PROBLEM OF THE WEEK

Solution of Problem No. 1 (Fall 2000 Series)

Problem: Given that cos 36 = 1
4 + 1
4 5, show that (tan2 18 )(tan2 54 ) is rational.

Solution (by Mike Hamburg, 11th Grade, St. Joseph H.S., South Bend)

Since cos 36 = 14 ( 5 + 1), cos 72 = 2 cos2 36 1 = 18 ( 5 + 1)2 1 = 1
4
( 5 1).
Then !2
2 2 cos 36 cos 72
(sin 54 )(sin 18 )
(tan2 54 )(tan2 18 ) = 2 2
2
= cos 36 +cos 72
(cos 54 )(cos 18 ) 2
!2  2
5+1
51
1 1
= 4

4
= = rational.
5+1
+ 51 5 5
4 4

Also solved by:

Undergraduates: Heung-Keung Chi (Sr. EE), Haldun Kufluoglu (Sr. EE), James Lee (Sr.
MA/CS), Robert Manning (Fr Eng), Maxine Mbabele (Jr. EE), Jeffrey D. Mosov (Fr.
MA/CS), Stevie Schrauder (Jr. CS), Yee-Ching Yeow (Jr. Math)

Graduates: Srinivas R. Avasarala (CS), Ali Israr (ME), Chen Kai (MA), Wook Kim (MA),
Sravanthi Konduri (CE), Gorindarajao Kothandaraman (AAE), Chris Lomont (MA),
B. N. Reddy Vanga (Nucl E)

Faculty & Staff: Steven Landy (Phys. at IUPUI), William Wolber Jr. (PUCC)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., Laf), Jake Foster (Soph. Harrison H.S.,
WL), Ariel Steinweg-Woods (8th grade, East Tipp M.S., Laf)
 PROBLEM OF THE WEEK
Solution of Problem No. 3 (Fall 2000 Series)

Problem: If a given equilateral triangle 4 of side a can be covered by five equilateral

triangles of side b, show that 4 can be covered by four of side b.

Solution (by Mike Hamburg, Jr. St. Joseph H.S., South Bend)

Suppose 4 can be covered by 5 equilateral triangles of side b (henceforth b-triangles).

Then we assert a 2b. For if a > 2b, then the vertices and midpoints of the sides of 4
(6 points at all) are mutually separated by 12 a > b. But no 2 points on a b-triangle are
separated by a distance greater than b, hence no b-triangle can cover more than one of the
6 points.

But if a 2b then 4 b-triangles can be arranged to form a 2b-triangle which covers 4.

Also solved by:

Undergraduates: Kevin Darkes (Soph. A&AE), Haldun Kufluoglu (Sr. EE), James Lee
(Sr. MA/CS), Stevie Schraudner (Jr. CS)

Graduates: Gajath Gunatillake (MA), Chris Lomont (MA)

Faculty & Staff: Steven Landy (Phys. at IUPUI)

Others: Damir Dzhafarov, Jake Foster (Sr. & Soph., resp., Harrison H.S., WL)

There was one incorrect solution.

 PROBLEM OF THE WEEK
Solution of Problem No. 4 (Fall 2000 Series)

Problem: Let x1 , x2 , , xn be n points in space. Between any pair (xi , xj ) there is an

arrow either from xi to xj or from xj to xi (this is a complete oriented graph of size n).
Show that there is a path xa1 xa2 xan which includes all of x1 , . . . , xn and
proceeds in the direction of the arrows.

Solution (by the Panel)

Proof by induction on n. The assertion is trivial for n = 1 and n = 2. Assume it is

true for all k < n. Choose any k, 1 < k < n. Let A be the set of i for which xi xk , and
B the set of i for which xk xi . By the induction assumption the {xi } with i A can
be arranged as {xai } so that xa1 xa2 xak1 ; likewise the set {xi } with i B
can be arranged so that xak+1 xan . Then xa1 xa2 xak xan
is the desired path.

Solved by:
Undergraduates: Kevin Darkes (Soph. A&AE), James Lee (Sr. MA/CS), Yee-Ching Yeow
(Jr. Math)

Graduates: Vikram Buddhi (MA), Yalin Firat Celikler (MA), Gajath Gunatillake (MA),
Wook Kim (MA), Chris Lomont (MA), Mohammed Majidi (MA visitor)

Faculty: Steven Landy (Phys. at IUPUI)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., Laf), Jake Foster (Soph. Harrison H.S.,
WL), Mike Hamburg (Jr. St. Joseph H.S., South Bend)

There was one incorrect solution.

