sean (TD Ch =(— 1)" Ch Hence an 3 (—1)° (Cin)? = (— 1)" Cine (*) Let 3° If m is odd, then we put m = 2n + 1. The coefficient of z"*+ in the left member of the equality (*) is equal to att pags (TD ChaCha = 2 (9 Chau But the right member of the equality (*) shows that this coefficient must equal zero (since it is evident from the expansion that odd powers of x are absent). Therefore 2nt4 2 (1 Cras? =0 and equality 3° is proved. 4° We have two equalities Chat 2C2a*+...+nChc®=nz(1+2)"4, Cr+ Chat ...4-Chz™ = (1--2)" Solutions to Sec. 6 333 Multiplying them termwise, we find n a a sCia° Py Chat = nx (1-+2). Equating the coefficients of x* in both members of these equalities, we get the required identity. 73. Since the product (x — a) (x — b) is a second-degree trinomial, when divided by it, the polynomial f (x) will necessarily leave a remainder which is a first-degree polyno- mial in x, ax + B. Thus, there exists the following identity f@) =@—4@ @—BQ(e) + ax +B. It only remains to determine @ and f. Putting in this iden- tity first c =a and then z = b, we get f(@ =aa +B, f() =ab +B. But we know that the remainder from dividing f (x) by x — ais equal to f (a), therefore, {(@) =A, 70%) = Thus, for determining a and B we get the fgllowing system of two equations in two unknowns aa+p=A ab+p =B. Hence 4 a=; (4—B), p= aB—bA a—b 74, Reasoning as in the preceding problem, we conclude that the remainder will have the following form ax® + Ba + y. 334 Solutions For determining «, B and y we have the following system a+ pa+y=A ab+ ph+y=B ack + pe+y =C. On determining a, B and y, we may represent the requi- red remainder ax* + Ba + y in the following symmetric form (z—a)(t—c) p | (e@—a)(t—b) =a 09 B+ eae © 75. The remainder will be (t= 22) (23) .-. (t— tm) Cera) (aa) aim) AT ( 1) (— 23) + aay) (ea (t—2m) Grae Bt _(@= 21) (= 29)» (Z—tm-1) T Gin) m— #2) ++ m— Ema) 76. The required polynomial (see the preceding problem) takes the form (z—ay) (tas) ... (cam) (era) 6 a9) 2 (OO) ET (2— a4) (243) «.. (2a) Taga) (gaa) os (gig AEE ay) (2—as (z=am-1) yo (m= Sinai A 77. Our equality states the identity of two polyno- mials. For this purpose it is sufficient to establish that the polynomial Tam a4) (arm (2—29) (2—25) ... (2—2m) FO) Gia) ead) amy +1 (ta) ESSN od (%g— 24) (t2— 23) «+. (72—Zm) Ef (ty) EEE EB) = = 2m-1)_ 7 (2) (fm— 24) «++ (@m—Zm-1) is identically equal to zero. Since the degree of this polyno- mial is equal to m — 1, it suffices to establish that it vani- Solutions to Sec. 6 335 shes at m different values of z. Indeed, it is easy to check that this polynomial is really equal to zero at B=, Lp, Ly, ~~ ., Lm- 78. Obtained from the previous problem by equating the coefficients of 2-1. : 79. If we put in the preceding problem f(x) = 1, 2, z,..., 2™-*, then it will be proved that s, =0 if0< >m —1 it is possible to proceed in the following way. Suppose 2, 22, ...-, Zm satisfy an equation of degree m a py) + pa? +... + Pm + Pm=0, where TPr= y+ 22+ .--+2ms Po ByZq+2_%3t «6. + Emmy — Ps=2yat3+ vo +y Multiplying both members of our equation by a, we get at pat pombt 4 4 pn att pao = 0. Putting in this equality successively @ = 2, 22, ..-; Lm and adding, we find Smth + PiSmyh—1 + PaSmeho2 + +++ + Pm-sSae1 + Pmsn = 0. At k = 0 we have Sm + Pim = Consequently Sm = —Py =% +ag+... +2. At k =1 we obtain Sm+i + Pi8m + P28m-1 = 0. 336 : Solutions Further Smet = (t+ tg + 23+... + 2m)? ~ — (Gg + 6. + omit) = sata t+... tom te, t+ atst..., i.e. Sm+1 is equal to a sum of products of the factors — By Bay oy Lm taken pairwise. Here the factors may bé both equal and unequal. Similar results can be obtained for Sn42, Sm+3 and so on. The same results can be obtained using a more elegant method (Gauss, Theoria interpolationis methodo nova tractata). Put 4 Goa e-a) Gm 7 . 4 (aay) aa) em) 4 Om: (tm— 24) (%m— 22) +++ (%m—Zm-1) Then we have Sn U}Oy + Thag+ 2.42%: Let us form the poe expression P=qst tet (*) Using the formula for an infinitely decreasing geometric progression and assuming that z is chosen so that | 24z |< 1, Jaz [ [nz |<4, expand the sum in an in- finite series in ‘the following way Peay (A aye aa? 4 rie? +...) tog (4 + ee + agate Hag poe) ee Om (1 + Lind + nd? + mas +o +e qos eater ‘ Or PH(pt ot «6. + Om) + (e101 + Lae + 6. + Lm%m) 2+ + (thor + tide +... + tm %m) 2? + Le. P = 89 + 84% + 892" + 532° + Solutions to Sec. 6 337 Put for brevity (1 — a2) (1 — x2)... (L+ 2,2) = Q, Expanding Q in powers of 2, we can write = O42 $+ 92? +... 4Om2", HN t... +2, Oy = Llp + Hyg... + Emmy Multiplying both members of (*) by (1 — 242) (1 — 222)...% x (1 — 22), we have PQ =a, (1 — x92) (1 — 232) 22. (AL — om2) + + Ge (tL — 242) (L — 252)... (1 — m2) + + as (1 — 242) (L — 292) (1 — 292)... (1 — 2 mz) + 2. + + Gm (L — yz) (1 — 22)... (1 — 2-42). Thus, the product PQ is an (m— 1)th-degree polynomial in z. Let us show that it is simply equal to 2™7}, i.e. the following identity takes place PQ@=2"1, Indeed, the expression PQ—z”-! becomes zero at 2= =i. +, Gis a At rt we have (1-2) (1-2)... (1-2)-ge= yer yer-?. Let us show in the same way that PQ — z™-! vanishes atz =4, aries 4. But if a polynomial of degree m—1 a m. vanishes at m different values of the variable, then it is identically equal to zero. Thus, PQ — z™-!= 0. Conse- quently amt =P. 21S 338 Solutions Or gmk 1 1043+ 0922 — 6323 -+ SOF SSA Spa $ Spy" t If we expand the left member in an infinite series in powers of z, then this series will begin only with a term containing =e Oz 2-1, Therefore the coefficients of 2°, 21, ..., 2™-? must also be equal to zero on the right, i.e. we have So = 8 = Sp a2 = 0 Besides, the coefficient at 2™-1 in the left member is equal to 1. Therefore Sines Now our equality takes the following form zm-t T—o42 | ot? — 032) + Reducing both members by 2-1, we find 1 Topp iiaaey.. 2a BL SB Sng H=14- S24 Singer... or 1 (1 — 042 4 092? — oy? $ os. 2”) (LF Sy. + SipiZ? + oe) Arranging the right member in powers of 2 and equating the coefficients of these powers to zero (since the left mem- ber contains only 4), we find Sm— 0, =0, 62 — 08m +S ,y33 = 9, Thus, we get a possibility to compute Smtis S MEY ve ee However, to determine the general structure of s,,4;. let us consider Spe? Vere Solutions to Sec. 6 339 But, on the other hand, 1 : Ba ASne+ Sng? t+ +SingaZ* eo, therefore we get Smith = cake ee tates shstes okt Thus, we get the following final result: s,,4, is equal toa sum of products of k + 1 equal or unequal quantities taken from the totality x4, £2, ..., 2m. In particular Sai = TOE 2 bth ryt + 2yt3+ ... 