Revision Notes FP1-3

Further Concepts for Advanced Mathematics FP1
Proof
There are three main techniques of proof used in Mathematics
(i) A logical sequence of steps leading from some known fact to conclude with what we wish to prove.
(ii) Proof by contradiction. ie, showing that the assumption that a statement is false leads to a
mathematical contradiction.
(iii) Proof by induction .
The first two cases were dealt with earlier in the course.
Proof by induction
Suppose that P(n) is a statement concerning the natural number n, i.e. P(n) is a function whose
domain is , theset of natural numbers.
The PRINCIPLE OF MATHEMATICAL INDUCTION states that if
(i) P
(
n
0
)
istruefor somen
0
and(ii) P
(
k
)
P
(
k + 1
)
for anyk n
0
, thenP
(
n
)
istruefor all n n
0
Proof: If the above is not true then there exists a subset of the natural numbers consisting of those
numbers greater than , n
0
for whichP
(
n
)
isfalse. Bythewell-orderingpropertyof thenatural numbers
This subset has least member m say. i.e. , since m > n
0
andP
(
m
)
isfalse. However, P
(
m 1
)
must betrue
m is the least member of the subset and then, by the principle of induction P contradicting the (m) istrue
assumption that it is false. Hence, no such subset can exist.
Ex. Prove that for all
n ,
E
r=1
n
r =
1
2
n(n + 1)
Let P
(n) be the statement that
E
r=1
n
r =
1
2
n(n + 1)
When
n = 1,
E
r=1
n
r = 1 and
1
2
n(n + 1) =
1
2
1 2 = 1 so P(1) is true.
Assume P
(k) true, i.e.
E
r=1
k
r =
1
2
k(k + 1) then
E
r=1
k+1
r =
1
2
k(k + 1) + (k + 1) =
1
2
(k + 1)(k + 2)
But this is P
(
k + 1
)
soP
(
k
)
P(k + 1) andhenceP
(
n
)
istruefor all n byinductiononn
Another important use for the induction method of proof is to test divisibility statements.
Ex. Prove that 13
n
6
n2
isdivisibleby7for all positiveintegersgreater thanm, wher m istobe
found.
Let f(n) = 13
n
6
n2
thenf(1) = 13 6
1
whichisnot aninteger.
so f(2) is divisible by 7 so we take m =2. f(2) = 13
2
6
0
= 169 1= 168= 7 26
Assume now that f
(
k
)
= 13
k
6
k2
= 7X whereX isaninteger, for somek 2
Now f
(
k + 1
)
= 13
k+1
6
k1
= 13(13
k
6
k2
) + 13 6
k2
6
k1
= 13 7X + 6
k2
(13 6) = 7(13X + 6
k2
)
And since 13X + 6
k2
isaninteger for k 2it followsthat f
(
k + 1
)
isdivisibleby7.
Hence, true for n = k truefor n = k + 1andtheresult followfor all n 2byinductiononn.
1
Create PDF files without this message by purchasing novaPDF printer (http://www.novapdf.com)
Sequences and Series
A function whose domain is the set of natural numbers is called a SEQUENCE. The elements of the
range of the function are the TERMS of the sequence. There are two common methods of defining a
sequence
(1) By giving a formula for the general term. Which again is done in one of two ways.
Ex. defines the sequence 1, 3, 5, 7, ... f(n) = 2n 1
Ex. defines the sequence 0, 7, 26, 63, ... n
3
1
(2) By expressing the

term in terms of the previous term or sometimes the previous two n
th
or (n + 1)
th
terms. is commonly used to denote the term of a sequence. u
n
n
th
Ex. with defines the sequence 1, 1, 2, 6, 24, ... u
n+1
= nu
n
u
1
= 1
Ex. defines the sequence 1, 1, 2, 5, 13, ... u
n+1
= 3u
n
u
n1
These are called RECURRENCE RELATIONS
Series
The sum of the first n terms of a sequence forms a FINITE SERIES. If all the terms of a sequence are
added we have an INFINITE SERIES
e.g. is a finite series of six terms. 1+ 2+ 4+ 8+ 16+ 32
is an INFINITE SERIES 1+ 2+ 4+ 8+ 16+ 32+ ... + 2
n
+ ...
Behaviour of series
Sequences and series may behave in a number of ways as the number of terms increase. The sum of the
series may
(i) Tend to some finite limiting value. We say that it CONVERGES to that value
(ii) Increase in size without limit. We say it DIVERGES
(iii) OSCILLATE between finite or infinite limits.
(iv) Display PERIODICITY. i.e the same sequence of values repeat at regular intervals.
Ex. (i) converges to the value 2. We say the sum to infinity is 2 1+
1
2
+
1
4
+ ... +
1
2
n+1
+ ...
(ii) diverges. The sum tends to infinity. 1+ 2+ 4+ 8+ 16+ ... + 2
n
+ ...
(iii) oscillates 1 2+ 4 8+ 16 ...
(iv) displays periodicity {1 (1)
n
} = {2,0,2,0, ...}
Note that a sequence is a collection of numbers whereas a finite series has a unique finite numerical value.
An infinite series may or may not have a finite value.
We may occasionally relax the requirement that the domain of our function be the set of natural numbers
and allow integer values or omit the first few values. We commonly use the sigma notation when dealing
with series. i.e. which means that we add together the values of for all integer values of r
E
r=1
n
f(r)
f(r)
between 1 and n inclusive.
Ex. (i) (ii)
E
r=1
r=4
r = 1 + 2 + 3 + 4 = 10
E
1
5
(3r 1) = 2 + 5 + 8 + 11 + 14 = 40
Note also that we often omit the when it can cause no confusion. r =
(iii)
E
3
6
(r
2
5) = 4 + 111 + 20 + 31 = 66
Note also that the same series may be defined in many different ways
Ex. etc.
E
1
n
r(r 1) =
E
0
n1
r(r + 1) =
E
2
n+1
(r 1)(r 2)
2
Summation of Finite Series
Wehavepreviouslymet threespecial series, thearithmetic, geometricandbinomial series.
Wenowconsider somespecial techniques.
Sums of Powers of Natural Numbers
We already know that
E
r=1
n
r =
1
2
n(n + 1) and
E
r=1
n
r
2
=
1
6
n(n + 1)(2n + 1)
Thefirst result isimmediatefromthesumof anarithmeticserieswhilst thesecondmaybeobtainedas
follows: (r + 1)
3
r
3
= 3r
2
+ 3r + 1 r
2
=
1
3
(r + 1)
3
r
3
3r 1
E
r=1
n
r
3
=
1
3 E
r=1
n
(r + 1)
3
r
3
3
E
r=1
n
r n =
1
3
(n + 1)
3
1
3
2
n(n + 1) n
=
1
3
n(n + 1)|2(n + 1)
2
3n 2| =
1
6
(n + 1)(2n
2
+ n) =
1
6
n(n + 1)(2n + 1)
Similarly, startingwiththeidentity(r + 1)
4
r
4
= 4r
3
+ 6r
2
+ 4r + 1wecandeducethesumof the
cubesof thenatural numbers. Thisisleft asanexercise.
Theseresultscanbeusetosumother series.
Ex Findthesumof n termsof 1 1+ 2 3+ 3 5+ 4 7+ ... + n(2n 1)
As the r
th
termis r(2r 1) wehave S
n
=
E
r=1
n
r(2r 1) = 2
E
r=1
n
r
2

E
r=1
n
r
=
1
3
n(n + 1)(2n + 1)
1
2
n(n + 1) =
1
6
n(n + 1)(2(2n + 1) 3) =
1
6
n(n + 1)(4n 1)
The Method of Differences.
If, for eachr, wecanwritether
th
termof aseriesintheformf
(
r + 1
)
f
(
r
)
thenS
n
= f
(
n + 1
)
f
(
1
)
Ex
Provethat
E
r=1
n
r =
1
2
n(n + 1)
Asanalternativetousingthesumof anarithmeticseries, notethat r =
1
2
[
r(r + 1) (r 1)r
]
i.e. r =f(r + 1) f(r) where f(r ) =
1
2
(r 1)r, hence,
E
r=1
n
r =f(n + 1) f(1) =
1
2
(n + 1)n since f(1) = 0
Ex Findthesumof
1
12
+
1
23
+ ... +
1
n(n+1)
Bypartial fractions
1
n(n+1)
=
1
n

1
n+1
= f(n) f(n + 1) =
[
f(n + 1) f(n)]
wheref(n) =
1
n
hence,
E
r=1
n
1
n(n+1)
= f(1) f(n + 1) = 1
1
n+1
=
n
n+1
Sometimesaseriescanbereducedtoastandardrecognisableform.
Ex Findthesumof x + 2x
2
+ 3x
3
+ ... + nx
n
If S
n
= x + 2x
2
+ 3x
3
+ ... + nx
n
thenxS
n
= x
2
+ 2x
3
+ 3x
4
+ ... + nx
n+1
sobysubtractionS
n
xS
n
= x + x
2
+ x
3
+ ... + x
n
nx
n+1
=
x(1x
n
)
1x
nx
n+1
(bysumof G.P)
i.e. S
n
=
x[1x
n
nx
n
(1x)]
1x
=
x(1(1+n)x
n
+nx
n+1
)
1x
3
Complex Numbers
Though the number system has already been extended in many ways, starting with the natural numbers,
through the integers and rational numbers to the real number system, there are still many quite simple
equations which have no solutions within this system, e.g. x
2
+ 2= 0, sinx = 2, 10
x
= 5etc
Consider quadratic equations, we know that the general quadratic ax
2
+ bx + c = 0hassolutions
x =
1
2a
b b
2
4ac andthat therearenoreal rootsif b
2
4ac < 0
Suppose now that so we may write the b
2
4ac = k = k(1) wherek > 0, then b
2
4ac = k 1
solutions as x = p q 1
Let us now introduce a special symbol to denote 1. Weshall usej (thoughi isalsooftenused)
Thus, if the roots of a quadratic equation are not real, we may write x = p qj
We define a COMPLEX NUMBER to be a number of the form a + bj wherea andb arereal, the+ sign
being merely a means of linking the two parts of the number together.
a is called the REAL part and b the IMAGINARY part of the number, hence if then z = a + bj
. Note that the imaginary part of a complex number is real, since it is the Re
(
z
)
= a andIm
(
z
)
= b
coefficient of j and does not include j itself.
Manipulation of complex numbers
Operations on complex numbers are defined as follows by assuming the usual properties of numbers to
hold together with the assumption that j
2
= 1
thus: Equality a + bj= c + dj a = c andb = d
Addition/subtraction (a + bj) (c + dj) = (a c) + (b d)j
Multiplication (a + bj)(c + dj) = ac + (bc + ad)jbd = (ac bd) + (bc + ad)j
Division Since (a + bj)(a bj) = a
2
+ b
2
wehave
a+bj
c+dj
=
a+bj
c+dj