 PROBLEM OF THE WEEK
Solution of Problem No. 5 (Fall 2000 Series)

Problem: Prove that the polynomials

Pn (x) = x2n 2x2n1 + 3x2n2 2nx + 2n + 1 (n = 1, 2, ) have no real zero.

Solution (by Vikram Buddhi, Gr. Math)

Clearly Pn (x) > 0 for x 0, since the terms with negative coefficients are multiplied
by odd powers of x. Now

Pn (x) + xPn (x) = x(x2n x2n1 + x2n2 x + 1) + 2n + 1

so

x2n+1 + 1
Pn (x) > x + 2n + 1
x+1
Pn (x) > 2n + 1 for x > 0.

Also solved by:

Undergraduates: Heung-Keung Chai (Sr. EE), James Lee (Sr. MA/CS), Jeffrey D. Moser
(Fr. MA/CS), Yee-Ching Yeow (Jr. Math)

Graduates: Gajath Gunatillake (MA), Wook Kim (MA), Thierry Zell (MA)

Faculty: Steven Landy (Phys. at IUPUI)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., WL), Jake Foster (Soph. Harrison H.S.,
WL), Mike Hamburg (Jr. St. Joseph H.S., South Bend)

Two unacceptable solutions were received.

 PROBLEM OF THE WEEK
Solution of Problem No. 6 (Fall 2000 Series)
R
Problem: Evaluate the integral I = 0 Arctan(ax)Arctan(bx)
x
dx where a and b are positive
numbers. Hint: Express I as a double integral.

Solution (by Gajath Gunatillake, Gr. Math, and many others)

Z a
x
Arctan ax Arctan bx = dt.
b 1 + x2 t 2
Hence, Z Z Z Z
a a
1 x 1
I= dt dx = dt dx.
0 x b 1 + x2 t 2 0 b 1 + x2 t 2
Change of order of integration gives
Z aZ Z a Z
1 1 t
I= 2 2
dx dt = dx dt
1+x t b t 0 1 + x2 t 2
Z a
b 0
1 a
= ( 0)dt = log( ).
b t 2 2 b

Also solved by:

Undergraduates: Stevie Schraudner (Jr. CS), Yee-Ching Yeow (Jr. Math)

Graduates: Vikram Buddhi (MA), Yi-Ru Huang (Stat), Wook Kim (MA), Sravanthi Kon-
duri (CE), B. N. Reddy Vanga (Nucl E)

Faculty: Steven Landy (Phys. at IUPUI)

Others: Mike Hamburg (Jr. St. Josephs H.S., South Bend)

There was one incorrect solution.

Correction: A correction must be made to the published solution of Problem No. 5, as

pointed out by Vikram Buddhi.
x2n+1 + 1
Pn (x) > x + 2n + 1
x+1
should be replaced by
x2n+1 + 1
(1 + x)Pn (x) = x + 2n + 1, hence Pn (x) > 0 for x > 0.
x+1
 PROBLEM OF THE WEEK
Solution of Problem No. 7 (Fall 2000 Series)

Problem: Show that, for every positive integer n,

  2n1   2n+1
2n 1 2
2n + 1 2
< 1 3 5 (2n 1) < .
e e

Solution (by Vikram Buddhi, Gr. Math)

By considering the graphs of log x and log(x2) and partitioning the interval [3, 2n+1]
into equal subintervals of width 2, we find
Z 2n+1 Z 2n+1
log(x 2)dx < 2[log 3 + log 5 + + log(2n 1)] < log x dx.
3 3

Hence, integrating:

(2n 1) log(2n 1) (2n 1) + 1 < 2[log 3 + + log(2n 1)]

< (2n + 1) log(2n + 1) (2n + 1) 3 log 3 + 3.

Taking exponents gives

  2n1   2n1   2n+1  e  32   2n+1
2n 1 2
2n 1 2
1 2n + 1 2
2n + 1 2
< e < 13 (2n1) <
2 < .
e e e 3 e

Also solved by:

Graduates: Wook Kim (MA)

Faculty: Steven Landy (Phys. at IUPUI)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., WL), Mike Hamburg (Jr. St. Josephs
H.S., South Bend)

There were two incorrect solutions.