4- Hp E18m + tty - +. A Lm1Zms Smag = BRIA «2 Ap aire Oem + LpTetg+ ...: 80. Let us introduce the following notation ess eeneenitdnesisio) eeeeeaeeseatatats Sn (is ay +++ Fm) = Gama eiaa em) t a +cat eae tt zh Wa lags - (Zm— #4) (%m—22) +++» (Zm—Tm-1) * ay" Gia) a) em) Wayne (LE 4 as m1 : atone) 1 4y(t +) (t-+) Bytp tm! Ce) ee therefore it is obvious that i (=14 14 4 Son (Hip Lay oes Em) = Sane (q> nm’ sng) 81. The validity of the assertion follows from the iden- tity of Problem 77. The same identity yields At f (zs) ae) (183) En)” A 340 Solutions 82. Set up the expression (A=ay) (2 Ab) an) =) If all the terms are transposed to the left and reduced to a common denominator and then the latter is removed, then the left member becomes a polynomial in A of degree n — t. By virtue of existence of the given system of equations this polynomial vanishes at x different values of 2, namely at A = ay, dz, ..., Gy, Therefore it is identically equal to zero, and, consequently, the original equality (+) is also an identity. But then the equality (+) represents an expan- sion into partial fractions of the following fraction (A641) (A—ba) cy Therefore, the unknowns 2;, 22, ..., a where Ar= (rB+B) (rB-+28) --- (rB-+nB) (8 =B) (rB = 28) «-. [B= (r—1) BI [PB (r-F4) By]... (Bn) * It only remains to simplify this coefficient. 86. We have Chat — Ch = Acy, ie. Cray = Cy + Acy. 342 Solutions and formula 4° holds at n = 1. Assuming that it is true at n, let us prove its validity at n + 1. Indeed Cronus = Chon + AChin = a —1)(n—2 = (otf dert2G59 Arey AAG) Abcy ee + Arex) +A (cn +$Ae,-+2059 Aten +... + Arex ) = =at(4 $1) Aen + (*G5944) Att... pater = ent AE hey EEO" Men FAM, and the proposition is proved. Formula 2° is proved likewise. It is obvious that at n=41 it holds true. Let us assume that it is valid at n. Then we have A" ey = Acpen— 7 Mthgna tS Athenee vet +(—1)" Ac, = (Chant Chin) — > (Chen — Chane + ee Donen 1—Chen-2) + 2-1) (Chr — Cn) = enn "Een EO cheng — ot (— AM en. 87. It is not difficult to check the validity of this for- mula. We see that the right member is an nth-degree poly- nomial in z. Let us designate it by @ (z), i.e. let us put 70) + af (0) +2 SV ay Qt... + z(r—1)... al + fea nF) Ang (0) = G (2). Let in this equality 2 = 0. We get g (0) = f (0), atz =4 we find g (1) == (0) + Af (0) = F (4). Using formula 1* of the preceding problem we may state that in general gtk) =f(k) atk =0, 1, 2,.... 0. Solutions to Sec. 6 343 Thus, two polynomials [q (z) and f (z)] of degree n are equal to each other at n -}- 1 different values of the inde- pendent variable 2. consequently. they are equal identical- ly, and we have ¢ (t) = f (2) for any zx. And so, we have checked the validity of the formulas. It is not difficult to deduce this formula. Let f(z) be an nth-degree polynomial. First of all we assert that it is always possible to choose the coefficients Ao, Ay, An wee A,, such that the following identity takes place * (x) = Ao ~ Aye + Age (x — 1) + Age (x — 1)(x@ — 2) + +... 7 Aye (cw — 1) (@ — 2)... (rx —n +1). Indeed, let us divide the polynomial f (z) by (x — 1) x » (cx — 2)... (x — x). Since the last polynomial is also of degree n. the quotient will be a constant, and the remain- der a polynomial of degree not exceeding n — 1. Dividing this polynomial by z(z — 1)... (ft —n +1), we find the constant A,-; and so on. Let us now compute the constants Ay. Ay. Ag... Ant An: Put for brevity a(e—1) (ex — 2)... @— hk +1) = gy (2) (k = 4, 2,3, ...). Then we have Agr (2) =x (t +1) — @ (@) = ss(e@+ 1)re(e—1)...(@—k+2)— —x(x—1)...(@@—k +4) = sskex (a —1)... (@ —k + 2) = ka (2). To determine Aq, Ay, Az, .... A, proceed in the following way. Put in our identity z = 0. Since @, (0) = 0, we find Ay = f (0). 344 Solutions Let us now take the difference between the members of the identity. We obtain Af (2) = Ay Aqy (2) + Ae Age (@) +... + + Aq Agy (2) = Ay + 2dag (2) Ee. + An Pn-t (2)- Putting here c = 0, we have Ay = Af (0). Further A? (2) = 2AAqy (2) +... + Ay AQy-) (2) = = 24,4 2.2. $2 (2 — 1) AnQn—e (2). Hence _. Af (0) A Seaqcoa Continuing this operation, we find all the coefficients Ag. Ay... 6, Ane 88. Replacing z by z + 1, we have (uf AY = Ag+ Awe +E a (eA) + A 9 (eA) (22) pont $B (r—t) 2. (rn). Putting f (x) = (x + 1)" and using the result of the pre- ceeding problem, we find A, = Af (0). From formula 2° of Problem 86 we get the required expres- sion for A.. 89. Putting & = 0 in formula 2° of Problem 86, we get Meg = en — Fen +2 oe ee t(D)". Put oo= Grae and take 1 = aop CMG ee Solutions to Sec. 6 345 to prove our identity it only remains to prove that A nt 1 ft 1 (e-+ny (@+1)--. @ +n) {Stet --+saa}- Use the method of induction. At n=1 the formula is true. Assuming, as usual, its validity for n, let us prove that it is also valid for n+1, We have am ra! (Mee) = =\(apers cerern (aatept et +epct) }=segn esa (Stat +s} — erate eer (aptaget tape} “Ter. Aapaty {4 r n+4) (+ Fad “tee —*(Satsat ta )}= (n-+4)! =TeFD tran {(F tent +s} At z=1 our identity yields 1 Cue i 4 nt {7 tot tap pe eet o 1" GE 90. The expression 9, (x+y) is an nth-degree polynomial in x, Therefore we may represent it as (see Problem 87) Gn (@+ Y) = Ao Ais (2) + Anthea (2)> +--+ Ann (2), where A,= Aten (since @n(z-+-y) turns into @,(y) at z=0). ee it is known (Problem 87) that Ag, (y)= =Na-1(y), consequently Aon (Y= 2 (2 1) n-2(¥), A’pn (y) =n (n—A) ~.. 28+ 1) nis (Y)- 346 Solutions ‘Thus ; —) (a2)... (2-944 7 Aya BARD 2) 2 PFD Pael) — Cr, oy), and our formula is valid. However, the validity of this formula can he proved using other reasons. Let xz and y be positive integers greater than n, Then the following equalities take place (1-2) = 44 tT (1p2y= tt yet HED 22 4.4 (142) = 44 (ety) e+ feruerent) whe op pyle tya ety 2) oy ‘ 1-2-3 On the other hand, (f+ 2)8(1-2)Y = (1+ "4, 342 sie) at aes Equating the “coefficients of 2" in Ba members of this equality, we get @n (f+ Y) = On (2) + CnGn-1 (2) GY) He CHG (2) Gna (Y) + Gn (Y)- Let yo be a whole positive number exceeding n. Then Gn (t-Yo) aNd Gn (x) + ChPn—i (2) G1 (Yo) + « «+ Pn (Yo) are two nth-degree polynomials in x, and they are equal to each other at all whole values of x exceeding n. Consequent- ly, they equal identically at all values of z. But yy may attain all whole values exceeding n. Consequently, as in the previous case, we conclude that yp can attain any values and the equality Gn (LAY) = Pu (2) + CrGn-1 (4) G1 - +. cnt = eer Gn Ew a oo 1 (y) + Pn (y) is valid for any values of x ce ye Solutions to Sec. 6 347 91. First of all, both identities 1° and 2° can be readily proved using the method of mathematical induction. Indeed, at n = 1 identity 1° takes place. Suppose it takes place for all values of the exponent, not — n, so that we have n n na=3) y= pT Pty 3) prige— AO 9ORD peep g Multiplying both members of this equality by z-+y =p, we get a zy oo poe rat M3) na n(n—4) (n—5) fa GS preg — x Kp P+... Hence ai yl a phat pity p Md) m3) a _ n(n—4) (n—5) aeps ect —a(p— =p rag 4 (— Ho ) p-5g2 oe Bin n—8 aa a = prs — ott ph gt {rea “th pr ig— —A4) (n—5) (n—1) (n—4 oe hd mph PE peng g OAM) prcage nm $4) (n= 3) (n—4) 4 poet... and the theorem holds at n + 1. Proposition 2° can be proved just in the same way. Bear in mind that if x and y are the roots of a quadratic equation, then both formulas represent none other than 348 Solutions the expression of symmetric functions of the roots of this equation in terms of its coefficients. If we put in these formulas z=cosg-+ising, y= =cosp—ising, then a"+y"=2cosnp, p=z+y=2cosp, g=zy=1, enth_yntt _ sin(n +1) @ zy ——ssing i Thus, we obtain an expansion of cosnp and sete in powers of cos@. 