cdj
cdj
=
(ac+bd)+(bcad)j
c
2
+d
2
from these definitions we can see that the complex numbers of the form behave in exactly the same a + 0j
way as the real numbers and it is usual to omit the . Similarly, purely imaginary numbers 0j 0+ bj
are usually written simply as bj.
If z = a + bj, thenthemultiplicativeinversez
1
of z isgivenbyz
1
=
1
a+bj
=
abj
a
2
+b
2
providingz = 0
The number z
+
= a bj iscalledtheCOMPLEX CONJUGATE (or simplytheconjugate) of z
The real number z = a
2
+ b
2
iscalledtheMODULUSof z, thusz
1
=
z
+
z
2
or z
2
=
z
+
z
1
= zz
+
Some important results concerning complex conjugates are:
If z
1
= x
1
+ y
1
j andz
2
= x
2
+ y
2
j withz
2
= 0then
and (z
1
+ z
2
)
+
= z
1
+
+ z
2
+
; (z
1
z
2
)
+
= z
1
+
z
2
+
;
z
1
z
2
+
=
z
1
+
z
2
+ ; z
1
+ z
1
+
= 2Re(z
1
) z
1
z
1
+
= 2jIm
(
z
1
)
these results are most important and should be memorised
The square root of a complex number
The method is explained by an example.
Ex Find 21 20j
Let 21 20j= (a + bj)
2
= (a
2
b
2
) + 2abj thenwemust havea
2
b
2
= 21and2ab = 20
It is fairly obvious, by inspection, that the possible solutions are If you a = 5, b = 2anda = 5, b = 2
cannot see these solutions then you must solve simultaneously. a
2
b
2
= 21and2ab = 20
Thus we have . Note that since the complex numbers do not form an 21 20j = 5 2j or 5+ 2j
ordered set, we cannot say that one of these roots is positive and the other negative.
Consider the polynomial equation It can be proved a
n
x
n
+
n1
x
n1
+ ... + a
1
x + a
0
= 0withall a
i
real.
that such an equation always has at least one root in . Remember that , theset of complexnumbers
contains . Let sucharoot bez = a + bj. thena
n
z
n
+
n1
z
n1
+ ... + a
1
z + a
0
= 0
Taking the complex conjugate of both sides and using the standard results given above we have, since
a
i
+
= a
i
thena
n
z
n
+
+ a
n1
z
n1
+
+ ... + a
1
z
+
+ a
0
= 0 z
+
= a bj isalsoaroot of theequation
Thus, complex roots of a polynomial equation with real coefficients, always occur in conjugate pairs.
What are the important consequences of this ?
Ex Solve the equation z
3
+ z
2
4z + 6= 0giventhat 1+j isaroot.
If 1+j isaroot thensoalsois1j hencez (1+j) andz (1j) arefactorsof thepolynomial
and hence so is |z (1+ j)||z (1 j)| = |(z 1) j||(z 1) + j| = (z 1)
2
+ 1= z
2
2z + 2
4
and so z
3
+ z
2
4z + 6= (z
2
2z + 2)(z + 3) andthesolutionsarez = 1+j, 1j or 3
The Argand Diagram
Since a complex number is essentially just an ordered pair of real numbers there is a natural one-to-one
correspondence between complex numbers and points in a plane defined by the mapping x + yj(x, y)
Points on the x-axis correspond to the real numbers and points on the y-axis to the purely imaginary
numbers and so we refer to the real and imaginary axes.
Such a representation of the complex numbers is known as the COMPLEX PLANE or the ARGAND
DIAGRAM (J.R.Argand 1768-1822)
The point P(x,y) correspondingtoz = x + yj iscalledtheAFFIX of z
The Argand diagram gives us an alternative interpretation of j , not as a number, but an operator, an
instruction to perform a certain geometrical transformation.
Let under z = a + bj beanycomplexnumber withaffixP. Thenjz = ajb withaffixP
'
istheimageof P
a positive (anti-clockwise) rotation of . Thus j can be interpreted as an operator 90
o
about theorigin
which causes a positive 90
o
rotationabout theorigin.
It should be obvious from the definition of addition/subtraction that in the Argand diagram it is exactly
the same as adding/subtracting vectors. The MODULUS which has already been defined can now be seen
to be the distance of the affix of a number from the origin. Before considering multiplication and division
in the Argand diagram we need to introduce the concept of the ARGUMENT of a complex number.
The position vector of the point representing a complex number z in the Argand diagram can be described
by means of its length r and the angle 0 that it makeswiththepositivereal axis.
r isof course z themodulusof z. Theangle0, whichismeasuredanticlockwisefromthepositivereal
axis, usually in radians, is not however uniquely defined since adding any multiple of will give the 2v
same direction. It is usual therefore to take that value of which is called the 0 for whichv < 0 v,
PRINCIPAL ARGUMENT of z, denotedbyargz
If arg(z) = 0 wehavez = r cos0 + (r sin0)j
ie z = r(cos0 +jsin0) wheretan0 =
Im(z)
Re(z)
This is the MODULUS-ARGUMENT or POLAR form of z.
Ex. (i) 1+j= r(cos0 +jsin0) r = 2, tan0 = 1 0 =
v
4
so 1+j= 2 cos
v
4
+ j sin
v
4
(ii) 3 4j= 5(cos0 +jsin0) wheretan0 =
4
3
and
v
2
< 0 < 0
Be sure to get the correct quadrant for the argument.
Consider now z
1
= r
1
(cos0
1
+jsin0
1
) andz
2
= r
2
(cos0
2
+jsin0
2
)
Then z
1
z
2
= r
1
r
2
(
cos0
1
+ j sin0
1
)
(cos0
2
+jsin0
2
)
= r
1
r
2
(cos0
1
cos0
2
sin0
1
sin0
2
) +jsin0
1
cos0
2
+ cos0
1
sin0
2
)
= r
1
r
2
|cos(0
1
+ 0
2
) +jsin(0
1
+ 0
2
)|
Thus z
1
z
2
= z
1
z
2
andarg(z
1
z
2
) = arg(z
1
) + arg(z
2
) 2v
Similarly,
z
1
z
2
=
z
1
z
2
andarg
z
1
z
2
= arg(z
1
) arg(z
2
) 2v
Note that the 2v maybenecessaryinorder togiveaprincipal
argument.
Geometrically therefore, multiplication by z enlarges by a factor
z androtatesthroughanangleargz
This is called a SPIRAL DILATION.
In the diagram, I, P
1
andP
2
aretheaffixesof 1,z
1
andz
2
respectively. If each side of cOIP
1
isenlargedbya
factor z
2
andthetrianglerotatedthroughananglearg(z
2
)
then then I ismappedontoP
2
. Let P
3
betheimageof P
1
'
s OIP
1
andOP
2
P
3
aresimilar andthus
and hence, OP
3
= z
2
OP
1
= z
2
z
1
andIO
P
3=arg(z
1
)+arg(z
2
)
P
3
istheaffixof z
1
z
2
Thus, to construct similar to z
1
z
2
wedrawonOP
2
, atriangle OIP
1
, makingOP
2
correspondtoOI
5
P z
( )
x
y
u
r
P
3
( ) z z
1 2
P
2
( )
2
z
P
1
( )
1
z
x
y
O
I
Simple loci in the Argand diagram
From the connection between complex numbers and vectors we see that between z
1
z
2
isthedistance
the points representing the complex numbers Thus if z
1
andz
2
. A,B andP represent the
complex constants a,b andthecomplexvariablez then
(i) if z a = k wherek isreal, thelocusof P isacircle, centreA andradiusk.
(ii) if z a = z b thenthelocusof P istheperpendicular bisector of AB
(iii) if z a = k z b k = 1thenthelocusof P isacircle.
We prove this last result. Let z = x +jy, a = a
1
+ja
2
andb = b
1
+jb
2
then z a = k z b x a
1
+ j(y a
2
) = k x b
1
+ j(y b
2
)
(x a
1
)
2
+ (y a
2
)
2
= k
2
(x b
1
)
2
+ (y b
2
)
2
x
2
+ y
2
2a
1
x 2a
2
y + a
1
2
+ a
2
2
= k
2
x
2
+ y
2
2b
1
x 2b
2
y + b
1
2
+ b
2
2
(k
2
1)(x
2
+ y
2
) + 2(a
1
k
2
b
1
)x + 2(a
2
k
2
b
2
)y + k
2
(b
1
2
+ b
2
2
) a
1
2
+ a
2
2
= 0
and dividing by (since which is the k
2
1 = 1) givesanequationof theformx
2
+ y
2
+ 2gx + 2f y + c = 0
equation of a circle.
The centre of the circle is at which is the (g,f) i.e. thepoint representing
k
2
b
1
a
1
k
2
1
+j
k
2
b
2
a
2
k
2
1
=
k
2
ba
k
2
1
point dividing AB externally in the ratio k
2
: 1
(iv) if arg(z a) = 0. where0 isreal, thelocusof P isthehalf-linefromA, (excludingA itself)
proceeding in the direction 0, measuredfromthereal axis.
(v) if from BP arg
za
zb
= ,, thelocusof P isacircular arcfromA toB suchthat APB = ,, measured
to AP.
(iv) and (v) are illustrated in the following diagrams.
1
3
v
A (2+ j)

2
3
v
A (3 j) (1 2j) B
arg(z 3+ j) =
2
3
v arg
z2j
z+1+2j
=
1
3
v
6
O
x
y
1 2
3
1
1
2
3
0 1 2 1
2
1
Z
Matrices
A MATRIX is an ordered rectangular array of numbers. Matrices are classified by the number of rows
and columns, the ORDER of the matrix. Thus a matrix with 4 rows and 3 columns is said to be a 4 3
Matrix, read as four-by-three. We usually use capital letters printed in heavy type in examination papers
and most text books, to represent matrices except when the matrix consists of a single column when
standard vector notation tends to be used, since a vector may be considered to be an matrix. n 1
Arbitrary elements of a matrix are denoted by small letters often with one or two suffixes as in the
following examples.
Ex. A C =
1 3 5
4 1 2
B =
a
1
a
2
a
3
b
1
b
2
b
3
=
a
11
a
12
a
13
a
21
a
22
a
23
a
31
a
32
a
33
A and B are . 2 3matricesandC isa3 3matrix
Algebra of matrices.
Let A = a
ij
, B = b
ij
, C = c
ij
wherea
ij
representstheelement inrowi andcolumnj of A
Equality: A =B if, and only if, a
ij
= b
ij
for everyi,j. ClearlyA andB must havethesameorder.
Addition/subtraction: Providing A and B have the same order (are COMPATIBLE) then their sum and
difference will exist and A B = C wherec
ij
= a
ij
b
ij
for everyi, j.
Multiplication by a scalar: k A = ka
ij
Multiplication of matrices: If A has order r s andB hasorder s t thenAB will exist andwill have
order . We say that A and B are CONFORMABLE for multiplication. Note that the product BA will r t
only exist when r =t and even then AB and BA will not, in general, be equal. It is important to remember
that matrix multiplication is NOT commutative.
When it exists, the product AB is defined by AB =C where i.e. The element in the i-th
c
ij
=
E
r=1
s
a
ir
b
rj
row and j-th column of the product AB is the scalar product of the i-th row of A and the j-th column of
B. If the product does not exist then the matrices are NON-CONFORMABLE.
A square matrix is called the IDENTITY matrix a
ij
of order n witha
ii
= 1for all i anda
ij
= 0for i = j
of order n and denoted by I (or . This matrix always commutes with I
n
if wewishtoemphsisetheorder)
any square matrix of the same order. i.e. AI =IA =A
Any matrix in which all elements are zero is called a ZERO or NULL matrix and denoted by 0
Note that, unlike in basic algebra, AB =0 does NOT imply that either A =0 or B =0.
Ex. 0
3 2
0 0
0 0
5 6
=
0 0
0 0
=
Inverse of a square matrix: If A,B are square matrices of order n and if AB =I, then B is the INVERSE
of A and is more usually denoted by A
1
. Note that a matrix always commutes with its inverse,
i.e. AA
1
=A
1
A =I
Division : There is no operation of division defined on matrices though a similar effect is obtained by
multiplying by the inverse of a matrix. Thus, AB =C providing A
1
exists, but remember B =A
1
C
That you must be very careful with the order of your multiplications.
Ex. AB = C B = A
1
C but wemust NOT sayB = CA
1
Inverse of a 2 x 2 matrix: If A . =
a b
c d
then
a b
c d
= ad bc istheDETERMINANTof A
Alternative notations for the determinant are det A or Providing A ad bc = 0theinverseof A will
exist and is given by does not exist, A
1
=
1
A
d b
c a
. Youshouldmemorisethispattern. If A
1
which is when the determinant is zero, then we say that A is SINGULAR, otherwise A is
NON-SINGULAR
Simultaneous equations.
The simple simultaneous equations 7x 2y = 11and8x 3y = 9, maybewritteninmatrixformas
of
7 2
8 3
x
y
=
11
9
andtheinverseof
7 2
8 3
is
1
5
3 2
8 7
hence, pre-multiplyingeachside
7
the matrix equation by this inverse we have
1 0
0 1
x
y
=
1
5
3 2
8 7
11
9
=
3
5
Hence, x = 3andy = 5.
Matrices and transformations
An important application of matrices is to represent mappings or transformations. The standard
transformations of a plane into itself which leaves the origin invariant, such as reflections in lines through
the origin, rotations about the origin, enlargements with centre at the origin etc, may be represented by
matrices. The images of points of the plane are then obtained by pre-multiplying the position vector 2 2
of the point by this transformation matrix.
In particular, if T =
a b
c d
isthematrixof atransformation, then
T
1
0
=
a b
c d
1
0
=
a
c
andsimilarlyT
0
1
=
a b
c d
0
1
=
b
d
Remembering this connection between the columns of the matrix and the images of the points (1,0) and
(0,1) is the easiest way of identifying a transformation from its matrix, or of finding the matrix of a given
transformation.
If we interchange the rows and columns of a matrix A we form a new matrix, called the TRANSPOSE of
A, denoted by A
T
. If A =A
T
, we say that A is SYMMETRIC, whilst if A = A
T
, then A is
SKEW-SYMMETRIC and if AA
T
=I then A is ORTHOGONAL.
If A is a Hence every 2 2matrix, thenA +A
T
issymmetricandA A
T
isskew-symmetric.
2 2matrixcanbewrittenasthesumof asymmetricandaskew-symmetricmatrix.
A is a symmetric matrix and B is skew-symmetric of the same order. AB is skew-symmetric if, and only
if, AB =BA.
A and B are symmetric matrices of the same order, then AB is symmetric if, and only if,
AB =BA. What can be said about for any positive integer k? A
k
If A and B are 2 2matrices, thendet(AB) =(detA)(detB)
_____________________________
Invariant points and lines
Points which are mapped onto themselves by a transformation are called INVARIANT points of the
transformation.
INVARIANT lines may be of two types.
(i) POINTWISE INVARIANT lines in which every individual point is mapped onto itself, and
(ii) ordinary invariant lines in which any point of the line is mapped onto a point of the line but not
necessarily the same point.
We illustrate by examples
Ex. Find the invariant points under the transformation given by the matrix
3 4
1 2
If
x
y
isthepositionvector of aninvariant point thenwemust have
3 4
1 2
x
y
=
x
y
3x + 4y = x andx + 2y = y 2x + 4y = 0andx + y = 0 x = 0, y = 0soonlytheoriginisinvariant.
Ex. Find the invariant lines through the origin of the linear transformation with matrix
4 2
5 3
Consider the image of a typical point (
k,mk
)
ontheliney = mx
4 2
5 3
k
mk
=
4k 2mk
5k 3mk
=
{4 2m}k
{5 3m}k
Hence the image of the line y = mx istheset of points
(
{4 2m}k,{5 3m}k
)
i.e. theliney =
53m
42m
x
Thus the line y = mx isinvariant if
53m
42m
= m 2m
2
7m+ 5= 0 (2m 5)(m 1) = 0 m =
5
2
or 1
Hence, the invariant lines through the origin are y = x andy =
5
2
x
8
Ex. A transformation is defined by the matrix
5 2
2 0
. Findtheinvariant lines.
Any line has an equation of the form y = mx + c withageneral point withpositionvector
This is mapped onto and we want this point
X
mX + c
say.
5 2
2 0
X
mX + c
=
5X 2mX 2c
2X
also to be on the line y = mx + c, i.ewerequirethat 2X = m(5X 2mX 2c) + c
X(2m
2
5m+ 2) = c 2mc andif thisistoholdfor all X, thenwemust have
2m
2
5m+ 2= c(1 2m) = 0 (2m 1)(m 2) = c(2m 1) = 0 m =
1
2
Hence, all lines with a gradient of .
1
2
, i.e. equationsof theform2y = x + c areinvariant
9
Relations between the coefficients and roots of an equation.
Let the roots of then a
n
x
n
+ a
n1
x
n1
+ ... + a
1
x + a
0
= 0bec
1
, c
2
,..., c
n
a
n
x
n
+ a
n1
x
n1
+ ... + a
1
x + a
0
= a
n
(x c
1
)(x c
2
)...(x c
n
)
Comparing coefficients we have ......
E
i=1
n
c
i
=
a n 1
a n E
i,j=1
n
c
i
c
j
=
a n 2
a n H
i=1
n
c
i
= (1)
n
a
0
an
where means the product of
E
i,j=1
n
c
i
c
j
means the sumof all products of the roots taken in pairs and
H
i=1
n
c
i
all the roots. In particular for
(a) ax
2
+ bx + c = 0wehavec
1
+ c
2
=
b
a
andc
1
c
2
=
c
a
(b) ax
3
+ bx
2
+ cx + d = 0wehavec
1
+ c
2
+ c
3
=
b
a
, c
1
c
2
+ c
2
c
3
+ c
3
c
1
=
c
a
andc
1
c
2
c
3
=
d
a
Ex. If Deduce the c and aretherootsof ax
2
+ bx + c = 0, express(c 2)(2c ) intermsof a,b andc.
condition that one root of the equation is double the other. Find also the condition that one root is the
square of the other.
We have c + =
b
a
andc =
c
a
hence, (c 2)(2c ) = 2c
2
+ 2
2
5c = 2(c + )
2
9c =
2b
2
a
2