Correction of Problem No. 8:

Problem No. 8 is incorrect as stated. The hypothesis the ratio of the length of a side
to the length of the base is rational should be replaced by the length of a side and the
length of the base are rational.
 PROBLEM OF THE WEEK
Solution of Problem No. 8 (Fall 2000 Series)

Problem: Let 4 be an isosceles triangle for which the length of a side and the length of
the base are rational. Prove that the radius of the incircle of 4 is rational if and only if
the two right triangles formed by the altitude to the base are similar to a right triangle
with integer side lengths.

Solution (by the Panel)

Let r be the length of the radius of the inscribed circle, a the length of the side, 2b the
length of the base, h the length of the altitude to the base. The area A of the triangle can
be expressed in two ways:
2A = r(2a + 2b) = 2bh,

hence r = bh/(a + b), and r is rational if and only if h is rational.

Now if h is rational, then n N such that an, bn, hn are all in N, and form the sides
of a triangle similar to the original triangle with sides a, b, h.
Conversely, if the triangle with sides b, a, h is similar to one with integral sides then
t R such that b2 t2 = a2 t2 + h2 t2 ; bt, at, ht N, but a is rational, so t is rational and,
therefore, h is rational.

Also solved by:

Undergraduates: Jeffrey D. Moser (Fr. MA/CS), Yee-Ching Yeow (Jr. Math)

Graduates: Gajath Gunatillake (MA), Wook Kim (MA)

Faculty: Steven Landy (Phys. at IUPUI)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., WL), Mike Hamburg (Jr. St. Josephs
H.S., South Bend)

Late: A correct solution of Problem 7 was mailed on time from Japan, but was received
late. The solver was Tetsuji Nishikura, a physician of Hyougo Prefecture.
 PROBLEM OF THE WEEK
Solution of Problem No. 9 (Fall 2000 Series)

Problem: Define a strip S to be the open set of points in the plane lying between two
parallel lines. Let |S| be the width of S. Given an infinite sequence {Si } of strips, show
that there are points in the plane that are not in any of the Si if |Si | converges.

Solution (by Steven Landy, Fac. Physics at IUPUI)

Let 1 |Si | = w. Consider the intersection of the union of the strips with a cir-
cular disk of radius R. Each strip Si intersects the disk with a length 2R. So the
area of the intersection is 2R|Si | and the area of the intersection of Si with the disk
2R|Si | 2Rw. Choose R > 2w/ then the area of the circle in R2 > 2wR 2R|Si |,
so some of the points of the disk are not in any of the Si .

Also solved by:

Graduates: Gajath Gunatillake (MA)

Others: Mike Hamburg (Jr. St. Josephs H.S., South Bend)

There was one unacceptable solution.

 PROBLEM OF THE WEEK
Solution of Problem No. 10 (Fall 2000 Series)

p 1 1 p
Problem: Given a rational number , show that the equation + = has only finitely
q x y q
many positive integer solutions.

Solution (by Steven Landy, Fac. Physics at IUPUI)

1 1 1 1 p 2q
WLOG may assume . Then , so x . There are only finitely many
x y x 2 q p
2q
positive integers that are no larger than , and since with each solution x, there is only
p
one y, there are only finitely many solutions.

Also solved by:

Undergraduates: Stevie Schraudner (Jr. CS)

Graduates: Gajath Gunatillake (MA), Yi-Ru Huang (Stat)

Others: Jonathan Landy (Jr. Warren Central H.S., Indianapolis)

 PROBLEM OF THE WEEK
Solution of Problem No. 11 (Fall 2000 Series)

Problem: A particle moves in a vertical plane from rest under the influence of gravity
and a force perpendicular to and proportional to its velocity. Obtain the equation of the
trajectory, and identify the curve.

Solution (by Steven Landy, Fac. Physics at IUPUI)

Assume the mass is dropped from x = y = 0 at t = 0. Let y = positive down. We then

have (F = kv)
my = mg k x, mx = k y,

or
y = g wx, x = wy, where w = k/m.
...
Substituting gives y = w2 y,
or
B B
y = B sin wt y = cos wt + ,
w w
B
x = wB sin wt x = sin wt + Bt.
w
Substitution in original differential equation gives B = g/w.
Finally

g g
x= 2
sin wt + t
w w
g g
y = 2 cos wt + 2 .
w w

These determine a cycloid produced by a wheel of radius R = g/w2 rolling on the x axis
g
at speed where w = k/m.
w

Also solved by:

Undergraduates: Benjamin Zwickl (Fr. Phys)
 PROBLEM OF THE WEEK
Solution of Problem No. 12 (Fall 2000 Series)

Problem: Given a triangle with vertices A, B, C and points A1 , B1 , C1 on the sides

BC, CA and AB, respectively, prove that the circumcircles of the triangles 4AB1 C1 ,
4BA1 C1 , and 4CA1 B1 have a common point.