92. Put attyt=Sr, cy=q. We have to prove that Smt €mdS mit Crp ig?Sman to. + C3 t2q™ 4S = 1. Assuming the validity of this equality, let us prove that Srna Cine 9S mn + Cig 2P Sma h + «+ CBG" 289+ + Cimg™S y= 4. We may consider that x and y are thé roots of the quadratic equation «?—a+q=0. Hence Srsi==Sk— Spi for any whole k. Consequently Siti = Sm—GSm-1s Sin = Sina — GSn-2s Simi = Smo2— WS m=a» Ss=S.—951, S2=S,— So, Si=8, Let us multiply these equalities in turn by $, Crgigs Cing2?s 22 CEm tig 4, CEng and add them, Solutions to Sec. 6 349 Then we obtain in the left member SmsrtCm4 19m Cine 29° Sine b + «ECT 1g" 1S C3ing"S We only have to prove that the right member is equa) to 1. The right member is equal to Sm Chg 1 GS mas Cm 2G Snap «+ CTm 1g 8S + C3ng*S; — GS t — Cn PS mn — Cm 2PSm-a— += = Con 1q"So. or Sm (Cin + 1) GSmat-+(Cinvt + Ongt) PSm-a+ + (Cinta + Com 2) gS + Cig" 1— —CrpiG’Sm2— «+ —C3m og” *S4— Com 21g" So = = (Sm+CmgSmart Cn 1GSmaat +++ CSmtag™ Sy} + +CIng"S1— Clin -1g"So- But, by hypothesis, the braced expression is equal to 1 and C3mS, — Ci1,So = 0, since S; = 1, and Sy = 2. And so, the right member is equal to 1. Furthermore, it is appa- rent, that at m = 1 our equality is true. Now we can assert that it is valid for any m. 93. If ut-v=4, then uw AA Chv + Cpu? + ee + Chmigv™) tu™ (14 ChutCyw+t ...+ CS iw") = 4. Put Then u-+-v=14. Further 1 1 Tim = (GetCn pert Cs sett + Cima y+ + (Rech thar het + Chet tL). Hence we get our identity. 350 Solutions 94. It is easily seen that we can always choose constants Ay, Az, ..., so that the following identity takes place (md tm + A (A + Pt) + +AU Ae) +... Indeed, (1 + #)" is a polynomial of degree n in t. Divi- ding it by ¢*-+ 4, we obtain a remainder (a polynomial of degree not exceeding n — 1). We divide it by ¢ (é"-? + 4) and so on. It is clear, that the quotients thus obtained will be constants determined uniquely in the process of division. Putting? = + in the identity being formed, we find (cp y)" = 2" $y" Aggy (ey) fp Agty? (Ob YE To determine the coefficients Ay, Az, ... let us put in this identity z=cosp+ising, y =cosg —ising. Then we have (2 cos @)" = 2 cos np + 2A, cos (n — 2) 9 + + 2A,cos(n—4)p+.... Taking advantage of the known formulas for the expan- sion of cosine’s power in terms of the cosine of multiple ares (see Problem 10, 1° and 3°), we find the expressions for Ay, Ag, eee 95. Let y, and yz be the roots of some quadratic equation y+ py+q=0. Let us set up this equation, i.e. find p and gq. For this purpose we multiply the first equation by q, the second by p, the third by unity and add the results. We get 2(Y§ A PYs +9) +0 (Ys + PY2 +9) = 419 + GaP + a3 =0 since Yt PY td =Yit PY2ty=9- Nolutions to Sec. 6 351 We then multiply the second equation by g. the third by p and the fourth by unity: We get ery (Ys + Py + + taye Ya + P29) = aog + aspa, = 0. Thus. for determining p and qg we obtain a linear system aq + agp + a3 = 0, ang + asp + a, -= 0. On finding p and qg, we determine y, and y, from the equation y? + py +4 = 0. Knowing y, and ys, we then determine z,; and z2, say, from the first two equations. The general system is solved in the same way. Namely, suppose Yss Yoav - - ++ Yn are the roots of a certain equation of degree n: n Ye Py = pay? oe PntY Pn = 0. To set up this equation multiply equation (1) by pz, equation (2) by p, -, and so on, and, finally, equation (m + 1) by 1 and add the results. We get Dy + G2Pp-t + +++ + ante =O We then multiply equation (2) by p,, equation (3) by Pn-1 and so on and, finally, equation (nm + 2) by 4 and thus obtain a second linear relationship for determining p,, Pn-s, .+-+ + Continuing this operation, we finally get n linear equations for determining the unknowns py, po, .. .s Pn: If Py, P2. +++» Pn are found, then to determine y,, Yo, «+++ Yn We have to solve the equation yt Py oF Dnsy + Pn == 0. To find 2, xg, ..., Z, it only remains to solve a system of linear equations. Demonstrated below is the original method of solving this system belonging to S. Ramanujan. Consider the follo- wing expression 1 zn sires Easier caree@e rire if 90) = Ro eT TT, * 352 Solutions But Toya + Oy: + Fyi+ Oyi+ ...), Totyy ta (LO + OYE P OVE +. -), Tat = en (1+ Oy + OUR OVA +...) Consequently, D (8) = (rH tet +6 En) + (Bab Bayar oF TnYn) 0+ + (aah oe Prny2) OB ... + (ey hr t t aeyZP 38 ot pany!) 02-14 (cyt +... + amnyD Om... But by virtue of the given equations we get @ (8) =a, +0,0+ 2,02 +... Hag I+... . Reducing the fractions to a common denominator, we find 00) A StneEeE Hence (4+ a8 + 907... + danOP" +...) (1+ B04-B0?+ + +Bn0") = Ay AgO+ -.. + An". Therefore Ai=a, A= 0,-+ By, Ag = 3 -+ a2B,+ a,Bo, 0 = Gay: +a,B, +... + By, O = Qny2 + OnisBi +... + a2By, Since thé quantities a, dz, -.., Gn: Apts, «+ +; Gon are known, the last equations enable us to find first By, By, ..., B, and then. Ay, A2,..., An. Knowing the quantities Solutions to Sec. 6 353 A; and B;, we can construct a rational fraction @ (8) and then expand it into partial fractions. Let, for instance, the following expansion take place © 0) et tt tS Then it is clear that T= Pir Y= 13 T2= Per Yo= 425 Tny=Pnr Yn= n+ The system is solved. 96. Eor the given case we have 240-4302 4203+ 64 © 0) = p50 pO * Expanding this fraction into partial ones, we get the following values for the unknowns poe p=—41, 5 _ 184+ V5 _3+V5 He eoedOne eae ceeaeae? _1%-V5_ 8 VB Ce eee ees eee Ol Vozet Tuinsa oy /g ttc cae ceeanee —8-V5 va _tV5-+4 25” aac 97. 1° We have (m, pw) = — a) (122) aa) La BTS) (Lamy — 2m) (12) (122)... 124) (1—2) (1-22)... (1) : Hence it is clear that (m, p)=(m, m—y). 23-1225 354 Solutions 2° Indeed, 4) Gem) daa) oo ee HY) daa) (m pt d= (1—2) (1—22) ... (1a) (121) ee deem aay (aa B oh) 1—xm (42) (122)... (a) Thus (m, p+) =(m—1, p+) Sa = {ama 1g pm—pat = (m4, py) ee ght =(m—4, wyt) [tame t =(m—14, p41)+a™H-4) (m—4, p). 3° Using the result of 2°, we get a number of equali- ties (m, p+ A) =(m—41, pA) pamel(m ~4, py), (m—4, pt t)=(m—2, +4) amb? (m—2, pW), (H+2, pt 1)=(M+4, pt1)+2(H+4, p), (ut+4, p+4)=(H, py). Adding these equalities termwise, we find (m, p+1)=(bs w) +e (BHA, Wt. pam hot (m—4, je). (4) 4° It is required to prove that (m, ) is a polynomial. We have (m, 1)= =ttotot...f2™. Thus, our proposition is true at 4» = 1 and any m. Assuming that (m, k) is a polynomial at k < yp, by virtue of the for- mula (+), we may assert that (m, p + 1) is also a polyno- mial. And so, our proposition is proved by the method of mathematical induction. 5° Introduce the notation f (x, m) = 1 — (m, 1) + (m, 2) — (m, 3) + 2. + + (—1)" (m, m). T=2 } | | Solutions to Sec. 6 355 First let us prove that f (@ m) = (1 — 2) f(z, m — 2). 