9c
a
=
2b
2
9ac
a
2
One root is double the other if either c = 2 or = 2c andineither case (c 2)(2c ) = 0
hence the condition is that 2b
2
= 9ac
Similarly, one root is the square of the other if (c
2
)(c
2
) = 0
Now (c
2
)(c
2
) = c
3
+
3
c
2
2
c andc
3
+
3
= (c + )
3
3c(c + )
hence, and so the required (c
2
)(c
2
) =
b
a
3
3
c
a

b
a

(
c
a
)
2
c
a
=
b
3
+3abcac
2
a
2
c
a
3
condition is that b
3
+ ac
2
+ a
2
c 3abc = 0
Sums of powers of roots of an equation
In the previous example we required These could have been found as c
2
+
2
andc
3
+
3
follows:
Since so c and aretherootsof ax
2
+ bx + c = 0wehaveac
2
+ bc + c = 0anda
2
+ b + c = 0
by addition aS
2
+ bS
1
+ 2c = 0whereS
n
denotesc
n
+
n
Thus, since S
1
=
b
a
wehaveaS
2
b
2
a
+ 2c = 0 S
2
=
b
2
2ac
a
2
Since neither of the roots is zero, they must also satisfy the equation , hence, putting ax
3
+ bx
2
+ cx = 0
x = c andx = andaddingasbeforewehaveaS
3
+ bS
2
+ cS
1
= 0soS
3
=
bS
2
cS
1
a
=
3abcb
3
a
3
Ex. Find the sum of the cubes of the roots of the equation x
3
2x
2
+ 3x + 1= 0
We cannot find S
2
bythisnewmethodsomust proceedasfollows:
If root are c, and, thenc + + , = 2, c + , + ,c = 3andc, = 1
hence, c
2
+
2
+ ,
2
= (c + + ,)
2
2(c + , + ,c) = 4 6= 2
Now we can use our new method to give S
3
2S
2
+ 3S
1
+ 3= 0 S
3
= 2S
2
3S
1
3= 4 6 3= 13
Equations having given roots.
Reversing the earlier procedure, if the roots of an equation are then we must have c
1
, c
2
..., c
n
(x c
1
)(x c
2
)...(x c
n
) = 0 x
n
E
c
i
x
n1
+
E
c
i
c
j
x
n2
+ ... + (1)
n
H
c
i
= 0
A very common situation is where we are required to find the equation whose roots are symmetrical
functions of the roots of a given equation. The following examples illustrate the technique:
Ex. If c, , , aretherootsof x
3
3x + 2= 0, findtheequationswhoserootsare
(i)
2
]
,
2
and
2
,
(ii) c
2
,
2
, ,
2
(iii) c + , + ,, , + c
(i) writing y =
2
x
wehavex =
2
y
Substitutingfor x inthegivenequationwill produceanequationwith
the desired roots. i.e. i.e. equation is
2
y
3
3
2
y
+ 2= 0 8 6y
2
+ 2y
3
= 0 2y
3
6y
2
+ 8= 0
(ii) writing y = x
2
wehavex = y soy y 3 y + 2= 0 y y 3 y = 2 y
3
6y
2
+ 9y = 4
10
(iii) Since required equation c + + , = 0wehavec + = ,, + , = c and, + c = sotherootsof the
are c, and, hence, writingy = x wehavex = y andy
3
+ 3y + 2= 0
i.e. equation is y
3
3y 2= 0
11
Curve Sketching
Most of the techniques have been dealt with earlier e.g. Intersections with axes, Stationary points,
Transformations, Use of standard graphs etc.
Thefirst extratechniqueisthefindingof asymptotes. Thesearelineswhichthecurveapproachesbut
never actuallyreachesandaresometimeslooselyreferredtoastangentsat infinity
Weconsider herethecasesof vertical andhorizontal asymptotes.
Vertical asymptotesoccur for valuesof x whichcannot exist. Theseareusuallyvaluesof x whichmake
thedenominator of arational functionzero.
Ex y =
x
(x1)(x+2)
hasvertical asymptotesat x = 1andx = 2
Horizontal asymptotesoccur wheny tendstosomefinitevalueasx tendsto
Ex y =
1
x
hasahorizontal asymptoteat y = 0sinceasx , y 0
Ex y =
3x+2
x1
=
3+
2
x
1
1
x
3asx since
2
x
and
1
x
both0sohorizontal asymptoteisy = 3
Thenext techniqueistoinvestigatethebehaviour of thefunctionfor large
(
+ or
)
valuesof x
Ex. y =
x
2
3x+7
x+1
=
x(x+1)4(x+1)+11
x+1
= x 4+
11
x+1
but
11
x+1
0for largex soy x 4
Thisisinfact anexampleof anobliqueasymptote.
Notehowever that it ispossiblefor acurvetocrossanasymptoteat somefinitepoint.
Inparticular youshouldbeabletoeasilysketchaquadraticcurvebyexpressingthefuncionin
completedsquareform.
Ex Tosketchy = 3x
2
6x + 2wenotethat thismaybewrittenasy = 3
(
x 1
)
2
1
y obviouslyhasaminimumvalueof 1whenx = 1, sox = 1isanaxisof symmetryand(1,1) isthe
lowest point of thegraph.
To summarise, with more complex equations, some or all of the following techniques may be required
1. Inspect for symmetry about the axes.
An even function is symmetrical about the y-axis
An odd function has rotational symmetry of order two
If there are only even powers of y then we have symmetry about the x-axis
2. Check for asymptotes
If y =
f(x)
g(x)
theng
(
a
)
= 0 x = a isanasymptoteandthecurvehasadiscontinuityat x = a
If y k asx theny = k isanasymptote
3. Determine intersections with axes and asymptotes
4. Investigate stationary points
5. Determine any restrictions on the values of x and y by solving for y and x respectively
The following example illustrates these techniques
Ex. (i) y =
x(x+1)
x1
(1) There is no symmetry about either axis (note that negative information can be very useful)
(2) x = 1isclearlyanasymptote,. alsoit isclear that y asx sotherearenoasymptotes
parallel to the x-axis. It is however possible to investigate for other asymptotes as follows
y =
x(x+1)
x1
=
x
2
+x
x1
=
(x1)(x+2)+2
x1
= x +2+
2
x1
whichtendstox + 2asx
Hence, y = x + 2isanasymptote.
It is also helpful to see from which side the curve approaches the asymptote. This often identifies turning
points without the need for differentiation.
Consider the asymptote : if which will be a large x = 1 x = 1 h whereh issmall, theny =
(1h)(2h)
h
negative number. i.e. as x 1fromtheleft, y
12
Similarly, as Again, considering we can see that for x 1fromtheright y + y = x + 2 x > 0, y > x + 2
since the largest neglected term is positive, and so the curve tends to this line from above if
x > 0andsimilarlyfrombelowif x < 0
(3) x = 0 y = 0but alsoy = 0whenx = 1
You should now be able to draw a very good sketch of the graph. Try for yourself and then compare with
the graph below.
(4) It is difficult to differentiate and in view of what we already know it is not necessary.
(5) Again it is not really necessary to look for restrictions but we will do it as an example of the
technique.
It is obvious that y existsfor all valuesof x except x = 1
Solving for x in terms of y we have xy y = x
2
+ x x
2
+ (1 y)x + y = 0
x =
y1 (1y)
2
4y
2
=
1
2
y 1 1 6y + y
2
Clearly, the curve can only exist for values of y such that y
2
6y + 1 0 (y 3)
2
8
i.e. for y 3+ 8 or y 3 8
y = x + 2
13
1 0 1
2
Further Methods for Advanced Mathematics FP2
Polar Coordinates
You are familiar with the representation of a point in a plane by cartesian coordinates (x,y)
The position of a point P is also given by its distance r from the origin and the angle between the x axis 0
and the line OP. These are the POLAR COORDINATES of P and are defined with respect to an origin
O, called the POLE and a fixed line Ox, called the INITIAL LINE.
The relation between the cartesian and polar coordinates of a point P should be obvious. (x,y) (r,0)
. x = r cos0, andy = r sin0, r
2
= x
2
+ y
2
hence, cos0 = x/r, sin0 = y/r, tan0 = y/x andr = x
2
+ y
2
Note that it is conventional to take . When sketching a curve, given by a polar equation r 0 r = f(0)
negative values of f may occur for some values of . In such cases we use the corresponding positive (0) 0
value of f . This is perfectly valid since the same point (0) andaddv to(or subtract v from) thevalueof 0
is given by for any integer value of n, leaves the position (r,0) and(r, 0 + v). Theadditionof 2nv to0,
of the point unaltered and so we choose a principal value of . 0 intherangev < 0 v
The polar coordinates of a point P may be compared with the modulus and argument of the
complex number represented by the point P in the complex plane.
To sketch a curve from a given polar equation we either work out values of r corresponding to simple
v
al
u
es of 0 suchasmultiplesof
v
6
or
v
4
or byconvertingtoacartesianequation.
Ex. Sketch the curve r = 2asin0
Method 1. We draw up a table of values.

Clearly for values of so describing the curve 0 fromv to2v r will takethecorrespondingnegativevalues
twice.
Method 2. which we may recognise to be the r = 2asin0 x
2
+ y
2
=
2ay
x
2
+y
2
x
2
+y
2
2ay = 0
equation of a circle centre (0,a) andradiusa.
Sometimes we can do it by simple reasoning rather than working out values. In the case above we could
argue as follows:
As 0 increasesfrom0tov/2, r increasesfrom0to2a.
as 0 increasesfromv/2tov, r decreasesfrom2a to0
This certainly tells us that we have a closed curve and one or two actual points should convince us that it
is a circle.
Unless the curve retraces itself when r < 0 as in the case above, it is usual to draw such portion with a
dotted line.
14
0 a a 2 a 3 2a a 3 a 2 a 0 r
v 5v/6 3v/4 2v/3 v/2 v/3 v/4 v/6 0 0
2a
Areas in Polar Coordinates
Consider a curve with the polar equation and let P and P' be adjacent points on the curve r =f(0)
with polar coordinates The curve being such that both are increasing as (r,0) and(r + ^r,0 + ^0). r and0
we move from P toP
'
.
The area is approximated by the area Â swept out bytheradiusvector inmovingfromOP toOP
'
of the triangle . Ignoring the product of OPP
'
i.e. Â ~
1
2
r(r + ^r)^0 sincesin^0 ~ ^0 for small angles
small quantities this gives and in the limit as Â ~
1
2
r
2
^0 P P
'
andboth^r and^0 0
dA
d0
=
1
2
r
2
andso, areaswept out as0 increasesfromc to is
]
c
1
2
r
2
d0
Ex. Find the area of the region enclosed by the curve r = 3+ 2sin0
We first sketch the curve.
r has a maximum value of 5 when and a 0 =
1
2
v
minimum of 1 when and when 0 =
3
2
v r = 3 0 = 0or v
so we have a kidney shaped curve as shown.
Area
=
]
0
2v
1
2
r
2
d0 = 2
1
2
]
0
v
(9+ 12sin0 + 4sin
2
0)d0
makinguseof symmetry.
=
]
0
v
(9+ 12sin0 + 2 2cos20)d0
=
[
110 12cos0 sin20
]
0
v
= 11v
15
u
ou
r
r + r
P
P'
O
3 3
1
5
0
The Inverse Trigonometric Functions
You will remember that for a function to have an inverse it must be one to one, hence, if the
trigonometric functions are to have inverse functions defined, we must restrict their domains so that they
become one to one functions. The usual restrictions are:
for for and for sinx
1
2
v x
1
2
v cosx 0 x v tanx
1
2
v < x <
1
2
v
Note the strict inequality in the last case which is necessary since is not defined for tanx x =
1
2
v
The inverse functions are denoted by the prefix 'arc', e.g. arcsin, arccos, arctan etc. or by using
the index 1, e.g. sin
1
, cos
1
, tan
1
etc. wethushavesin
1
x = y siny = x for 1 x 1
cos
1
x = y cosy = x for 1 x 1andtan
1
x = y tany = x for all real x
The graphs of the inverse trigonometric functions are easily obtained by reflecting the graphs of the
trigonometric functions in the line providing equal scales are used on both axes. y = x
y = sin
1
x y = cos
1
x y = tan
1
x
Integration using inverse functions
We first consider the derivatives of the inverse trigonometric functions.
y arcsinx x siny 1 cosy
dy
dx
(implicit differentiationwithrespect tox)
so
dy
dx

1
cosy

1
1sin
2
y

1
1x
2
but from the graph of arcsinx wecanseethat it hasapositivegradient throughout itsdomain
hence we take only the positive root in the denominator to give
d
dx
(arcsinx)
1
1x
2
Alternative justification for taking only the positive root is that y arcsinx
v
2
y
v
2
andsocosx
must be positive.
In a similar manner y arccosx cosy x
dy
dx

1
siny

1
1x
2
This time we see that arccosy has a negative gradient throughout
Finally y arctanx tany x
dy
dx