Solution (by the Panel)

The circumcircles of 4AB1 C1 and 4BA1 C1 have the point C1 in common, hence have
another point P in common unless they are tangent (to be discussed later). There are two
cases to be considered.

a) P lies inside 4ABC, then we have quadrangles AB1 C1 P and BP A1 C1 inscribed in the
circles. It follows that B1 P C1 = 180 B1 AC1 and A1 P C1 = 180 A1 BC1 . So
B1 P A1 = 180 A1 CB1 ; thus the quadrangle B1 P A1 C has a circumcircle and P lies
on the circumcircle of 4B1 CA1 .

b) If any pair of the circumcircles intersects in a point other than A1 , B1 , or C1 , relabel the
original triangle so these are the circumcircles of 4AB1 C1 and 4BA1 C1 . Now the quad-
rangles AB1 C1 P and BP A1 C1 are not convex, and B1 P C1 = B1 AC1 and A1 P C1 =
A1 BC1 . The quadrangle CB1 P A1 is convex and B1 P C1 + A1 P C1 = 180 B1 CA1 ;
therefore this quadrangle has a circumcircle which must be that of 4B1 CA1 , so P lies on
this circle.

c) If two of the circumcircles are tangent, say at point C1 , then C1 is a limit point of points
for which such tangency does not occur, and the result is obtained by continuity.

Partially solved by:

Graduates: Gajath Gunatillake (MA)

Faculty & Staff: Steven Landy (Phys. at IUPUI), Sebastien Mercier (Research, Chem.)

Others: Mike Hamburg (Jr. St. Josephs H.S., South Bend)

 PROBLEM OF THE WEEK
Solution of Problem No. 13 (Fall 2000 Series)

Problem: Show that in the (x, y) plane, for odd integers A, B, C, the line Ax+By +C = 0
cannot intersect the parabola y = x2 in a rational point.

Solution (by Steven Landy, Fac. Physics at IUPUI)

The x-coordinate of the point of intersection of the line and parabola is found from
the equation
Ax + Bx2 + C = 0.

Assume x is rational, x = p/q, where p, q are integers, not both even. Then

Apq + Bp2 + Cq = 0.

If both p, q are odd, we have a contradiction, because the sum of three odd numbers cannot
be zero. If p is even, q odd, we have again a contradiction. The same is true if p is odd
and q is even. Hence, x cannot be rational.

Also solved by:

Undergraduates: Jeffrey D. Moser (Fr. MA/CS), Stevie Schraudner (Jr. CS)

Graduates: Vikram Buddhi (MA), Gajath Gunatillake (MA), Chris Lomont (MA)

Faculty: Sebastien Mercier (Research, Chem.), Ralph Shines (GAANN Fellow, MA)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., WL), Jonathan Landy (Jr. Warren
Central H.S., Indianapolis)

There was one incorrect solution.

 PROBLEM OF THE WEEK
Solution of Problem No. 14 (Fall 2000 Series)

Problem: Consider the equations N = x3 (3x + 1) = y 2 (y + 1)3 , where x, y are relatively

prime positive integers. Show that there is only one possible value for N . Find it.

Solution (by Steven Landy, Fac. Physics at IUPUI)

Since (x, y) = 1 it follows that x3 |(y + 1)3 , hence x|(y + 1), x y + 1. Similarly
y 2 /(3x + 1), so y 2 3x + 1. Combining the inequalities gives x2 5x = 0. As x is positive,
only x = 1, 2, 3, 4, 5 are possible. Trying these values in x3 (3x + 1) = y 2 (y + 1)3 , we find
that only x = 5, y = 4 work, so N = 42 53 = 2000 is the only solution.

Also solved by:

Undergraduates: Stevie Schraudner (Jr. CS)

Graduates: Vikram Buddhi (MA), Gajath Gunatillake (MA), Yi-Ru Huang (Stat), Chris
Lomont (MA)

Faculty & Staff: Sebastien Mercier (Research, Chem.), Ralph Shines (GAANN Fellow,
MA)

Others: Damir D. Dzhafarov (Sr. Harrison H.S., WL)

Two incorrect solutions were received.