1=1, (m, 1)=(m—14, 1) +27, (m, 2)=(m—1, 2)+2"* (m—1, 1), (m, 3) =(m—4, 3)+27-3 (m—4, 2), (m, m—1) =(m—1, m—1)+2(m—1, m—2), (m, m)=(m—1, m—1). We have Multiplying these equalities successively by +4 and adding the results, we get f(z, m)=(1—2™") —(m—4, 1) (1—2™*) + (m—1, 2) x x (t—2™)— (1) (m1, m—2) (1—2). But (1-24) (m—1, 1) =(1—2"-4) (m—2, 1), (1-24) (m—1, 2)=(1— 24) (m—2, 2), ‘Therefore eee ell f(a, m)=(1—2™") {1—(m—2, 1) 4+(m—2, 2)—...4 +(=1)" (m—2, m—2)} = (1—2") f (@, m—2). Thus f(z, m)=(1—2"") f(t, m—2), f (s,m —2)=(1—2""*) f(z, m—4), First let us assume that m is even. We get f(z, m)=(1— 2") (1-24) 4-2)... (1— 2) f(z, 2). But fle )=A—@, 142, 2)=2—42# ate. Consequently, indeed, fe m)=(1—2"™ (1a)... (128) (1-2) if m is even. 356 Solutions If m is odd, we have F(a, m)=(t—2"4) (1-2)... (1-24) f(z, 1). But f(z, 1)=0, consequently f(z, m)=O0 for any odd m. However, the last fact can be readily established imme- diately from the expression for f(x, m) f(z, m)=1—(m, 1) +(m, 2)—(m, 3)+ ...+(—1)"(m, m). 98. 1° Put R(h+1) (4—2n) (4-2)... (dan) 143 eee 2 gk — F(n), Then att RGA) paced (antl) (1-2)... (4—anekta) SEE) Fet)=1+> “aa ey h=1 Hence F(n+1)—F (x)= 7 h(R+4) a ato tt (+4) (42) arheay yg 2 gnth n hik-bAy any (Lh) FO gh =3 Gaadoay em eh Atal) + == 4 (nAyin-+2) pap mens eae Ht a = h(h-1) =a" 3) (f=27) (1am (120) ae ie (fa) a =a) n(n+4) gate 2 2h = a2 (n). And so ae ) : F(n+1)—F (n) =22" F (n), ie. F (v+1)=(1-+4+ 20) F (n). Solutions to Sec. 6 357 Therefore F (n) =(1 +22") F (n—1), F(n—1)=(1 +22") F (n—2), F (3) =(1 +224) F (2), F (2) = (1+ 22%) F (1), F(A)=14 22. Multiplying these equalities, we actually get F (n) = (4 + 22) (1 + 22)... (4 + 2%). 2° is proved similarly. From 1° it also follows that (—a2n) (4—an-ty ... (1 an-t}) (1—2) (22)... ah) is a polynomial in x (see Problem 97). From the same equality we can obtain Newton's binomial formula as well. Indeed Aaanht — Afata®t...fank tok tape y... pat Therefore, at x=1 the last expression attains the value — . Consequently, we may consider that the expres- sion (127) (4 — and (4-2) (1— at z=1 turns into n(n—1) ... (n—k-+4) Ak 4- ok ¢ and formula 1° at x=1 yields han (14+2)"=14 > Che* (Euler). R= 99. Readily obtained from 1° of Problem 98 at z = —1. 100. Put Cot Ci +2) + Cn (8 +2) + Hei +O +2) = oO). 358 Solutions We then have fp tne @n (#22) =n (2) (expressing @, (2) in terms of a product). Making use of @n (2) expressed as a sum, we find with the aid of the last identity Cra?) (1 — 922%) = Coys (1 — 22H) (k=0, 4, 2, ...,”—4). Furthermore, it is obvious that C, == x”*. Putting in the last relation the following values for k in succession: n—1,n—2,...., 0 and multiplying the equalities thus obtained, we find C, = cet) (J—22nt2heay |. (1— rin) zk (1—22) (1—a4) ... (1— 220-2) (k=0, 4, ...,n—1). 101. 1° Put cosx +ising =e. Then cosz — ising =e, Further cos lx + isin lx = e!, cos lx — isin lx = e'. Consequently . el i sin le = 5, (1—e-*), Substituting this value of sin lz into the expression for Up, we find (hg?) (1 gt) 6. (gant) — P aon—my a) =a) a) q u where g = 27%. The required sum is rewritten as follows 1—uy+ug—ust+ ...+Ugn = 1+ an pyr dagay dt +2 ‘) ($9) ( Ur 1 ~ 7 h2n—k) Solutions to Sec. 6 359 Now let us take advantage of formula 1° of Problem 98 4 and, replacing in it n by 2n, put z =q and z = —q"%, We then have an ‘ue L—uy+u,—us+... +e =[[—-g A= hat 2n n =] t—e) = 7] (fe) (ft) = hat hat n n ==] 2 [4 —cos (2 —1) 2] = 2" [] [1 —cos (2—1) 2]. h=t k=t 2° Put (as in Problem 97) (14) (1 gent (9) (14 (2—g2n-h+1y —gk) = (2n, k). Then 1 uy = (2n, yg 2B, where ¢=cos 2z— isin 2z, We have to compute the following sum Qn 2n D (— 1) uk = Dy (— A)" (2n, Regen, h=0 k=0 where (2n, 0)=1. From Problem 98, 1° we have 2n ROA) (192) (1g?) -.. (1 g?"2) = 3 (—1)* 2n, ka 2 gk, Put (1 —gz) (19?) ... (1-92) = n (2, 9). We then have Gn (2, 9)*Pn(—2, 9) = On (9, 2°). Hence R(R+1) 2n s(s+1) an Dy (— 1 On, Hg 2 a. 3 Qn, s)q 2 gt (= 1)" {2n, m}qrmrnz™, 360 Solutions where {2n, m} is obtained from (2n, m) by replacing q by q*. Consider the coefficient of 2?" in both members of this equality. On the right this coefficient is equal to (—A)"{2n, nj gh», In the left member we obtain the following expression R(R+4) a s(s+1) D1 (—1)* (2n, A) (2n, 8) gq? : R}S=2n But (2n, 2n—k) = (2n, k), therefore the last sum is equal to 2 genttn S (—1)* (2n, k)? gh —2n8, k=0 And so, we have 2n gnttn > (—1)* (2n, ky? gh? 2nk — (— 1)" {2n, ny grtn, R=0 But (2n, by = uhg?nt-™, hence 7 2n By (— Dhak = (= 1" Qn, n. Further 4 (Qn, n)=ung?”, (2n, n}=tng-™, where u, is obtained from un by replacing x by 2z. Finally, Qn cee n sin (2n+2)2sin (2n-4+4)z... sin 4nz DS eed at reer ore Cees y or ‘ h=0 We proceeded from an DL (= 1)" (2n, By? ght-2e* = (— 1)" {2n, n} grr. k=0 Likewise we can obtain the following formula anti 3, (Cn, beg ern = 0, 0 he Solutions to Sec. 7 364 If we put g=4, then (n, k) turns into Ck and we get the formulas 2n 2ntt DOCH = (OCS) Chaal?=0. Likewise, if we take advantage of the identity Gn (2 9) -Pn (q"% 4) = Pan (2; 9), we get n >) (n, ky? gk? = (2n, n) ef ) and hence n Dy C= etn (see Problem 72). SOLUTIONS TO SECTION 7 1. We have to prove that Gta bye ~ afb aye However, this equality is equivalent to the following bac (ea) Oe) (a+b) (@Fe) or b—a_ c—b b+ce a+b’ ie. 7 Pat =o? — Be. The last equality follows immediately from the condition of the problem. 2. Ifa, is the nth term and a,, the mth term of the arithme- tic progression, then we have a, =a + a(n — 1), Gm = a +4 (m — 1), where d is the common difference of the progression 362 Solutions Hence Gp, — Om = (n — m) d. By hypothesis, we have the following equalities b—c=(q—7)4, c—a=(r—p)4, a—b=(p—q)4. Multiplying the first of them by a, the second by 6, and the third by c, we get diq—rnatr—p)b+(p—gad= =a(b—c) +b(c—a) +c(a—b) =0, whence (q—Nat(r—p)b+(p—qe =0. 3. We have a, —a, = (p — 49) 4, where d is the common difference of the progression. Since, by hypothesis, a) =, a =p, then a, — a, =q— Pp, therefore qg—p=(p—4, and, consequently, d=~—1 (we assume p — q 0). Further Gp, — ay = (m — p) da, hence Am = ap + (m— p)d=q—m-+ p. 4, We have An+k = U% + pd. Let k in this equality attain successively the valves: 4, 2,3, ..., qg. Add termwise the g obtained equalities. We get Qpt1 1 Appz + ves + Apt = =a +a, +... +a, + pad, Solutions to Sec. 7 363 But Ans, Open +... + aptg = Sptq — Spy Oy + dg+... + aq = Sq, therefore we have Speq = Sp + Sq + pad On the other hand, it is known that aya aya, Sp= S Pp, Sy=—s 0 Hence 2S, Sq Sa = = (p94 or 2 (DS —PSa) Sites Consequently 2(9Sp—PSq) _ (P+H Sp — (P+ 9S, ee Finally + Soq= F=F (Sp—Sq) = —(P +9)» 5. Follows from Problem 4. However, the following method may be applied. We have ata ata, Sp=aztp, Sy=za hence +a, a+aq 2 ?P 2 or [2a, + d(p —1)] p = [2a, + d (q — 1)] q, 2a (p — 9g) + d(p? —p— a + 9) = 0, 2a, + d(p+q—1) =0, Hence a + ap+q = 0, 364 Solutions since Apeg = 4 +4 (p +q—'1). But ay + a, Spg= ay (n+ 9)- Consequently, indeed, Spiq=0 6. We have Sm At 8m im, Sn= From the given condition follows: a+ am m aq+an on? le. Qay+(m—1)d _ m Qayt(n—1hd a Hence 2a, (n—m) +{(m—1) n—(n—1) m}d=0, therefore Am = a)-+(m—1)d=4 + (m—1)d= ane Aig cates and finally 7. It is necessary to prove that at the given m and k (positive integers k > 2) we can find a whole s such that the following equality takes place (2s + 4) + (2s + 38) +... + (2s + 2n — 1) = nt. The left member is equal to (2s + n) n. Therefore it remains to prove that it is possible to find an integer s such that the following equality takes place Qstnynant, s= Reheat) | But n can be either even or odd. In both cases s will be an integer, and our proposition is proved, Solutions to Sec. 7 365 8. Let a, =d. Then a, = a +d (k —1) =d(k — 1), since, by hypothesis, a, = 0. Consequently 2 3 n—1 1 - Saqtgt +E (t+ g+ tac) = TS ett ag 1 n-2 1 = Pd etaa= Dd (14+4)- k=1 h=t k=1 (n—2) a a ae a ant 1 an ae a tad 9. Multiplying both the numerator and denominator of each fraction on the left by the conjugate of the denomina- tor, we get sa Va-V4 Va. Vas— Van 4 Ven= Vann ag— ay an— an 5 (Va—VaitVa—V at ...+-Van—V aa) = _Va-Va =e, since yp — Gy = 03 — Og =... =Gn— Any =. Hence ga Vin=Va nay nt : a (Vent Va) Vent Var” 10. We have a} — a3 = (4,— a) (a, +2) = —d (a, +42), a} — a4 = (a3—4,) (a3+ 4%) = —d (a3+4,), 3p 4 — OB = (Aen-1— Gan) (G2n-1 + Gon) = —d (dan-1+ Gen). Therefore Sd (Oy + Oy $34 04 oe + agp-4 + ayy) = —G AEM 2K, 366 Solutions But Qgp=G,+d(2k—1), a, —agn= —d (2k—1), consequently, S = —d(2k—1) SO 4p S* (ata). 141. 1° We have S(Q+2)—S (0+ 1) = Anse, S (n-43)—S (0) Anes + anya dna Consequently, we only have to prove that Onst + Ange + nys— 3ani2 = 0. But it is possible to prove that a, a, set Hasse {if r and s are of the same parity). Indeed, G; + d,= 2a, 4-(s—1)d+(r—1)d= r+s =2]a,+{———1) d|=2a,,,, [ov (S2—1) d]—2ee therefore nti t Ons = 24n42, and, consequently, ns + On4a+ Onsg— 3dniz = 0- 2° First of all S (Qn) —S (7) = anise + Opn = BE on, Now we have S (Bn) = ay + dye + Ont (dns + +++ + an) Fansite + Ff yy = EEE nt (ay + anes) + (nat danaa) + +» + (ar +a32)- But since the sum of two terms of an arithmetic progres- sion equidistant from its ends is a constant, we have Qu bangs = Gna t Gongs = + =O + Gan = Ongip an: Solutions to Sec. 7 367 Therefore S (Gn) = EH nn (ayy fgg) on = 3 faethe py =3(S (2n)—S (n)). 12. According to our notation we have Sh=r-tyngi t+ Qa-wnggt +++ + dans : Sit = Gantt Ghnaat ++ + Gasnn Consider the difference Sasi — Spe We have Sirti Sh = [danen — Aan] + +++ + (thnga — Ga-pnael + + [aansi~ Qantas But since Am — a = (m—l) d, we have Shit —Sypand+ ...4+-nd+nd=n'd. 13. We have b—a=d(q—p), c—b=d(r—q), c—a=d(r—p); on the other hand, a=uo?, b=uot, c=uo", where u, is the first term of the geometric progression, and © is its ratio. Therefore alm. 8-4. 02-9 — ghar , pder-P», pdcP-@ = Usa-ryeder-Py+d(P—a) QUA) PDH P-PVA- D4 (P-DT-D) | But it is easily seen that d(g—r)+d(r—p)+d(p—g)= (a—r)(p—1) + (r—p) (G1) + (p—9) (r—1) = 9. And so aP~*.bF-4 68? = 1, 14. We have Ltet+e4+...fa%= onto zt * 368. Solutions Consequently ia (tote... pata (Et y-e= _ (entt_4y2—an(e@— 1)? z2mt2_gentt p—gnt242ent1_—gn a (1% ~ (1)? =e Pate ttt aise aber) Ok x (t-o-ot$... +2). 15, Let the considered geometric progression be Uy, Ua, ey Way Ungty ++ ey Urry Urns «ees Wane Hence Ssn— Son =Uantit +++ Wan, San—Sn = Unsit +++ + one But u=ug', us =uygs. Therefore Ups Use, Uansn = Mag”, consequently, San— San = Uanstt ve. + lan = GP" (ir $ Yat «sf Un) = P"Sny Son — Sy = Uns +++ Man =Q" (Urt let... tun) =Q"Sn- Therefore Sn (Ssn—Son) =G"Siy (San —Sn)?=9""Si, and the problem is solved. 16. Using the formula for the sum of terms of the geometric progression, we get and —% ro et . sash, ga Consequently But, on the other hand, P? = (ayy... Qn) = (Asan), Solutions to Sec. 7 369 hence n 2 P=(3)- 17. Let us consider Lagrange’s identity mentioned in Sec. 1 (see Problem 5) (EP ai btn) EME et = (AY Daye + «6 + TnsYn—1)? = (TY2— Toys)? + + (2ays— sys)? + «+ «+ (En-2Yn1— Yn-2F ns)? Put Ty= G4, Te G2, +--+, Fn = An-13 Y= 42, Yo Ag, +++, Yn = An. We then have (tai... baka (a-talt ...+at)— — (as@y + apg... + nsdn)? = (0103 — a3)? + + (aya, — 302)? +... + (Gn-2@n— aR-1)?. (#) The bracketed expressions on the right have the following structure GAs — Oye, and k+s=k'+s’. It is evident that if a, a, ..., an form a geometric progression, then (provided k+s=k'+- +8) Ayds — Ay Oy = Indeed a =ag'*, a= ayy’, Ay = age -', dy = aygr'. Therefore aya, = alg** and toh he Apydy = aig* +s'—2, Ops = aps". Thus, if a,, a2, ..., a, form a geometric progression, then all the bracketed expressions in the right member of the equality («*) are equal to zero, and the following rela- 21225 370 Solutions tion takes place (ab tat+ baka) (taht ...+ah) = = (4102 + 203+ '..- + an-sdn)*. Now let us assume that this relation takes place. It is required to prove thaf the numbers a, a2, ..., a, form a geometric progression. In this case all the bracketed expres- sions in the right member of the equality (*) are equal to zero. But among these expressions there is the following one (ajay — Gran)? (k = 3, 4, ..., n)e Therefore we have Sh oe = eae (k=3, 4, ..., n), i.e. the numbers a, a, ...,@, really form a geometric progression. 18. 1° It is known that Soy = 2094 Let us make up the required sum. We have Sit Sot... 4+Sn= on pda andy —1 qi q—i _lectert stan)q_ aun (Gng—ay) gan q-1 ~q=t ate ai 9° 4 4 aaah tae 3° 4 +. == we eat an -4) — gk at f 19. Let the given progression be a1, a2, ..., dy. Let a, designate the kth term from the end of the progression. Then a; =a, —(k—4)d, a, =a, + (k—1)d. Solutions to Sec. 7 371 Consider the product a,a;. We have ana, = aya, — (k — 1)? d*? + (k — 1) d (a, — a) = =a, — (k — 1)? d* + (k — 1) (n—1) And so aya; = aya, + ? {(k — 1) (n — 1) — (& — 1)*y. It only remains to prove that the expression P, = (k — 1) (n — 1) — (k — 1)? . s * n n+1 increases with an increase in n from 4 to zr yz: 2 We have Py =(k-1)(n—h), Parr = k(n —k — 14). Hence : Pry, — Py = n — 2k. Consequently, P+; > P, ifn — 2k>O, ie. ifk< > and our proposition is proved. 20. Let a, @2, ..., @ be an arithmetic progression, and Ww, Ua, ..-, U, a geometric progression. By hypothesis, a; = WY, a, = Uy. Let the ratio of the progression be equal to q. Then Un = Ugg”! = an. Put Atat...+a=s, wut... +uy,=9,. Prove that Sp SS Op. We have n-1 pat Sa tty EF yg AE —ung— m4 nt n= ot: Since, by hypothesis, a,>0, it only remains to prove that gut t+ grt q—t _—p nn. <3 372 Solutions Let us write the left member of the supposed inequality in the following way Patt et. tghtt ght ghia HF YEH PAE EEE et +@ +0). Let us prove that htgh tit. Indeed gh gh ht —1— ght = (gt —1) +g" (1—g') = = (¢—1) (1-9) <0, since if g> 4, then gt — 41 >0, 1 — qr*-1 <0, and if q<1, then g —1<0,1~—qr*?