1
sec
2
y

1
1+tan
2
y

1
1+x
2
Reversing these results we have
]
1
1x
2
dx arcsinx + c, and
]
1
1+x
2
dx arctanx + c
More generally by making the substitution we can show that x asinx
]
1
a
2
x
2
dx arcsin
(
x
a
)
+ c
and putting x a tanx wehave
]
1
a
2
+x
2
dx
1
a
arctan
(
x
a
)
+ c
Ex. Differentiate (i) sin
1
3x (ii) tan
1
x
(i) (function of a function)
d
dx
(
arcsin3x
)
3
1
1(3x)
2

3
19x
2
(ii)
d
dx
(arctan x )
1
2 x

1
1+x

1
2(1+x)) x
16

0
1
0 1
0
1
1
2
t t
t
2
2
2
t
t
Ex. Evaluate (i)
]
0
2
d0
4+0
2
(ii)
]
1/3
2/3
dx
49x
2
(i)
]
0
2
d0
4+0
2
=
1
2
tan
1
0
2 0
2
=
1
2
[tan
1
1 tan
1
0] =
1
8
v
(ii)
]
1/3
2/3
dx
49x
2

1
3
]
1/3
2/3
dx
4
9
x
2

1
3
arcsin
3x
2 1/3
2/3
1
3
sin
1
1 sin
1
1
2

1
3
v
2

v
6

v
9
Harder Integrals
Integrals of functions of the form where
]
P(x)dx
ax
2
+bx+c
P(x) isapolynomial inx
If P(x) isof equal or higher degreethanthedenominator thenweperformdivisionuntil wehavea
remainder of lower degree so we need only consider that case. So suppose P(x) = Ax + B
If logartithmic A = 2a andB = b thenthenumerator isthederivativeof thedenominator andwehavea
integral. Otherwise we write giving a logarithmic integral and an integral of the P(x) = k(2ax + b) k
'
form . We therefore need only consider
]
k
'
ax
2
+bx+c
dx
]
1
ax
2
+bx+c
dx
We do this by completing the square in the denominator to produce one of the following forms.
The first and third of these may now be done by using partial fractions to
]
dx
X
2
A
2 ]
dx
X
2
+A
2
or
]
dx
A
2
X
2
produce two logarithmic integrals whilst the second case gives an inverse tangent .
Integrals of the form Again, by similar techniques as before we can reduce the problem to ]
P(x)dx
ax
2
+bx+c
two types of integral.
and again, by completing the square inside the
]
2ax+b
ax
2
+bx+c
dx 2 ax
2
+ bx + c + k and
]
dx
ax
2
+bx+c
square root we obtain one of the following forms:
]
dx
A
2
X
2
,
]
dx
A
2
+X
2
or
]
dx
X
2
A
2
The first of these gives an inverse sine result. For the second by substituting we have X = Asec0

dx
d0
= Asec0tan0 sotheintegral becomes
]
Asec0 tan0d0
Atan0

]
sec0d0 lnsec0 + tan0 + k
= ln
X
A
+
X
2
A
2
A
+ k ln X + X
2
A
2
+ k
'
The third case is dealt with later in this book.
17
Maclaurin Series
Suppose a function is expressible as a polynomial in f(x) x of degreen
i.e. are all determined if the value f(x) = a
0
+ a
1
x + a
2
x
2
+ ... + a
n
x
n
, thenthecoefficientsa
i
(i = 0ton)
of , for we then have a set of n + 1 f(x) andall of itsderivativesareknownfor ssomeparticular valueof x
equations in the which may be solved simultaneously. This is especially n + 1unknownsa
0
, a
1
, a
2
, ..., a
n
easy if the particular value of x is zero for then we immediately have
f(0) = a
0
, f
'
(x) = a
1
+ 2a
2
x + 3a
3
x
2
+ ... + na
n
x
n1
f
'
(0) = a
1
f
''
(x) = 2a
2
+ 2.3a
3
x + 3.4a
4
x
2
+ ... + (n 1)na
n
x
n2
f
''
(0) = 2a
2
and in general f
(r)
(0) = r!a
r
a
r
=
f
(r)(0)
r!
thus f(x) =f(0) +f
'
(0)x +
f
''
(0)
2!
x
2
= ... +
f
(r)
(0)
r!
x
r
+ ... +
f
(n)
(0)
n!
x
n
To prove that this technique can be extended to an infinite series is beyond the scope of this
course but we will accept it as true providing the infinite series converges to a finite limit. The infinite
expansion of a function obtained in this way is called the MACLAURIN SERIES for the function.
Ex. Find the Maclaurin series for if n is negative or fractional. (1+ x)
n
Assume (1+ x)
n
= a
0
+ a
1
x + a
2
x
2
+ ... + a
n
x
n
+ ...
Writing f(x) = (1+ x)
n
wehavef(0) = 1
f
'
(x) = n(1+ x)
n1
f
'
(0) = n
f
''
(x) = (n 1)n(1+ x)
n2
f
''
(0) = n(n 1)
and, in general f
(r)
(x) = n(n 1)(n 2)...(n r + 1)(1+ x)
nr
f
(r)
(0) = n(n 1)(n 2)...(m r + 1)
and so the Maclaurin expansion of (1+ x)
n
isgivenby
providing this series is (1+ x)
n
= 1+ nx +
n(n1)
2
x
2
+
n(n1)(n2)
3!
x
3
+ ... +
n(n1)...(nr+1)
r!
x
r
+ ...
convergent.
Investigation of convergence is outside our syllabus but it is interesting to consider one standard
test that will suffice for most of the series we shall meet.
First there is the fairly obvious requirement that the terms must tend to zero eventually,
i.e. .
r
lim u
r
= 0, whereu
r
denotesther -theterm
This however is not in itself a sufficient condition for convergence as can be seen by considering the
series 1+
1
2
+
1
3
+
1
4
+ ... +
1
r
whichdoesnot converge.
If however , we also have
r
lim
u
r+1
u
r
< 1thenconvergenceisguaranteed.
Applying this to our series for (1+ x)
n
wehave
u
r+1
u
r
=
(nr+1)x
r
x asr
The condition for convergence is thus satisfied if and clearly we also have x < 1 u
r
0
Note that if n is a positive integer then the series terminates at the term and is identical to the (n +1)
th
series we have met earlier. In this case of course, there is no restriction on the value of x.
Ex. Find the first four terms in the expansions of (a) (1+ x)
1
(b) (1+ x)
2
(c) (1+ x)
1/2
(d) 1
x
2
1/2
(e) state also the range values of x for which your series is valid. (3+ 2x)
1
(a) for (1+ x)
1
= 1+ (1)x +
(1)(2)
2!
x
2
+
(1)(2)(3)
3!
x
3
+ .. = 1 x + x
2
x
3
+ ... x < 1
(b) (1+ x)
2
= 1+ (2)x +
(2)(3)
2!
x
2
+
(2)(3)(4)
3!
x
3
+ ... = 1 2x + 3x
2
4x
3
+ ... for x < 1
(c) (1+ x)
1/2
= 1+
1
2
x +
1
2

1
2
2!
x
2
+
1
2

1
2

3
2
3!
x
3
+ ... = 1+
1
2
x
1
8
x
2
+
1
16
x
3
+ ... for x < 1
(d) 1
x
2
1/2
= 1+
1
2

x
2
+
1
2

1
2
2!

x
2
2
+
1
2

1
2

3
2
3!

x
2
3
+ ... = 1
1
4
x
1
32
x
2
1
128
x
3
+ .
for x < 2
(e) (3+ 2x)
1
= 3
1
1+
2x
3
1
=
1
3
1+
(
1
)
2x
3
+
(1)(2)
2!
2x
3
2
+
(1)(2)(3)
3!
2x
3
3
+ ...
=
1
3

2
9
x +
4
27
x
2
8
81
x
3
+ ... for x <
3
2
18
Note especially the method of the last example.
In general, (a + bx)
n
= a
n
1+
bx
a
n
if a = 1andisvalidfor x <
a
b
The binomial series is also very useful for estimation.
Ex. Estimate the value of
3
997 givingyour answer correct to6decimal places
3
997 =
(
1000 3
)
1/3
= 10
(
1 0.003
)
1/3
= 10 1+
1
3
(
0/003
)
+
1
3

2
3
2!
(
0.003
)
2
+ ...
= 10(1 0.001+ 0.000001 ...
It is clear that the next term will not affect the 6th decimal place so
3
997 ~ 10 0.9989990= 9.989990
The following expansions are especially important and should be memorised
(1+ x)
1
1 x + x
2
x
3
+ ... (1 x)
1
1+ x + x
2
+ x
3
+ ...
(1+ x)
2
1 2x + 3x
2
4x
3
+ ...
What can we do with if x does not lie between ? (1+ x)
n
1
Consider and since x is large, must +(1+ x)
3
whenx islarge. Wemaywrite(1+ x)
3
= x
3
1+
1
x
3
1
x
be small and so we have for
1
x
3
1
3
x
+
6
x
2

10
x
3
+ ... =
1
x
3

3
x
4
+
6
x
5

10
x
6
+ ... x > 1
i.e. we have an expansion in descending powers of x
Ex. Express
x
(1x)(2x)
asaseriesof (i) ascending(ii) descendingpowersof x
(i) We use partial fractions
x
(1x)(2x)
=
1
1x

2
2x
= (1 x)
1
2(2 x)
1
now (1 x)
1
= 1+ x + x
2
+ x
3
+ ... and(2 x)
1
=
1
2
1
x
2
1
=
1
2
1+
x
2
+
x
2
4
+
x
3
8
+ ...
hence,
x
(1x)(2x)
=
(
1+ x + x
2
+ x
3
+ .
)
1+
x
2
+
x
2
4
+
x
3
8
+ . =
x
2
+
3x
2
4
+
7x
3
8
+ ...
(ii) For large x, (1 x)
1
=
1
x
1
1
x
1
=
1
x
1+
1
x
+
1
x
2
+
1
x
3
+ ... =
1
x

1
x
2

1
x
3

1
x
4
...
and(2 x)
1
=
1
x
1
2
x
1
=
1
x
1+
2
x
+
4
x
2
+
8
x
3
=
1
x

2
x
2

4
x
3

8
x
4
...
so
x
(1x)(2x)
=
1
x

1
x
2

1
x
3

1
x
4
... 2
1
x

2
x
2

4
x
3

8
x
4
... =
1
x
+
3
x
2
+
7
x
3
+
15
x
4
+ ...
Other important standard results. These should be committed to memory.
e
x
= 1+ x +
x
2
2!
+
x
3
3!
+ ... +
x
r
r!
+ ... validfor all valuesof x.
sinx = x
x
3
3!
+
x
5
5!

x
7
7!
+ ... +
(
1
)
r x
2r+1
(2r+1)!
+ ... validfor all valuesof x
cosx = 1
x
2
2!
+
x
4
4!