>0. At gq =1 it is clear that the product contained in the left member of our inequality is equal to zero. And so, indeed, gitar ht ie. On Say which solves the problem. 24. Let the first common term of the progressions be a, and the second b. Then the nth term of the arithmetic pro- gression will be equal to a+ (b — a) (n — 4), and the corresponding term of the geometric progression has the form a a(z)"- And so, we have to prove that a+(b—a)(n—1) 1, 2 ) ayiss— > Onur. kat k= Transform the right member as follows nt nti Di Gna — >) axa — Quy + dn ytinss = =2 k=2 n+l o> a caus (44 — Gn) Ur — Q1ls + nln = nth 5 =D) dunt anyttnss— atts, k=2 where d is the common difference of the arithmetic progres- sion. Thus nt s(qg—t)= -a>y Unt anstinsi— ly, 8 (9-1) = Gnsnys— ay —d es 2 Finally Qnsilns1— O44 Un ast so EEE Et g@ Se q~1 (q— 1) 24. The required sum can be rewritten in the following way 1 4 1 PHatp $a Spae ttn. Summing each of the geometric progressions separately and joining the partial sums thus cbtained, we have 1\2 1\% 41 \2 (e+5) 4+(@+a)4+.-4 (+4) = (ent? +4) (z2n—1) (fan +2n. Solutions to Sec. 7 375 25. The sum S, is readily computed by the formula for an arithmetic progression. Let us now compute S2. Consider the following identity (a + 1)8 — 2 = 327 + 82 +1. Putting here in succession x = 1, 2, 3, ..., m and sum- ming up the obtained equalities termwise, we have n n n a Dd (+1 2=3 Di 43D etn. sexed wat x=t wai Or {B+ 34... P+ (nt 13} {P+ 3+... += =35,43S;, +n. And so 38,+35,+n=(n-+1)?—1. But 5,=20t0 ‘ Now we find easily Spa Tt) Gn +t) ou aa Ona The formula for S, is deduced in a similar way. We only have to consider the identity (z + 1) — ct = 42° + 62? + 4e +4 and make use of the expressions for S; and S, found before. 26. We have identically (2AM ah (b+ 4) oh EE OE ght 4 EEOREND eae kb) etd. Putting here successively z = 4, 2, 3,..., m, and sum- ming up, we get the required formula. 376 Solutions 27. Consider the following square table: 4F | 2*) 3" | 48 nk ge 2k] 3h Lak nk 4h 2k BR LAR nk Aeon Oe aden a sum of terms of each line is equal to 1* + 2* + + n* = S, (n). Thus, the sum of all the terms of it table will be nS}, (n). On the other hand, summing along the broken lines, we get the following expression for the sum of all the terms of the table AA (1842.24) + (18 + 2% 43.3") + (1 4 284 344.4%) + Bee (Uh 2k 4384 2.4 (n— 1)" tren’) = = 14 [Sp (1) +2") + [Se (2) + 3°] + [Sy (8) + 4+. + +1Sn(n— 1) +n] = = Sp (1) +S2 (2) +... + Sa (21) + EAM Ot ght weeny, And so Sp (1) = Snss() +Sp (N~1) +84 (2 —2)+ ... +Sp (2) +S2(1)- 28. Both 1° and 2° are readily obtained from the formu- la of Problem 26. Let us rewrite it as k k(k—14 Sa= —F Sp Spa 0 8 ay 7 (n+Ayke—4 + Sepia (+) Atk=4S;=14243+...4n="2"=5ntg dn. Thus, both propositions (4° and 2°) are valid at k = 1. Suppose they hold true for any value of the subscript less than k and let us prove that they are also valid at the sub- script equal to k. Since, by supposition, S,-; is a polynomial in n of degree k, S,-. a polynomial of degree k — 1, and so on, it is easily seen from the formula (#) that S, is indeed Solutions to Sec. 7 377 a polynomial of degree k + 4. Further, since S,_4, Sy_2, ..-5 So do not contain the term independent of n, it follows er that S, also does not contain such a term oe. when expanded in powers of n, will not contain a constant term). As is evident from the same formula («), the coef- ficient of the term of the highest power in the expansion of : .. 4 . S, in powers of n will be Fe1: It only remains to prove that the coefficient of the second term, i.e. B, is equal to + In the expansion (+) there exist only two terms contai- ning n*. One of them is contained in — 48,4, and the A+ other in ere From what has been proved we have eerie aed z= — | {En + paz ents Further (m+tyet—t Ake ak eqreepseqeaigeee Repeat tnt. Hence, it is obvious that Bat. As to the structure of the rest of “che coefficients (C, ae Ay we may assert the following: the coefficient of nett will be equal to cay 4 Chi hq , where A is independent of k. This proposition is proved using the method of induction with the aid of the formula (+). 29. S, can be computed using, for instance, the formula from Problem 26. However, we may also proceed in the following way. From the result of the previous problem it follows that S.= dns tony Cr+ Dn? + En. It only remains to determine C, D and E£. Since the last equality is an identity, it is valid for all values of n. Put- ting here in succession n = 1, 2, and 3, we get a system of equations in three unknowns C, D and E. Namely, we have 378 Solutions 3 C+DAE =i, 8C-4+-4D+42E =" ,274.049D43E=8. Hence 4 4 C=4, D=0, E=-. It only remains to factor the expression and the required result will be found. The remaining three formulas are obtained similarly. 30. The validity of the identities is established by a direct check, using the expressions for S, obtained before. 31. Put k = 1. We have (B +4) — BY = 2, B, + 2B, +1 —B, = 2, Consequently, By = 4. Then. put k = 2. We get (B + 1)? — BS = 3, or i.e. By+3B,+3B,+4—B,=3, B,=+. Proceeding in the same way, we get the following table ie. 4 3617 B=y, B=, Bu=0, By=— ap el 694 Br=e, Br=0, Ba=—gay, Br =0; 1 B 43867 By=0, Be=—y, Bo=0, = TR 4 7 = Be=—3, Bo=0, Bu= Gy» By =0. 5 By=0, Bo=gq, B=0, Knowing this table, we may easily solve Problem 29, i.e. arrange S,, Ss, S, and S, according to powers of n. These numbers play quite an important role in many fields of mathematics and possess a number of interesting properties. They are called Bernoulli's numbers (J. Bernoulli, Ars Conjectandi). We can show that for odd k's exceeding unity B,, will be equal to zero. And Bernoulli’s numbers with an Solutions to Sec. 7 379 even subscript will increase rather fast. Let us consider the value of Bigg. If we put Bis = — 2 , then it turns out that N =.171390, Z=62753 13541 04611 93672 55310 66998 93743 60315 30541 53311 89530 55906 39107 01782 46402 41378 48048 46255 54578 57644 21158 35788 96086 55345 32214 56098 29255 49798 68376 27052 31316 61171 66687 49347 22145 80056 7A247 06735 79434 16524 98443 87718 34415 Thus, the numerator of this number contains 215 digits (D. H. Lehmer, 1935). Let us now prove relationship 2°. On the basis of the results obtained in Problem 28 we may put (+4) (184 2*-+ 3% 4... 4+n¥) = anh LEEPER Cat Dnt... +L, where C, D, ..., L are independent of n, but undoubtedly depend on k. Put (e+ 1) (AP 2" 4384... tn') =n! 40h yn® ACRyroen tL... ECR en an! We may then write the following symbolic equality (e+ 1) (182% 4... +n") =(nt+a)*t—ah, On removing the brackets in the right member by replacing a by a, (s=0, 1, 2, ...), we pass over from the symbolic equality to an ordinary one. Since this equality is an identity with respect to n, we may put in it nm +1 instead of m and obtain (FAA) E+ OR + + (nt A) = (n+ 1 toyed oh Subtracting from the last equality the preceding one, we find (41) +1 = (n+ 1a)h— (nto). Putting here n=0, we have (a+ 4)ht—oh =k. R Charagn. 380 Solutions Besides, it should be remembered (see the solution of Prob- lem 28) that a's are independent of k and that a, = + 7 And so, the numbers a, and B, are determined by one and the same relation, and a, = By. Therefore a, = By for any k. 32. Let d be the common difference of our progression, Then m= a +d(k—1). From the first equality we have mat) _ atin n=, na,+d (+) On the other hand, ah =a? 2ad (k—1)4 @ (k~1). Therefore, from the second relation we get >, ah = nazi + 2ad 3 (K-14 @ 2 (k—4)? =B?. Ke Hence nai+ 20d 20) 4 geen On~N _ yp (1) (see Problem 25). Squaring both members of the equality (+) and dividing by n, we find nzi+2ad nat) print a ded es Subtracting (2) from 7 we get Pn (n2—1) _ wna? 12 n (2) Consequently 2V3 0% s+. nV Substituting d into the equality (+), we find 2, and, con- sequently, we can construct the whole arithmetic progres- sion. Solutions to Sec. 7 381 n 33. 1° Put s= >) Pz", Hence a-s= > kz", = =A Subtracting the first equality from the second, we find nti s(a—N= > (kta yatta See ae Consequently a n s(e—1) = DS) (k= 1)22"1 4 2a" — 5) ek, Ret = a 0 8(2—1) =n?2"— S) (2k—1) 2h = nz" —2 D) kat + R=h kat D at arta" —2 aoyetna™ "(nt 4) appt St kat (see Problem 22). Finally 14424924... niet ntzn (x—1)2—2nzn (x—1) + (ar—4) (2 +4) see eee eee 2° Proceed as in the previous case. Put s= 19423243924... 4+n92™ 1 = => Bek, Make up the difference n n sa—s=n3z"—3 DS Pat143 kot — Rt REA n=t Substituting the expressions obtained before for the sums on the right, we have n2zn (2—1)2—2nzn (2—1) + (2n—1) (e+4) s(z—1) =n3x"—3 Gao + nzrtt—(nfA)enf+4 amd +38—— =p z—1 382 Solutions Finally s(z—1)* = 32” (a — 1)? —3n22” (x — 1)? + + 3nz" (x? —1)—(x"—~1) (a? + 42+ 1). 34. To determine the required sums first compute the following sum 143245274 ...4(2n— 1) 2" = >) (2k—-1) a= kat n n = eg 4 __ 2nxt (x—1) (2 +1) (e"—1) =2>) kat — 5) gad ae eine), For computing the first of the sums put in the deduced formula ane ae then have 5 2n— 4 145+EtE at = ger (8 (2"~1)—2n}. And putting z= ae we find 2 3 5 7 jp 2n—1 2-4 (—1)n+1 (6) 1 1~$4$—-T4 4 (oy et Be 35. 1° First assume that n is even. Put x= 2m. Then 1-243-—4+4...4(-1)""= —243-—4+4...4+(2m—1)—2m=(14+34...4 +2m—1)—(2444...+2m)=—m=- 4. Now let be odd and put n-=2m—1. Then our sum takes the form (1—243—44 ...—(2m—2)]+ (2m—1)= = —(m—1) + 2m—1-= matt Thus, if we put 1-243~—44...4(=1)"'n=S then S=—+ if nis even, S=2t* if n is odd. However, this result can be obtained in a simpler way. Indeed, if n is even, we have S = {t—.2]+ [3--4] + [5—6] 4- ... + [(2m—1)—2m] = =—1-m=—m=—+. Hence we also get the result for odd n. Solutions to Sec. 7 383 2° First assume that n is even and put n=2m. We have 2294 3®— 4 (1) nt = (2-2) + + (8-4) +... + [(2m — 1)? — (2m)?] = — (142) — —(84+4)—...—(Qm—14 2m) = —[14243444...4 42m —14.2m) = — Onto nes 1) Thus, if is even, then P2432. ¢ (Aint 2D If n=2m-+1 is odd, then 492874 38.4 (Ay nt 22741 32 (2m) + (2m 1)? = BEF) (om 4 1) = 2 n(n—4) _ n(n+4) 2 Seeceeraseaeceect 3° The required sum is equal to —8n?. The result is obtained as in the previous case. 4° Rewrite the required sum as =n’ n n n 4) (8n2 + TIn+ 2) S@P+M=F BLD year lat ert n+ h=1 k=1 nat (see Problem 25). 36. The considered sum may be rewritten as 40—1 , 10?—1 , 103-14 107—1 spo tg tpt to wherefrom we easily find its value = (10 tn}. 37. Consider the first bracketed expression on the right and rewrite it in the following way 2x2" 2 gAye 4 Day Day? — gM g2nt2 4 y2nt2 apy? — 21, =2r 384 Solutions The second bracketed expression arises from the first one as a result of permutation of the letters x and y, therefore 2n+21 y2n+2 it is equal to yy yt, Squaring both obtained expressions and adding the results, we easily prove the validity of the identity. 38. The required product is equal to (t-a+4-a?4 ...44-a"3)-+ (aa? 4... + aa") + +(@P+...+@-a™)4 ...4 0a"? art= =a(l+a+...ta")+a@(4ta+...+a™%)+ +a (4ta+...fa"4)+...40" 5 (14a)+ar3= ant =a a—4 ot ges a Sota’ ar- ae At" fatty oe. 1 gin-yy _ (a"'—4) (am —a) +e) = GT * 39. The sum on the left may be rewritten as follows (phate te ESE) pay 22 tne) al fn. The first bracketed expression is equal to z{(n—~1) o@—nan 144] 2% (@—1)2 A letQe+... +(n—1) 24] = (see Problem 22). The second bracketed expression is obtained from the first one by replacing x by+. Hence, we get the required result, Solutions to Sec. 7 385 40. 1° We have TMm+!) na nti" Adding the right and left members, we get the required result. 2° The required sum may be rewritten in the following way n 4 s= D FeRD ETS * k=t 1 44 4 ceed But k(RF) (+2) 2°k kT mo ko Therefore 1 1 1) 4/4, 4 4 4 sag(tt gto ty +33 rat... tapiteps) ea i 1 4 4 —(gtgte ty tag) aa(tt+e)+ 1,4 1 14f_4 1 1 1 +(gtate te) te (sar tags) 2a 44 1 et ey —(g+gt-- te) =a+aa ta 3° Solved as the preceding problem. 41. The sum is equal to n is S= > 4h2—1 * b= 25-1225, 386 Solutions Hence ott 2k4-1)—(2k~1 16S = 5 tote 161. foeatet a -3 (4 +1) +4 3 oe hat hat 18 Oe ee +E £3 (ate wat): 2n (n+) (2n+1) 14 19g— ROP Her Dints (greg tet t "Rat — mt Sf? 2n (n-+4) (2n-+1) n 165 = SOE tnt sare Finally i +2) 4 fos 22 where m= 2n+14. 42. We have 4 1 M%tan__ 1 4 ede wig ate tee ata (ae tw) fare ret rere Ai (ed 4 tytn dt Bai Bhi ae eee (a te) Ay + An = Ag+ Any = 03+ Ang=...- Therefore, adding our equalities termwise, we find 4 4 4 2 4 1 1 tent aaa tae akan (ay tag t tae): 43. 4° It is obvious that the following identity takes place 4 eed n+k—1 FRADE OEE FHT * Putting k=1,2,..., p+4 and adding the obtained equa- lities termwise, we prove that n | nt n+p 4 TIT tp On ht Gp EAT HW EPPO Solutions to Sec. 7 387 2° We have n n n n way tarat + agai < eat n+1 n+p ae aa 4 tant tpt Wa ee (see 1°). Therefore 1 1 1 14 1 wait wat twee =—tan>, a a m 2cot 5 —cot p= —tan Tt a a 2cotz— = cot sar = — tan ma 1 4 Multiplying these equalities in turn by 1,4 DP Rr and adding termwise, we get the required result. 51. Consider the following formula cos [a + (k — 2) h] — cos [a + kh] = = 2sinh sin [a + (k — 1) hl). Putting k = 1, 2, 3,..., n —1, n, we find 2 sin h sin a = cos (a — h) — cos (a +h), 2 sinh sin (a + h) = cos a — cos (a + 2h), 2 sin h sin (a + 2h) = cos (a +h) — cos (a + 3h), Qsinhsinlat+(n—2A) = E = cos [a + (n — 3) h] — cos [a + (n — 1) Al, 2sinhsinfa+(n—1)h]) = = cos [a + (n — 2) h] — cos [a + nhl. 392 Solutions Adding these equalities term by term, we find 2sinh{sina+sin(a+h)+sin (a4 2h)+ ...+sin[a+ +(n— 1) h]} = cos a+cos (a —h) —cos (a+ nh) —cos [a+ (n- — 1) h] = {cos a—cos [a+ (n—1) h]}+ + {cos (a—h)--cos (a+ nh)} == =2sin 24 hsin(a+75* hk) +2sin (a+ ea a) x X sin nt h=2sin(at z h)-2sin 2 cos 4 Hence sina+sin(a-+h) + sin (a+ 2h)+ ...+sin[a+(n—1) h]= sin (a+ nt 1) sine eee ae oaCaE TE CC ETEE sin The second formula is obtained similarly. However, it can also be readily obtained from the above deduced formula by replacing a by = —a. ala 52. Putting in the previous formulas a=0, h = we get S=cotz, S'=0. in 53. Taking advantage of the results of Problem 54, we have . . . sin na sin na sina+sin3a+...4 sin [(2n— 1) a] = , sin na cos no cosa -+-cos 3%+-...