x
6
6!
+ ... +
(
1
)
r x
2r
(2r)!
+ ... validfor all valuesof x
ln
(
1+ x)
)
= x
x
2
2
+
x
3
3
... +
(
1
)
r+1x
r
r
+ validonlyfor 1< x 1
Noteparticularlythelast oneasit istheonlyonewithout factorial denominatorsandtheonlyoneother
thanthebinomial seriesthat hasarestrictedrangeof valuesof x.
19
Complex Numbers again
De Moivre's Theorem
We have previously established that (
cos0 + j sin0
)(
cos,+ j sin,
)
= cos
(
0 + ,
)
+ j sin
(
0 + ,
)
hence, if 0 = , then(cos0 + j sin0)
2
= cos20 + j sin20
Similarly (cos0 + j sin0)
3
= (cos0 + j sin0)
2
(cos0 + j sin0) = cos30 + j sin30
which suggests the general result
(
cos0 + j sin0
)
n
= cosn0 + j sinn0
This result is known as DE MOIVRE'S THEOREM after Abraham de Moivre (1667-1754)
Formal proof is left as an exercise.
Ex. If z = cos0 + j sin0, findthevalueof z
5
+ z
5
We have z
5
=
(
cos0 + j sin0
)
5
= cos50 + j sin50
and z
5
= (z
1
)
5
= (cos(0) + j sin(0))
5
= cos(50) + j sin(50) = cos50 j sin50
Hence, z
5
+ z
5
= 2cos50
Ex. Evaluate cos
v
12
+ j sin
v
12
4
By de Moivre's theorem cos
v
12
+ j sin
v
12
4
= cos
v
3
+ j sin
v
3
=
1
2
+ j
3
2
Ex Express cos40 intermsof cos0
cos40 = Re(cos40 + j sin40) = Re(cos0 + j sin0)
4
= Re
(
cos
4
0 + 4j cos
3
0 sin0 + 6j
2
cos
2
0sin
2
0 + 4j
3
cos0sin
3
0 + j
4
sin
4
0
)
= cos
4
0 6cos
2
0 sin
2
0 + sin
4
0 = cos
4
0 6cos
2
0(1 cos
2
0) + (1 cos
2
0)
2
= cos
4
0 6cos
2
0 + 6cos
4
0 + 1 2cos
2
0 + cos
4
0
= 8cos
4
0 8cos
2
0 + 1
Can we form a complex power of a real number? In particular, what meaning can we attach to if z is e
z
complex?
First let us consider a purely imaginary power We may assign a meaning to this by simply requiring e
j0
that it has a series expansion, thus:
e
j0
= 1+j0 +
1
2!
(j0)
2
+
1
3!
(j0)
3
+ ... = 1+j0
1
2
0
2
1
3!
j0
3
+
1
4!
0
4
+ ...
= 1
0
2
2!
+
0
4
4!
... +j 0
0
3
3!
+
0
5
5!
... = cos0 +jsin0
Hence, for the general complex power we have so is the e
z
=e
x+jy
= e
x
e
jy
= e
x
(cosy +jsiny) e
x
modulus and y is the argument of the complex number e
z
The n-th roots of unity
cos
p
q
0 + j sin
p
q
0
q
= cosp0 +jsinp0 =
(
cos0 + j sin0
)
p
so cos
p
q
0 +jsin
p
q
0 isaq -throot of
(
cos0 + j sin0
)
p
i.e. avalueof
(
cos0 + j sin0
)
p
q
thus, for rational n, cosn0 +jsinn0 isonevalueof
(
cos0 + j sin0
)
n
The reason why we only say "one value of" will soon be apparent.
z
n
= 1
(
cos0 + j sin0
)
n
= 1wherez = cos0 +jsin0 cosn0 +jsinn0 = 1 n0 = 2kv for k = 0,1, 2,...
Thus for integral values of are the distinct n-th roots of unity. z = cos
2kv
n
+jsin
2kv
n
k from0ton 1
Taking . k > n 1merelygivesarepetitionof anearlier root
Ex. Solve z
3
1= 0
By the result just obtained the solutions are the cube roots of unity
i.e. z = cos0+jsin0, cos
2v
3
+jsin
2v
3
, cos
4v
3
+jsin
4v
3
or 1,
1
2
(
1 j 3
)
Ex Solve z
4
+ z
3
+ z
2
+ z + 1= 0
Since it follows that the required solutions are the complex fifth roots (z 1)(z
4
+ z
3
+ z
2
+ z + 1) = z
5
1
of unity, i.e. z = cos
2v
5
+jsin
2v
5
,cos
4v
5
+jsin
4v
5
, cos
6v
5
+jsin
6v
5
, cos
8v
5
+jsin
8v
5
20
Again note that if we write c = cos
2v
5
+jsin
2v
5
, thentheother rootsarec
2
, c
3
andc
4
We may also note that since c
r
c
nr
= c
n
= 1, thecomplexrootsoccur inconjugatepairs.
The equation z
n
=c
Let c be any complex number. If u is any one root of then we may write z
n
= c
(
z
u
)
n
= 1
Hence, all roots are given by u cos
2kv
n
+ j sin
2kv
n
for k = 0,1,2, ..., n 1
Now write If -th root of r then c intheformr
(
cos0 + j sin0
)
wherer = c r
1
n
denotesthereal n
r
1
n
cos
0
n
+ j sin
0
n
isann-throot of c
Taking u to be this root, the n-th roots of c are given by r
1
n
cos
0+2kv
n
+ j sin
0+2kv
n
for k = 0ton 1
Ex. Solve the equation z
5
= 32j
r = 32hencer
1
5
= 2j= cos
v
2
+jsin
v
2
hencez = 2 cos
v
2
+2kv
5
+ j sin
v
2
+2kv
5
for k = 0to4
i.e. and z = 2 cos
v
10
+ j sin
v
10
, 2 cos
v
2
+ j sin
v
2
, 2 cos
9v
10
+ j sin
9v
10
, 2 cos
13v
10
+ j sin
13v
10
2 cos
17v
10
+ j sin
17v
10
Application of de Moivre's theorem to series
Certain types of trigonometric series may be summed by making use of de Moivre's theorem.
Ex. Prove that
E
r=1
n
cosr0 =
cos
1
2
(n+1)0 sin
1
2
n0
sin
1
2
0
for 0 = 2kv, andfindanexpressionfor
E
r=1
n
sinr0
Putting z = cos0 +jsin0 thenz
r
= cosr0 +jsinr0
so writing
C +jS =
E
r=1
n
z
r
we have C =
E
r=1
n
cosr0 and S =
E
r=1
n
sinr0
Now by the sum of a geometric series C+jS =
z
(
1z
n
)
1z
and 1 z
n
= 1 cosn0 jsinn0 = 2sin
2
n0
2
2jsin
n0
2
cos
n0
2
= 2jsin
n0
2
cos
n0
2
+ j sin
n0
2
so putting n = 1wehave1 z = 2sin
0
2
cos
0
2
+ j sin
0
2
Hence C+jS =
(cos0+j sin0) 2j sin
n0
2
cos
n0
2
+j sin
n0
2
2j sin
0
2
cos
0
2
+j sin
0
2
soequatingreal andimaginaryparts
C =
E
r=1
n
cosr0 =
cos
1
2
(n+1)0 sin
1
2
n0
sin
1
2
0
andS =
E
r=1
n
sinr0 =
sin
1
2
(n+1)0 sin
1
2
n0
sin
1
2
0
21
Complex Numbers in Geometry
Simple Loci
By simple analogy with vector geometry we can see that is the distance between the points z
1
z
2
representing Thus, if represent the complex constants and z
1
andz
2
intheArganddiagram. A, B andP a
b andthecomplexvariablez then
(i) z a = k wherek isreal, definesacirclecentreA, radiusAP = k
(ii) z a = z b definestheperpendicular bisector of thelineAB
(iii) z a = k z b withk = 1isacirclewithcentredividingAB externallyintheratiok
2
: 1
(iv) , proceeding in the arg(z a) = 0 with0 real definesthehalf-linefromA, excludingthepoint A itself
direction 0 measuredfromthex-axis
(v) defines a circular arc from < measured from i.e. arg
za
zb
= 0 A toB suchthat APB = 0 BP toAP 0
is the angle of AP tothex-axisminustheangleof BP tothex-axis.
We prove the least obvious one which is (iii)
Let z = x +jy, a = a
1
+ja
2
andb = b
1
+jb
2
then x a
1
+ j(y a
2
) = k x b
1
+ j
(
y b
2
)
(x a
1
)
2
+ (y a
2
)
2
= k
2
(x b
1
)
2
(y b
2
)
2
x
2
+ y
2
2a
1
x 2a
2
y + a
1
2
+ a
2
2
= k
2
x
2
+ y
2
2b
1
x 2b
2
y + b
1
2
+ b
2
2
(k
2
1)(x
2
+ y
2
) + 2(a
1
k
2
b
1
)x + 2(a
2
k
2
b
2
)y + k
2
b
1
2
+ b
2
2
a
1
2
+ a
2
2
= 0
dividing by (since gives an equation of the form which we (k
2
1) k = 1) x
2
+y
2
+2gx +2fy +c = 0
know to be the equation of a circle with centre at the point which is the
k
2
b
1
a
1
k
2
1
+j
k
2
b
2
b
1
k
2
1
=
k
2
ba
k
2
1
point dividing AB externallyintheratiok
2
: 1
22
Matrices Again
Inverse of a 3 x 3 matrix
Consider M . =
a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3
andlet = det M = a.(b c) = 0
Let L be a matrix such that LM =I then Mr =i LMr = Li r = Li (thefirst columnof L)
Similarly Lj and Lk are the second and third columns of L respectively.
M
x
y
z
= i x =
i.(b c)
a.(b c)
=
A
1
, y =
B
1
, z =
C
1
sothefirst columnof L is
1
A
1
B
1
C
1
Replacing i by j and then k in the above then leads to L
=
1
A
1
A
2
A
3
B
1
B
2
B
3
C
1
C
2
C
3
The matrix
A
1
A
2
A
3
B
1
B
2
B
3
C
1
C
2
C
3
istheADJUGATE or ADJOINT of Mandisformedbyreplacingeach
element of M by its cofactor and then transposing (i.e. Changing rows into columns and v.v.)
It is denoted by adjM, hence, the inverse of M, M
1
=
1
adjM
Ex Find the inverse of the matrix M =
4 5 3
3 3 4
5 4 6
We first calculate the cofactors thus:
A
1
=
3 4
4 6
= 18+ 16= 2, A
2
=
5 3
4 6
= 30 12= 18, A
3
=
5 3
3 4
= 20 9= 11
Similarly B
1
=
(
18+20
)
= 2, B
2
= 2415= 39, B
3
=
(
169
)
= 25
and C
1
= 1215= 3, C
2
=
(
16+25
)
= 41, C
3
= 12+15= 27
hence, = 4
(
2
)
+ 3
(
18
)
+ 5 11= 7
adjM =
2 18 11
2 39 25
3 41 27
soM
1
=
1
7
2 18 11
2 39 25
3 41 27
Eigenvalues and eigenvectors
Definition: If s isanon-zerovector suchthat Ms = zs, whereMisamatrixandz isascalar, thens is
calledanEIGENVECTOR of Mandthescalar z isthecorrespondingEIGENVALUE
To find eigenvectors we need to solve the equation Ms = zs
Now Ms = zs Ms zs = 0 Ms zIs = 0
(
M zI
)
s = 0
For non-zero solutions we must have det
(
M zI
)
= 0. ThisisknownastheCHARACTERISTIC
. Det expands to give a polynomial in EQUATION of M
(
M zI
)
z, theCHARACTERISTIC
POLYNOMIAL. (Note! Eigenvalues and eigenvectors are also known as characteristic values and
vectors)
Ex Find the eigenvalues and corresponding eigenvectors of the matrix M=
1
2
3
2
3
2

1
2
We form the characteristic equation det i.e.
(
M zI
)
= 0
1
2
z
1
2
z
3
4
= 0
23
are the eigenvalues.
1
4
+z
2
3
4
=0z
2
=1soz =1
To find the corresponding eigenvectors we proceed as follows, solving (
M zI
)
s = 0
If s =
x
y
istheeigenvector correspondingtoz = 1thenwerequire

1
2
3
2
3
2

3
2
x
y
= 0
i.e. -
1
2
x +
3
2
y = 0and
3
2
x
3
2
y = 0 x = 3y and 3x = 3y whichisthesameasx = 3y
so the eigenvector is a vector along the line x = 3y for examplethevector
3
1
and is a POINTWISE INVARIANT LINE. x = 3y
z = 1
3
2
3
2
3
2
1
2
x
y
= 0
3
2
x +
3
2
y = 0and
3
2
x +
1
2
y = 0
i.e. 3x + 3y = 0and 3x + y = 0sotheother invariant lineisy = 3x witheigenvector
1
3
3 by 3 matrices are dealt with in a similar manner.
Ex Find the eigenvalues and corresponding eigenvectors of the matrix M =
3 0 2
1 2 1
0 0 1
The characteristic equation is
3 z 0 2
1 2 z 1
0 0 1 z
= 0
i.e.
(
3 z
)(
2 z
)(
1 z
)
= 0soz = 1,2or 3
z = 1
4 0 2
1 3 1
0 0 0
x
y
z
= 0
4x + 2z = 0
x + 3y + z = 0
soz = 2x andx = 3y, soeigenvector isr =
3
1
6
z = 2
1 0 2
1 0 1
0 0 3
x
y
z
= 0
x + 2z = 0
x + z = 0
3z = 0
sox = z = 0, y = anysoeigenvector isr =
0
1
0
z = 3
0 0 2
1 1 1
0 0 4
x
y
z
= 0
2z = 0
x y + z = 0
4z = 0
soz = 0, x = y soeigenvector isr =
1
1
0
An important application of eigenvalues and eigenvectors is to express a square matrix M in the form
where D is a diagonal matrix making it very easy to obtain powers of a square matrix. PDP
1
Ex Consider the matrix M dealt with above. and hence find =
3 0 2
1 2 1
0 0 1
ExpressMintheformPDP
1
M
3
andM
4
Let i.e. a diagonal matrix with the eigenvalues as diagonal elements. D =
1 0 0
0 2 0
0 0 3
Take P=
3 0 1
1 1 1
6 0 0
i.e. columnsaretheeigenvectors. thendetP = 6soP
1
=
1
6
0 0 1
6 6 2
6 0 3
24
PDP
1
=
1
6
3 0 1
1 1 1
6 0 0
1 0 0
0 2 0
0 0 3
0 0 1
6 6 2
6 0 3
=
1
6
3 0 1
1 1 1
6 0 0
0 0 1
12 12 4
18 0 9
=
3 0 2
1 2 1
0 0 1
= M
i.e. PDP
1
= M
Thus M
3
= (PDP
1
)(PDP
1
)(PDP
1
) = PD
3
P
1
=
1
6
3 0 1
1 1 1
6 0 0
1 0 0
0 8 0
0 0 27
0 0 1
6 6 2
6 0 3
i.e. M
3
=
1
6
3 0 1
1 1 1
6 0 0
0 0 1
48 48 16
162 0 81
=
27 0 14
19 8 11
0 0 1
andM
4
=
1
6
3 0 1
1 1 1
6 0 0
1 0 0
0 16 0
0 0 81
0 0 1
6 6 2
6 0 3
=
1
6
3 0 1
1 1 1
6 0 0
0 0 1
96 96 32
486 0 243
=
81 0 .40
65 16 35
0 0 1
The Cayley-Hamilton Theorem
Continuing with the above matrix, the characteristic equation was
(
z + 1
)(
z 2
)(
z 3
)
= 0
i.e. z
3
4z
2
+z+6= 0
M
2
=
9 0 4
5 4 3
0 0 1
andsoM
3
4M
2
+ M +6I =
27 0 14
19 8 11
0 0 1

36 0 16
20 16 12
0 0 4
+
3 0 2
1 2 1
0 0 1
+
6 0 0
0 6 0
0 0 6
i.e. M
3
4M
2
+ M +6I =
0 0 0
0 0 0
0 0 0
= 0
This illustrates the Cayley-Hamilton theorem which states that a square matrix always satisfies its own
characteristic equation.
The main use of this theorem is to find higher powers of a square matrix.
Example With M as above, find M
6
Solution
We have M
3
4M
2
+ M +6I = 0 M
3
= 4M
2
M 6I M
4
= 4M
3
M
2
6M
i.e. M
4
= 15M
2
10M 24I
M
5
= 15M
3
10M
2
24M = 50M
2
39M 90I
soM
6
= 50M
3
39M
2
90M = 161M
2
140M 300I
i.e. M
6
=
1449 0 644
805 644 483
0 0 161

420 0 280
140 280 140
0 0 140

300 0 0
0 300 0
0 0 300
=
729 0 364
665 64 343
0 0 1
25
Hyperbolic Functions
We define the hyperbolic cosine, sine and tangent by
; and coshx chx
1
2
(e
x
+ e
x
) sinhx shx
1
2
(e
x
e
x
)
tanhx thx
shx
chx

e
x
e
x
e
x
+e
x
also sechx =
1
coshx
; cosechx
1
sinhx
andcothx
1
tanhx
The following give some justification for the 'invention' of these functions.
(i) a number of integrals which otherwise cannot be obtained, are expressible in terms of them
(ii) A uniform chain hanging freely between two fixed points takes the form of a CATENARY with
equation y = k cosh
x
k
wherek isconstant, referredtosuitableaxes.
(iii) If x = acosh0 andy = bsinh0 then
x
2
a
2

y
2
b
2
= 1 cosh
2
0 sinh
2
0 = 1
1
4
(
e
0
+ e
0
)
2
1
4
(
e
0
e
0
)
2
= 1
so are parametric equations for the hyperbola x = acosh0 andy = bsinh0
x
2
a
2

y
2
b
2
= 1
This last result indicates the close analogy with the circular functions In fact sin0, cos0 andtan0 etc.
there is a hyperbolic identity corresponding to each trigonometric one.
We have already seen that cosh
2
0 sinh
2
0 = 1.
Consider now sinh
2
0 + cosh
2
0 =
1
4
(e
0
e
0
)
2
+
1
4
(e
0
+ e
0
)
2
=
1
2
(e
20
+ e
20
) = cosh20
Thus sin
2
0 + cos
2
0 = 1andcos
2
0 sin
2
0 = cos20
but sinh
2
0 + cosh
2
0 = cosh20 andcosh
2
0 sinh
2
0 = 1
OSBORN'S RULE enables us to write down any hyperbolic identity from the corresponding
trigonometric one: "Change cos to cosh, sin to sinh, and change the sign of any term involving the
product of two sines"
Thus sin(A + B) = sinAcosB + cosAsinB sinh(A + B) = sinhAcoshB + coshAsinhB
but cos(A + B) = cosAcosB sinAsinB cosh(A + B) = coshAcoshB + sinhAsinhB
and tan2A =
2tanA
1tan
2
A
translatestotanh2A =
2tanhA
1+tanh
2
A
Note that tan
2
A =
sin
2
A
cos
2
A
translatesto
sinh
2
A
cosh
2
A
= tanh
2
A
Ex. Write down and prove the hyperbolic identity corresponding to 1+tan
2
x = sec
2
x
By Osborn's rule the identity is 1 tanh
2
x = sech
2
x
Proof Left hand side = 1
e
x
e
x
e
x
+e
x
2
=
(e
x
+e
x
)
2
(e
x
e
x
)
2
(e
x
+e
x
)
2
=
4
(e
x
+e
x
)
2
=
1
cosh
2
x
= sech
2
x
Equations involving hyperbolic functions may be solved either by using similar methods to those for
trigonometric equations or by using the definitions of the hyperbolic functions as in the following
example.
Ex Solve the equation 2sinhx coshx = 1
2sinhx coshx 1 (e
x
x
)
1
2
(e
x
+ e
x
) 1e
x
3e
x
2e
2x
2e
x
3= 0
hence, (e
x
3)(e
x
+ 1) = 0but e
x
cannot benegativesoe
x
= 3istheonlysolution, i.e. x = ln3
Differentiation and Integration
and by the quotient rule
d
dx
(sinhx)
d
dx
1
2
(
e
x
e
x
)
1
2
(
e
x
+ e
x
)
= coshx, similarly
d
dx
coshx sinhx
with corresponding integral results.
d
dx
tanhx sech
2
x
26
The Inverse Functions
Consider first the graphs of sinhx, coshx andtanhx whicharereadilyobtainedfromthegraphsof e
x
and e
x
y = coshx
y = tanhx
y = sinhx
We can see from these graphs that sinhx isonetoonewithrangeanddomainbothequal to
We therefore define the inverse hyperbolic sine, such that sinh
1
x, for all valuesof x tobethat valueof y
x sinhy. Thusy sinh
1
x x sinhy
1
2
(e
y
e
y
) 2x e
y
e
y
e
2y
2xe
y
1 0e
y
x x
2
+1
Since we must have e
y
> 0it followsthat onlythepositiveroot isacceptableandso
sinh
1
x y ln x + x
2
+ 1 for x
coshx isnot onetooneunleswerestrict thedomain, theusual restrictionbeingtoallowonlyx > 0
We can then define cosh
1
x tobethepositivevalueof y suchthat x = coshy.
Thus y cosh
1
x x coshy
1
2
(
e
y
+ e
y
)
e
2y
2xe
y
+ 1 0e
y
x x
2
1
Now x x
2
1 x + x
2
1 x
2
(
x
2
1
)
1 x x
2
1 x + x
2
1
1
The domain of cosh
1
x is{x : x 1} andfor x > 1x + x
2
1 > 1 x x
2
1 < 1
and and so we cannot have e
y
< 1 y < 0. But fromour definitionof cosh
1
x wemust havey 0
e
y
= x x
2
1. Hencecosh
1
x = y = ln{x + x
2
1}
tanhx isclearlyonetoonefor all real x andsowedefinetanh
1
x tobethat valueof y suchthat
The logarithmic form for x tanhy for 1< x < 1. tanh
1
x isleft asanexercise.
Derivatives and Integrals
y sinh
1
x x sinhy
dx
dy
coshy
dy
dx

1
coshy

1
x
2
+1
Only the positive root need be considered since sinh
1
x hasapositivegradient for all valuesof x
Similarly (why only the positive root this time?) and
d
dx
cosh
1
x
1
x
2
1
d
dx
tanh
1
x
1
1x
2
Verify these results for yourself. You could be asked to derive them in an examination.
Hence,
]
dx
x
2
+1
sinh
1
x + c lnA x + x
2
+ 1
]
dx
x
2
1
cosh
1
x + c lnA x + x
2
1
and this one could also be done by partial fractions
]
dx
x
2
1
tanh
1
x + c
1
2
lnA
1+x
1x
What restrictions must be imposed on the values of x in these three integrals?
We can now complete our review of integrals of the form
]
dx
ax
2
+bx+c
and
]
dx
ax
2
+bx+c
27
1
1
0
0
1
2
3
4
1
2
3
By completing the square of the quadratic in the denominator we obtain one of the following:
]
dx
X
2
+A
2
=
1
A
arctan
X
A
;
]
dx
X
2
A
2
=
1
2A
lnk
X+A
XA
;
]
dx
A
2
X
2
=
1
2A
ln
AX
A+X
where X hastheformx b unlessax
2
+ bx + c < 0for all x inwhichcasetheintegral doesnot exist.
Note that a logarithmic form is usually preferred to an inverse hyperbolic tangent.
;
]
dx
)A
2
X
2
= sin
1
X
A
]
dx
A
2
+X
2
= sinh
1
X
A
;
]
dx
X
2
A
2
= cosh
1
X
A
28
Curve Sketching Again
Most of the techniques were dealt with earlier. The remaining ones are (i) curves with equations of the
form y
2
= f(x) and(ii) equationsof theformy = f(x)
(i) To sketch y
2
= f(x) westart bysketchingy = f(x) andthenplot thetwosquarerootsof eachordinate.
Ex. Sketch y
2
=
x(x+1)
x1
We first sketch y =
x(x+1)
x1
shownbyadottedlineinthediagrambelow
We note that x = 1isanasymptote.
There is no horizontal asymptote but writing the
equation as we see that y = x + 2+
2
x1
y = x + 2
is an asymptote. Considering how the curve
approaches these asymptotes it is quite easy to
sketch the curve.
Clearly, the curve can only exist for
1 x 0andx > 1
Between x = 1andx = 0wemust haveaclosed
loop. Note that, in general, if 0< y < 1then y > y
while if and the square root y > 1then y < y
curve must intersect the original curve at any point
where y = 0or 1
Ex. Sketch y
2
= x
2
(1 x)
This example is included to illustrate one final situation.. It is easy to sketch the graph of
y = x
2
(1 x) asit issimplyacubic, touchingthex axisat theoriginandpassingthrough(1,0)
The square root curve obviously only exists for x 1andweagainmust havealoopfor 0 x 1
The curves y = x
2
(1 x) andy
2
= x
2
(1 x) alsointersect wherey = 1, but howdoesthesquareroot
curve pass through the origin and how does it go off to infinity.
At the origin there are really just three possibilities
Since this is an important feature of the graph we consider how the curve behaves for small values of x
We have
y
2
= x
2
x
3
y
2
~ x
2
for small x sincethex
3
termmaybeneglectedasbeingverysmall compared
so the middle diagram must be the correct one. withx
2
y ~ x for verysmall valuesof x
To deduce how the curve goes off to infinity consider y = x
n
(i) If n = 1wehaveastraight line(ii) if n > 1thelinewill get steeper asx increases
(iii) if 0< n < 1thecurvewill get flatter asxincreases
In the above example, since the dominant term for large x isx
3
it followsthat thecurvewill get steeper
as x but not toorapidly. Canyousketchthefinal curve?
29
y
x
1 0 1
Graphof y = f
(
x
)
This is simply a matter of sketching and reflecting any part of the graph that is below the x-axis y = f
(
x
)
in that axis.
Inequalities
We encountered the solution of inequalities earlier . An alternative method is to use a sketch graph as
follows.
Ex. Solve the inequality
2x1
x1

9
x+1
Solution: This is equivalent to Note that since we do not know (2x 1)(x + 1)
2
(x 1) 9(x 1)
2
(x + 1)
whether the denominators are positive or negative we must multiply by the squares of the denominators
to ensure that the inequality sign remains unchanged.
Thus we have (2x 1)(x + 1)
2
(x 1) 9(x 1)
2
(x + 1) 0 (x + 1)(x 1)|(2x 1)(x + 1) 9(x 1)|
i.e. so we plot (x + 1)(x 1)(2x
2
8x + 8) 0 2(x + 1)(x 1)(x 2)
2
0 y = 2(x + 1)(x 1)(x 2)
2
This is easily sketched since it crosses the x- axis where x = 1and1andtouchestheaxisat x = 2
Clearly also, y ispositivefor largepositiveor negativevaluesof x sincethedominant termisx
4
Since we require y 0wecaneasilyseethat thesolutionis1 x 1or x = 2
You should also be aware of the standard cartesian and parametric equations of the conics which are as
follows:
Parabola: y
2
= 4ax, x = at
2,
y = 2at for aparabolasymmetrical about x axispassingthroughorigin.
Ellipse:
x
2
a
2
+
y
2
b
2
= 1, x = a cos0, y = bsin0 for symmetricallyplacedellipse.
Hyperbola:
x
2
a
2

y
2
b
2
= 1, x = asec0, y = btan0 or x = a cosh0, y = bsinh0
Rectangular Hyperbola: xy = c
2
, x = ct, y =
c
t
30
x
y
1 0 1 2
Further Applications of Advanced Mathematics FP3
Further Vector Geometry - The Vector Product
The vector product of two vectors a and b at an angle to each other, is defined to be the vector of ab 0
magnitude in a direction perpendicular to both a and b such that a, b and a b form a absin0
right-handed set of vectors. i.e. An ordinary (right- handed) screw rotated from the direction of a to the
direction of b would move in the direction of a b.
Thus where is a unit vector in the direction of a b ab= (ab sin0)n n
An immediate consequence of the definition is that a b = b a so vector multiplication is
anti-commutative.
a b = 0 In particular a = 0 or b = 0 or a isparallel tob. a a = 0hence, i i = j j =k k =0
whilst i j =k, j k =i, k i =j, j i = k, k j = i andk i = j
The distributive law can be shown to hold and so, in component form, with a (b + c) = a b +a c
usual notation,
a b = ( a
1
i + a
2
j +a
3
k) (b
1
i +b
2
j+b
3
k) = (a
2
b
3
b
2
a
3
)i +(a
3
b
1
a
1
b
3
)j +(a
1
b
2
a
2
b
1
)k
The following diagram is a convenient aid to memory The three triple products
a
1
a
2
a
3
a
1
a
2
b
1
b
2
b
3
b
1
b
2
i j k i j
sloping downwards from left to right are positive and the three sloping upwards from left to right are
negative.
Ex. Given , find the components of the vector in the directions of the p = i +j andq =2i j +2k p q
axes. Find also the magnitude of the resolved part of p q inthedirectionof i +j k
and the components are thus 2i, 2j and p q =(2 0)i +(0 2)j +(1 2)k = 2i 2j 3k
3k. The magnitude of the resolved part in the direction of
r is(p q).r = (2i 2j 3k).
1
3
(i + j k)
=
1
3
3= 3
Ex. Find a unit vector perpendicular to both a =2i +2j k andb =4i +j k
a b =(2+ 1)i +(4+ 2)j +(2 8)k = i 2j 6k
Thus are perpendicular to both a and b and so a unit vector is i 2j 6k andi +2j +6k
1
41
(i +2j +6k)
Applications of the vector product.
1. Line of intersection of two planes. The vector product of the normals to the planes gives us the
direction vector for the line of intersection. Finding the coordinates of any one common point of the
planes then enables us to write down the equation of the line of intersection.
Distance of a point from a line
Problem
To find the distance of the point P(x,y, z) fromthe
liner = a +zb
Let A be any point on the line.
Required distance is PN
The vector product gives APAN= AP AN sin0
soAP b
= APsin0 i.e. PN =
AP b
b
31
A
P
N
u
b
Special case. (2 dimensions)
Consider a point and a line in the plane z = 0
If equation of line is ax + by + c = 0andP isthepoint (x
1
,y
1
,0) thenwemaytakeA tobe 0,
c
b
,0
b=
b
a
0
andsoAP b
=
x
1
y
1
+
c
b
0

b
a
0
=
0
0
ax
1
+ by
1
+ c
hence,PN =
AP b
b
=
ax
1
+by
1
+c
a
2
+b
2
Triple Products.
There are two types of triple product.
Shortest distance between two skew lines (i.e. lines which do not intersect)
Let A and B be any points on each of the two lines
Shortest distance between the lines will be the length of the
common perpendicular PQ hence, since APQ and BQP
are right angles, PQ is the projection of AB on the common
perpendicular to the two lines.
The direction of PQ is given by where r and s are r s
the direction vectors of the two lines, hence,
PQ =
AB.(r s)
r s
Distance of a point from a plane
Problem
To find the distance of the point P from the plane
r = a +zb +uc
Solution
NP hasdirectiona b thenormal totheplane.