+ cos [(2n—1) a] = sin a The rest is obvious. 54. The required sums can be computed, for instance, in the following way. Make up the sums S;, and S%. It is easi- ly seen that Sh+Sn=2n. On the other hand, S;, — Si, = cos 2r+ cos 4z+...+cos 4nz, Solutions to Sec. 7 393 Using the second formula from Problem 51, we find in 2nz cos (2n-+1) z sing Ee cos 2x-+ cos 4+ ...+cos4nz = And so + gw __ sin Bn cos (2n-+1) Sn—Sn= sing Sit Sn=2n. Hence sin 2nz cos (2n-+1) 2 2sinz Syant » sin 2nz cos (2n-}-1) 2 Sa=8———Saing : 55. Let us make use of the formula sin A sin B= [cos (A—B)—cos (A+ B)]. We then have P ~ mai. natt S= > sin sin = ist t (m+n) ni edie AG But if mtn is divisible by 2(p+4), then cos mE ==1 and 4 (m—n)ni 4 S=z os —zP. i Using formula 2° from Problem 51, we easily find P ; > cos OM _ a, pri iat Hence All the remaining cases are proved analogously. 394 Solutions 56. We have arctan (+ 1) x+ arctan (—kz) = ke e—ke z T= (e412 (—ka) THkkpha since (k+ 1)x(—kz) <1 (see Problem 25, Sec. 3). Hence =aretan = arctan z arctan 2z— arctan x= arctan Tyrie z arctan 3x2 — arctan 2x=arctan Teri arctan (n ++ 1) z —arctan nz = arctan Taps : Adding these equalities termwise, we find that the required sum is equal to nz arctan (n + 1) —arctan x = arctan Thespa : 57. It is obvious that Gp ~ Gp. T+ andy, arctan a, + arctan (—ap_y arctan r = arctan ————_—_.. 1 ann Now we find easily that our sum is equal to as Oni — 1 arctan Poem: 58. Put 144k —2y, cty=2k. (This is done to use the formula ty 4—zy arctan ssarctanz-+arctany if zy < 1.) Then arctan EF =aretan (H?-LA +1) — arctan (k* — k + 1), Solutions to Sec. 7 395 therefore n > arctan hat Eu = arctan 3— arctan 1-+ arctan 7 — —arctan3+ ...+ arctan (n*-+n-+4)—arctan(n?—n+1)= =aretan (n?+n+1)—4. 59. Let & be one of the numbers 1, 2,...,2—4. Multi- ply the first equation by sin ka, the second by sin nae, the third by sin k2© and, finally, the last one by sing etx eerncae: find . Adding the obtained products termwise, we At, + Anta +o. + Ancitnen = ay Sin hE fag sink 4,4 iE fay«sink 2s , And Ay=sin 1 sin k 2+ sin 12% sin k 2% 4 sin sink 4 +...+sin pes sing C92 Taking ‘advantage of formula 2° of Problem 54, let. us prove that A,=0 if lAék, A=5 if I=k. Hence .. —t ty=2 (a,sink 2+ apcosk B+... +an4sink 2") (k=1, 2,3, ...,2—1). 396 Solutions SOLUTIONS TO SECTION 8 4. We have deed 1 1 1 1 4 4 th Be? atm ERP WTO Adding these inequalities termwise, we find 1 1 1 t 1 access ated wet tape te to > tat ta maT 2. It is obvious that ee eee eee (nk 1) (n+ ky > (nF)? > (np k—1) (n+) * But 1 ae) 1 (nk+1) (ak) n+k” n+k+1’? 4 1 4 (n+k—1)(n+h) nPk—-1 nk’ therefore 4 4 4 4 4 ntk n+k+i < (nh? < n+k—-1 n+k* Summing these inequalities (from k=1 to k=p), we get the required relation. 3. Let us have n fractions (n> 4) Mec mipctiaer tegen 4 4 eee Let us assume 2acbk41. Consequently b>a+i1, cS>a}2, d>a+3, ..., loatn—. Therefore i. 4 1 1 4 oe ptet teat epipt Solutions to Sec. 8 397 Hence 1 4 4 n Palisa ate ere (a1) (@-fn—f * But a—1>1, atn—ASn41, (@—1) (@+n—1)Sn4+1 and 1 4 4 ptt ta Sag n" and Wri >V a. 5. Since ae a0. Hence (VA+2) (V4—2) 7. SS VA, A-a@ 7 /7= A-a@ Sar a+g55- Let us now prove the second inequality. For any x there exists the following inequality 2(l—a)=r—arct, 398 Solutions Indeed, we have e-e—ta - (2-4) <0. It is obvious that we have an equality only at z = + Since itis possible to assume that A —a 7 , we have > =z 1 1-4-0) Vt A— -VW4-) > 0, we find (2a+1)(VA—a)—(A~a) <4. Whence finally 4 VA eV nFi—2V a, since — ye 1 4 -i-Va=———~ < 1 .. Vat -Va- Daa 2/V2-2, —>2V3—-2/2, vi >2V3—-2V = >2V4-2V3, a >2V4-2y qe eV Fi—2Vin. Adding these inequalities, we obtain the required result. Solutions to Sec. 8 399 12 4 2s—2 4A>e7' eGo? 1.3 5 2s—1 oie ecm Multiplying these deat we find A = >> Ie 8. Since and tan @ = oa 1—tan? > we have ty cot? 5 cot? > cot 0 = —_;—— = a 2 aaa 2eot > 2 Consequently 8 cog! 8 41+ cot 8—cot >= 14+——— - ety = 2 2eot > = zea a {cot?}—2cot $44} = — 400 Solutions since cot $00 <00, since C is an obtuse angle. And so tan A} tan B 4—tan AtanB >. But since’A and B are less than %, it follows that 2 tan A+ tan B>0, and hence 1—tanAtanB>0, tanAtanB<1. 10. Indeed tanO—tanp _ (n—1) tang tan (0—9)=TTtandtang ~ tEnton2@ Therefore tan? (0—@ (n—1)? (n—1)2 (nt (cote yntan 2 (cotg—ntange+4n = In 141. We have 1—tan? y T+ tan?y * To prove that cos 2y <0, it is sufficient to prove that 1— tan? y <0. cos 2y = But we have {—tan® y= cos? a ct eee asin BP We only have to prove that cos* a cos? B — (4 + sin a sin B)? <0. But cos? % cos? B — (4 + sina sin B)? = = (1 —sin? a) (1 — sin? B) — (4 + sina sin B)? = = — (sina + sin f)? <0. Solutions to Sec. & 404 12. Let m be the least and M the greatest of the given fractions. Then m y, x > z. Therefore, if 4—2>0, 26-1225 402 Solutions then « wh? yh® SO and g-? — 2-2 >0, and, consequently, for 4 > 2, ak —y+—2>0, ie. zk > yA + 2 We prove in the same way that cyto if W<2. 16. (See Problem 7, Sec. 1). It can be proved, for instance, in the following manner. If a? + b? =1, then, obviously, wercan find an angle @ such that a=cosg, b =sing. Likewise we can find an angle g’ such that m=cosqg’, n=sing’. Then we have | am + bn| = | cos pcos g’ + sin psin g’| = = | eos (p — g')| <1. 17. We have at > d* — (b — 0), b? > b? — (ec — a), * ce >ct — (a — b)*. Multiplying, we get abc? > (a + b —'c)? (a +c — by? (b +c — a)® Hence follows the required inequality. 18. It is known that if A + B-+C =n, then A B A c B c tan> tan>+tan> tan>-+ tan> tanz=1 (see Problem 40, 4°, Sec. 2). Put A B Cc tany=2, tanz = Ys. tan It only remains to prove that P+yp+e>i Solutions to Sec. 8 403 if zy + az + y2 = 1. But we have 2 (2 + y? + 2°) — 2 (zy + a2 + yz) = =(c—yP + (@— 2)? + (y—aP Se. Hence 2@+y4+7)—-2<0 e+yp+eSt. 19. We have SHEA —) (P=) iB ray 2) | sing V EE, sing sn § EE De Consequently, it is sufficient to prove that (p—a) (p—b) (p—c) — 1 ae SB But ae athte ga btena p-—c= Therefore, we have to prove only the following (bene) (ape=H(etb—9 <4, = abe provided b-+-c—¢>0, a+-c—b>0 and a{b—c>0 (see * Problem 17). This inequality can be proved in a different way. Put sin 4 sin 2 sin & =f; nz sin > sin->=6; then we have 1 A—B ALB ALB B= x (cos =F“ —cos > ) cos 5 Hence cost 448 —cosAZ cos At 26* 404 Solutions Consequently ace Ace +f cost Ao ag cos 2 ad z Since cos a and cos 4? are real, there must be cos ATB —8§>0, A-B 1 8Eab-+cd+2V abed, i.e. that aed ch +- ad >2 Vcbad. But act _ an cb-ad—2Y chad = (Y cb—YV ad)’ >0. 22. We have a® +-b®— 2ab = (a—b)?>0. Hence @—ab-+b'>ab, a? 4 bab (a+). Consequently 3a3 + 3b? > 3a*b + 3ab*. Add a3+0? to both members of the last inequality. Solutions to Sec. 8 405 We have 4a} + 4b? >(a +b). And so, indeed, Seo (eH). 23. 1° It is required to prove that the arithmetic mean of two positive numbers is not less than their geometric mean. Indeed, ath Vab=4(a4+b—2V a) = 4 (Va—Vo)'S0. 2° To prove that o

You might also like