PN= APcos0 =
AP.PN
PN
=
(
p a
)
.
(
a b
)
a b
Note. is positive if P is on the same side
(
p a
)
.
(
a b
)
a b
of the plane as the origin and negative if on opposite side.
If the equation of the plane is ax + by + cz + d = 0andP isthepoint (x
1
,y
1
,z
1
) anormal totheplaneis
a
b
c
sotakingA tobethepoint
0
0
d
c
wehaveAP.PN =
x
1
y
1
z
1
+
d
c
.
a
b
c
= ax
1
+ by
1
+ cz
1
+ d
note similarity to result for distance of point from line in 2 dimensions. hence, PN =
ax
1
+by
1
+cz
1
+d
a
2
+b
2
+c
2
Scalar Triple Product
The second and third results above both involve an expression of the form a.b c
We saw there that it measures the volume of a parallelepiped with edges defined by a,b and c
and is calculated as for a 3 by 3 determinant. i.e. a.(b c)
=
a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3
32
P
Q
A
B
r
s
r s
P
N
A
u
Note that a cyclic change of the letters does not affect the value but a non-cyclic change multiplies the
product by 1.
Geometrical applications
An immediate application is to test whether two lines intersect. If they do then the shortest distance
between them is zero and hence the scalar triple product in the previous work is zero. Conversely, if the
scalar triple product is not zero then the lines do not intersect.
To test whether points A,B,C and D are coplanar.
If they are, then the parallelepiped defined by edges AB,AC and AD will have zero volume and so the
scalar triple product (
b a
)
.
[(
c a
)
(
d a
)]
= 0
Testing for right or left-handedness
, a.
(
b c
)
> 0 a, b andc formaright handedseta.
(
b c
)
< 0 a, b andc formaleft handedset
Remember that but for example a.
(
b c
)
=b.
(
c a
)
=c.
(
a b
)
a.
(
c b
)
= a.
(
b c
)
33
Multivariable calculus
There are many situations where we are dealing with a function of more than one variable.
e.g. thevolumeof acylinder isgivenbyV = vr
2
h whilst asurfacein3-dimensional spacewill havean
equationof theformz = f
(
x,y
)
If z = 2cosx y
2
definessuchasurface, lettingx takeafixedvalue0 say, wehavez = 2cos0 y
2
whichrepresentsacurveinthey,z plane, i.e. asectionof thesurfaceperpendicular tothex axis
whilst givingy afixedvaluewill produceasectionof thesurfaceperpendicular tothey axis.
If z hasafixedvaluewehaveasectionparallel tothex y planewhichisusuallycalledaCONTOUR.
Inthisexamplethesectionswithx constant will beinvertedparabolas, thosewithy constant will be
cosinecurveswhilst thecontoursz = constant will beof differenmt formsaccoprdingtothevalueof z.
Thediagramshowsafewcontoursfor thissurface.
z = 2givesaseriesof pointsat x = kv
0< z < 2givesaseriesof circlescenteredat x = kv
k = 2givesapair of cosinecurvesintersectingonthex -axisat x = (2k + 1)v
k < 2andthecurvesdonot intersect theaxis.
34
x
x x x x
2
0
t
t
t
t 2
x x
x
x x x
x x x x x
x
x x
x x
Partial Differentiation
Supposez = f
(
x, y
)
andlet x increasebyasmall amount ^x whiley remainsconstant, thentheincrease
inz isf
(
x + ^x,y
)
f
(
x,y
)
andtheaveragerateof changeof z withrespect tox is
f(x+^x,y)f(x,y)
^x
Thelimit of
f(x+^x,y)f(x,y)
^x
as^x 0, if it exists, iscalledthePARTIAL DERIVATIVE of z with
respect tox andisdenotedby
cz
cx
(Note! Becareful not toconfusethesymbolsc andd)
Similarly
cz
cy
isthepartial derivativeof z withrespect toy,x remainingconstant.
Ex Find
cz
cx
and
cz
cy
whenz = y
2
lnx
Treatingy asaconstant wehave
cz
cx
= y
2
1
x
=
y
2
x
treatingx asaconstant
cz
cy
= 2y lnx
Geometricallyspeaking,
cz
cx
tellsusthegradient of asectionof thesurfaceperpendicular tothey axis
. whilst
cz
cy
givesthegradient of asectionperpendicular tothex- axis
Tangent Planes
Consider a surface z = f
(
x, y
)
andsupposethat at thepoint x = a, y = b onthissurfacethevaluesof
cz
cx
and
cz
cy
arep andq respectivelythenthevectors
1
0
p
and
0
1
q
must lieintheplanetangential tothe
surfaceat thispoint, hence, anormal totheplaneisgivenby
1
0
p
0
1
q
=
p
q
1
andsothe
equationof thetangent planeispx qy + z = ap bq + f
(
a,b
)
Ex Find the equation of the tangent plane to the surface z =
x
3
y
2
at thepoingivenbyx = 4, y = 8
x = 4, y = 8 z = 1.
cz
cx
=
3x
2
y
2
=
48
64
=
3
4
at (4, 8,1),
cz
cy
=
2x
3
y
3
=
128
512
=
1
4
Sotwovectorsinthetangent planeare
1
0
3
4
and
0
1
1
4
A normal vector is
1
0
3
4
0
1
1
4
=

3
4
1
4
1
sotangent planeis
3
4
x
1
4
y + z = 3+ 2+ 1= 0
i.e. tangent planeis3x + y 4z = 0
Directional derivatives
Tofindthegradient onthesurfacez = f
(
x,y
)
at thepoint A
(
a, b,c
)
inthedirectiondefinedbythe
horizontal unit vector u=
cosc
sinc
weput dx = cosc anddy = sinc intheequationof thetangent plane
at A togivedz =
cf
cx
cosc +
cf
cy
sinc, thepartial derivativesbeingevaluatedat A. Sincethisvalueof dz
isthevertical changeneededtoget backtothetangent planeafter aunit horizontal step, it isalsoa
measureof therequiredgradient. ThisisknownastheDIRECTIONAL DERIVATIVE inthe
directionof u.
The vector grad f
Thedirectional derivativecanalsobewrittenasthescalar product
cosc
sinc
cf/cx
cf/cy
35
Thesecondof thesevectorsiscalledgradf, or f . ispronouncednabla
grad f isavector normal tothecontour throughA andisalsothemaximumgradient onthesurface.
Summarising, if A
(
a,b,c
)
isapoint onthesurfacez = f
(
x,y
)
andthecontour withequationf
(
x,y
)
= c
isdrawninthex y plane, thenthevector grad f, drawnstartingat point A
'
(thefoot of theperpendicular
fromA tothex y plane) isnormal tothecontour, pointsinthedirectionof greatest slope, andhas
magnitudeequal tothegreatest gradient at A.
Implicit functions
To find the derivative of an implicit function f
(
x,y
)
wenotethat because
cf/cx
cf/cy
isanormal tothe
curve, consideredasacontour, then
cf/cx
cf/cy
must betangential andsogivesthederivative.
i.e.
dy
dx
=
cf/cx
cf/cy
Stationary points
A STATIONARY POINT on a surface is a point where the tangent plane is horizontal.
i.e. z = constant
cf
cx
and
cf
cy
arebothzero.
It isalocal maximumif f
(
a + h, b + k
)
< f
(
a,b
)
for all sufficientlysmall h,k (not bothzero)
It isalocal minimumif f
(
a + h, b + k
)
> f
(
a,b
)
It isaSADDLEpoint if f
(
a + h,b + k
)
< f
(
a, b
)
Functions of more than two variables
which gives an approximation for the change If w= g
(
x,y,z
)
thenwehave^w~
cw
cx
^x +
cw
cy
^y +
cw
cz
^z
inw duetosmall changesinx,y andz.
Thevector grad gisnowdefinedtobe
cw/cx
cw/cy
cw/cz
andmaybeinterpretedasthevector whose
magnitudeanddirectiongivesthegreatest rateof changeof w andthedirectioninwhichit occurs.
The surface g(x,y,z) =k
Theset of points
(
x,y, z
)
for whichw= k formsasurfacewithequationg
(
x,y,z
)
= k, thisisthethree-
dimensional equivalent of acontour of afunctionof twovariables. It issometimesknownasa
CONTOURSURFACE
If A
(
a,b, c
)
isapoint of thesurfacew = k, andu isanyunit vector tangential tothesurfaceat A, then
sincew remainsconstant inthesurface, thedirectional derivativeinthedirectionof u iszero.
i.e. u.grad g= 0 grad gisperpendicular tou andhencetoall vectorsinthetangent planesoit isa
normal vector for thetangent plane.
Thetangent planethushasequation
cw
cx
x +
cw
cy
y +
cw
cz
z = constant =
cw
cx
a +
cw
cy
b +
cw
cz
c
andequationof normal at A isr= a+zgrad g
36
A
A'
gradf
Ex A surface has equation z = x
2
y 2xy y
2
3y + 8
(i) Find
cz
cx
and
cz
cy
(ii) Find the co-ordinates of the three stationary points on the surface.
(iii) Find the equation of the tangent plane to the surface at the point (0,1, 4)
(iv) The normal line at the point (2,0, 8) meetsthesurfaceagainat thepoint P.
Findtheco-ordinatesof P.
(i) z = x
2
y 2xy y
2
3y + 8
cz
cx
= 2xy 2y and
cz
cy
= x
2
2x 2y 3
(ii) At stationary points
cz
cx
=
cz
cy
= 0i.e. xy y = 0 y = 0or x = 1
substitutingin
cz
cy
= 0wehavex
2
2x 3= 0if y = 0i.e.
(
x 3
)
(x + 1) = 0 x = 3or 1
x = 3, y = 0 z = 8andx = 1,y = 0 z = 8
andy = 2if x = 1 z = 2+ 4 4+ 6+ 8= 12sostationarypointsare(1,0,8),(3, 0,8) & (1,2,12)
(iii) At (0,1,4) wehave
cz
cx
= 2and
cz
cy
= 5sotangent planeisz 4= 2(x) 5(y 1)
i.e. 2x + 5y + z = 9
(iv) Let g
(
x,y,z
)
= x
2
y 2xy y
2
3y z +8
At (2,0, 8)
cg
cx
= 0
cg
cy
= 3and
cg
cz
= 1so
0
3
1
isanormal vector
Equation of normal line is thus r =
2
0
8
+ z
0
3
1
whichmeetssurfaceagainwhen
12z12z9z
2
9z(8+z) +8=09z
2
+10z =0z =0or
10
9
Co-ordinates of P are thus 2,3
10
9
,8
10
9
i.e. 2,
10
3
,
62
9
Ex. (i) Given that g(x,y, z) = 3x
2
+ y
2
+ 2z
2
xy + 2xz 9, find
cg
cx
,
cg
cy
and
cg
cz
A surface has equation 3x
2
+y
2
+3z
2
xy +2xz 9= 0
(ii) Find the equation of the normal line to the surface at the point P(1,1, 2)
(iii) This normal line meets the surface again at the point Q. Find the co-ordinates of Q, and show that the
normal line at P is also the normal line at Q.
(iv) Find the co-ordinates of the two points on the surface where the tangent plane is parallel to the plane
x = 0
(i) g(x,y,z) = 3x
2
+y
2
+2z
2
xy +2xz 9

cg
cx
= 6x y + 2z,
cg
cy
= 2y x and
cg
cz
= 4z + 2x
(ii) at P(1,1, 2),
cg
cx
= 3,
cg
cy
= 3and
cg
cz
= 6
Hence, equation of normal is r =
1
1
2
+ z
1
1
2
= t
1
1
2
(iii) Parametric equations of normal are x = t, y = t, z = 2t
and this meets the curve where 3t
2
+t
2
+8t
2
+t
2
4t
2
9= 0t = 1
t = 1correspondstothegivenpoint sotheother point Qis
(
1, 1,2
)
at Q(1,1, 2),
cg
cx
= 3,
cg
cy
= 3and
cg
cz
= 6soequationof normal at Q isr = s
1
1
2
whichisclearlythesamelineasthenormal at P.
(iv) tangent plane is parallel to x = 0where
cg
cy
=
cg
cz
= 0 2y = x and4z = 2x
i.e. at pointsof theform
(
2k,k,k
)
such points lie on the surface if 12k
2
+ k
2
+ 2k
2
2k
2
4k
2
9= 0 k = 1
hence, therequiredpointsare(2,1,1) and(2, 1,1)
37
Differential Geometry
Envelopes
Tofindtheenvelopeof asystemof curvesof theformf
(
x,y,p
)
= 0wherep isaparameter wefind
thepartial derivativeof f withrespect totheparameter andsolvef
(
x, y,p
)
= 0and
cf
cp
= 0
simultaneously.
Arc Length
The positive sense along a curve is the senase in which a point moves as the independent variable
(or parameter) increases.
ByPythagoras, inanelemental trianglewehave^s = (^x)
2
+ (^y)
2
hence,
ds
dp
=
dx
dp
2
+
dy
dp
2
so arc length =
]
p=a
p=b
dx
dp
2
+
dy
dp
2
dp =
]
x=a
x=b
1 +
dy
dx
2
dx = ]
y=a
y=b
1 +
dy
dy
2
dy
dependingonhowtheequationisdefined.
I npolar coordinates wehaves =
]
0=c
0=
dr
d0
2
+ r
2
d0
Solids of revolution
. Consider acurvey = f
(
x
)
rotatedthroughfour right anglesabout thex axis, thenwehave
Volume generated between x = a and x = b is v
]
a
b
(f(x ))
2
dx
Curved surface area produced is
]
A
B
2vy ds =
]
a
b
2vy
ds
dx
dx =
]
p 1
p 2
2vy
ds
dp
dp =
]
p 1
p 2
2vy
dx
dp
2
+
dy
dp
2
dp
Intrinsic equations
TheINTRINSICequationof acurveistheequationconnectingthearclengths withtheangle, (psi)
betweenthetangent tothecurveandafixeddirection, usuallyahorizontal axis.
Curvature
TheCURVATURE at apoint P istherateof changeof , witrhrespect tos at P andisdenotedbyt
i.e. t =
d,
ds
. Curveiscurvingtotheleft of thepositivetangent if t > 0.
t = 0at apoint of inflectionbut theconversedoesnot apply.
If theintrinsicequationisknownthendifferentiatingwithrespect to, andinvertingwill givethe
curvature. If wedonot havetheintrinsicequationit israther moredifficult.
Fromcartesianequation, t =
d
2
y
dx
2
1+
dy
dx
2
3/2
or fromparametricequationst =
x
'
y
''
y
'
x
''
(x
'2
+y
'2
)
3/2
Inthislast form, theprimesdenotedifferentiationwithrespect totheparameter.
The Circle of curvature
Thecirclewithitscentreonthenormal throughP, whichtouchesthecurveat P, hascurvaturet
andiscalledthecircleof curvature. Itsradius, p =
1
t
istheradiusof curvatureat P.
i.e. radiusof curvaturep =
1+
dy
dx
2
3/2
d
2
y
dx
2
=
(x
'2
+y
'2
)
3/2
x
'
y
''
y
'
x
''
Centre of curvature
Let t
betheunit vector inthedirectionof thepositivetangent andn theunit vector inthedirectionof

. thepositivenormal, i.e. 90
o
anticlockwisefromt
If C isthecentreof curvaturethenPC = pn andthepositionvector of C isgivenbyc = r +pn

Incoordinateform, x = a psin,, y = b + p cos, whereC
(
x,y
)
isthecentreof curvatureat P
(
a,b
)
38
The evolute of a curve
TheEVOLUTE of acurveisthelocusof thecentreof curvatureasP movesalongthecurve, alternatively
it istheenvelopeof thenormalstothecurveat thepoint P. Thisisusuallytheeasiest waytofindit.
Involutes
If a string is wrapped round the evolute and then unwound, keeping the free part taut and straight, then
theendof thestringdescribestheoriginal curve, whilst anyother point of thestringdescribesaparallel
curve, theseparallel curvesarecalledINVOLUTESof theevolute.
Ex (i) For the point P(0,1) onthecurvey = e
2x
, calulate
(a) the radius of curvature.
(b) the co-ordinates of the centre of curvature.
(ii) (a) Find the equation of the straight line passing through the points (t,0) and
(
0, t
3
)
(b) As t varies, these straight lines define an envelope. Find the cartesian equation of the envelope.
(i) (a) y = e
2x
dy
dx
= 2e
2x
and
d
2
y
dx
2
= 4e
2x
sot =
4e
2x
(1+4e
4x
)
3/2
=
4
5 5
at P
soradiusof curvatureis
5 5
4
(b) t
=
1 5
2/ 5
son=
2/ 5
1/ 5
hencecentreof curvatureisat
0
1
+
5 5
4
2/ 5
1/ 5
=
5/2
9/4
(ii) (a) equationof linej oining(t,0) and
(
0,t
3
)
isy =
t
3
t
(
x t
)
y = t
2
x + t
3
(b) Let g(x,y, t) = y + t
2
x t
3
then
cg
ct
= 2tx 3t
2
sosolvingy + t
2
x t
3
= 0and 2tx 3t
2
= 0
simultaneouslywehavet =
2
3
x (sincet = 0) y +
4
9
x
3
8
27
x
3
= 0 y =
4
27
x
3
Which is the equation of the envelope.
Ex A curve has parametric equations x =4t
1
3
t
3
, y =2t
2
8.
(i) Show that the radius of curvature at a general point 4t
1
3
t
3
,2t
2
8. onthecurveis
1
4
(
4+ t
2
)
2
(ii) Find the centre of curvature corresponding to the point on the curve given by t = 3.
The arc of the curve given by 0 t 2 3 isdsenotedbyC.
(iii) Find the length of the arc C.
(iv) Find the area of the curved surface generated when the arc C is rotated about the x-axis.
(i) x
'
= 4t
2
; y
'
= 4t; x
''
= 2t; y
''
= 4
sop =
(
x
'2
+y
'2
)
3/2
x
'
y
''
y
'
x
''
=
(
4t
2
)
2
+16t
2
3/2
4(4t
2
)+8t
2
=
(
4+t
2
)
3
4(4+t
2
)
=
1
4
(
4+t
2
)
2
(ii) At t = 3, x = 3; y = 10;x
'
= 5andy
'
= 12so
dy
dx
=
12
5
andp =
169
4
t
=
5/13
12/13
n=
12/13
5/13
socentreof curvatureisat
3
10
+
169
4
12/13
5/13
=
36
25/4
i.e. centreof curvatureisat 36,
25
4
(iii) Length of arc is
]
0
2 3
x
'2
+ y
'2
dt =
]
0
2 3
(
4+ t
2
)
dt = 4t +
1
3
t
3
0
2 3
= 8 3 + 8 3 = 16 3
(iv)
Areaof curved surfaceis 2v
]
0
2 3
y x
'2
+ y
'2
dt = 2v
]
0
2 3
(
2t
2
8
)(
4+ t
2
)
dt
= 2v
]
0
2 3
(
2t
4
32
)
dt = 2v
2
5
t
5
32t
0
2 3
= 2v
2
5
32 9 3 64 3 =
512 3 v
5
39
Abstract Algebra
A binaryoperatrion onagivenset of elements, isarulewhichtakestwoelementsof theset (which
neednot bedistinct) andformsadefinitethirdelement (whichmaybethesameaseither of the
original ones). Typical examplesare:
Additionof numbers, vectors, matricesetc.
Multiplicationof numbersor matrices.
Combinationof permutations, functionsetc.
If theresult of theoperationisalwaysanelement of theoriginal set thenwesaytheoperationisCLOSED
If changingtheorder of thetwoelementsgivesthesameresult, i.e. a + b = b + a thenwesaythe
operationisCOMMUTATIVE
If a + (b + c) = (a + b) + c for all a,b,c intheset wesaytheoperationisASSOCIATIVE
If thereisanelement e intheset suchthat a + e = e + a = a for everyelement a intheset thene iscalled
anIDENTITY element for theoperation.
If for anelement a intheset, thereexistsanelement b suchthat a + b = b + a = e thenwesaythat b isthe
INVERSE of a, moreusuallydenotedbya
1
unlesstheoperationisoneof additionwhenweusea
for theinverse.
Groups
A non-emptyset S together withabinaryoperation+ issaidtoformaGROUP if thefollowingfour
axiomsholdtrue.
(C) + isclosed inS. i.e. a,b inS a + b andb + a areinS (notethat a + b andb + a maybedifferent)
(A) . + isassociative. i.e. a + (b + c) = (a + b) + c for all a, b,c inS
(N) Thereisanidentity or neutral element e inS suchthat a + e = e + a = a for all a inS
(I) Eachelement a inS hasaninverse element a
1
inS suchthat a + a
1
= a
1
+ a = e
If, inaddition, a + b = b + a for everya, b inS thenwesayit isaCOMMUTATIVE or ABELIAN group.
Immediate consequences of the axioms are:
(a) The identity element is unique. If e andf aretwoidentityelementsthenef = e andef = f e = f
(b) Inverses are unique. If a hastwoinverses, p andq thenp = pe = p(aq) = (pa)q = eq = q
(c) Cancellation laws hold: ha =hah
1
(
ha
)
=h
1
(
hb
)
(
h
1
h
)
a =
(
h
1
h
)
b a =b
(d) ax = b hasauniquesolution: ax = b a
1
(ax) = a
1
b (a
1
a)x = a
1
b x = a
1
b
But notethat ax = b doesnot implythat x = ba
1
unlessthegroupiscommutative.
The order of a group
A groupmaybeeither finiteor infinite. A finitegroupmeansthat it hasadefinitefinitenumber of
elements, theORDER of thegroup.
Isomorphisms
A mappingbetweentwogroupsof equal sizewhichpreservesthestructureof thegroupsiscalledan
ISOMORPHISM andthegroupsaresaidtobeISOMORPHIC.
A mappingf isanisomorphismif, for anya,b inthefirst set wehavef
(
a + b
)
= f
(
a
)
of
(
b
)
where+ isthe
operationinthefirst set ando istheoperationinthesecondset.
Subgroups
If thereisasubset of theelementsof agroupG =
(
S,+
)
whichhasthegrouppropertiesitself thenit is
calledaSUBGROUP of G.
A PROPER subgroup is any non-empty subgroup excluding the whole group itself. {e} and G are
TRIVIAL subgroups. All others are called NON-TRIVIAL
Cayleys Theorem
A finitegroupG of order n isisomorphictoasubgroupof thegroupof permutationsof n elements.
Lagranges Theorem
Thisstatesthat theorder of anysubgroupof afinitegroupG, must beadivisor of theorder of G.
See proof in text book. (page 155)
Cyclic groups
If all theelementsof afinitegroupG arepowersof someelement of G thenit iscalledaCYCLIC
group.
Groupsof primeorder arenecessarilycyclic.
Ex
G is a finite group which is commutative. S andT aresubgroupsof G, andS , T = {e}, wheree isthe
40
identityelement of G.
WedefineST tobetheset of all elementsof theformst, wheres S andt T.
(i) Show that ST isasubgroupof G.
(ii) Showthat if s
1
t
1
= s
2
t
2
wheres
1
,s
2
S andt
1
,t
2
T, thens
1
= s
2
andt
1
= t
2
.
(iii) Deduce that if S containsk elementsandT containsm elements, thenST containskm elements.
Nowlet G = {1,5,7,11,13, 17,19,23,25,29, 31,35}, wherethebinaryoperationismultiplication.
You may assume that G is a commutative group. modulo36.
Three cyclic subgroups of G are S = {1,13,25}; T = {1,17} andU = {1,35}
(iv) List the elements of the subgroup ST and show that ST is cyclic.
(v) List the elements of TU and show that TU is not cyclic.
(i) x,y ST x = s
1
t
1
andy = s
2
t
2
for somes
1
, s
2
S, t
1
,t
2
T
xy = s
1
t
1
s
2
t
2
= s
1
s
2
t
1
t
2
(
sinceG isacommutativegroup
)
but s
1
s
2
S andt
1
t
2
T soxy ST andwehaveclosure.
We are given that ST contains the identity element.
Associativity is inherited from G so ST is a subgroup og G.
(ii) s
1
t
1
= s
2
t
2
s
2
1
s
1
= t
2
t
1
1
ands
2
1
s
1
S andt
2
t
1
1
T sowemust haves
2
1
s
1
= t
2
t
1
1
= e
s
2
1
s
1
= e s
1
= s
2
andsimilarlyt
1
= t
2
(iii) From (ii) all the elements st are distinct so we have
(k 1) elementsof theformse andm 1of theform
, i.e. et + ee + (k 1)(m 1) of theformst givingatotal of (k 1)(m 1) + (k 1) + (m 1) + 1elements
Km.
(iv) ST contains1,13,17and25somust alsocontain13 17and25 17(mod36) i.e. 5and29
soST = {1,5,13,17,25, 29} andbyinspectionthisiscyclic, generatedbytheelement 5.
(v) TU contains1,17and35soonlyother element is17 35(mod36) = 19
hence, TU = {1,17,19,35} andsinceeveryelement isself inverseit cannot becyclic.
Ex. A non-Abelian group G consists of eight 2 2matrices, andthebinaryoperationismatrix
multiplication. Theeight distinct elementsof G canbe writtenasG = {I,A,A
2
,A
3
,B,AB,A
2
B,A
3
B}
are whereI istheidentitymatrix, andA,B 2 2matrices suchthat A
4
= I, B
2
= I andBA = A
3
B
(i) Show that (
A
2
B
)
(AB) = A and(AB)
(
A
2
B
)
=A
3
(ii) Evaluate the following products, giving each one as an element of G as listed above.
(AB)(A), (AB)(AB), (B)(A
3
)
(iii) Find the order of each element of G.
(iv) Show that
(v) Find the other two subgroups of G which have order 4.
(vi) For each of the three subgroups of order 4, state whether or not it is a cyclic subgroup.
Solution
(i)
(A
2
B)(AB) = A
2
(BA)B = A
2
(A
3
B)B = A
5
B
2
= A
4
AB
2
= IAI = A

(AB)(A
2
B) = A(BA)AB = A(A
3
B)AB = A
4
BAB = IA
3
B
2
= A
3
I = A
3
(ii)
(AB)(A) = A(BA) = AA
3
B = IB = B
(AB)(AB) = A(BA)B = A(A
3
B)B = A
4
B
2
= II = I
(B)(A
2
) = (BA)A = A
3
BA = A
3
A
3
B = A
6
B = A
2
B
(iii)
A andA
3
areof order 4, A
2
, B andAB areof order 2
(A
2
B)(A
2
B) = A
2
(BA)(AB) = A
2
A
3
BAB = ABAB = I soA
2
B isof order 2
(A
3
B)(A
3
B) = A
3
A
3
AB
2
B = A
2
BA
2
B = I soA
3
B hasorder 2
Obviously I has order 1
(iv) Using results already established we can construct
the table for
I, A
2
, B, A
2
B
From the table we see that we have closure, an identity
and every element has an inverse. Associativity is
inherited from G hence is a subgroup I, A
2
, B, A
2
B
of G.
(iv) other two subgroups of order 4 are {I,A,A
2
,A
3
}
41
I A
2
B A
2
B A
2
B
A
2
I A
2
B B B
B A
2
B I A
2
A
2
A
2
B B A
2
I I
A
2
B B A
2
I
and I, A
2
, A
3
B, AB
The first is obviously a cyclic group of order 4
and we verify the other by constructing a table
verifying that it is a subgroup
(vi) is cyclic, the other two are not {I,A,A
2
,A
3
}
42
I A
2
AB A
3
B A
3
B
A
2
I A
3
B AB AB
AB A
3
B I A
2
A
2
A
3
B AB A
2
I I
A
3
B AB A
2
I
Markov Chains
A MARKOV CHAIN is a sequence of events where the probability of an outcome at one stage depends
only on the outcome of the event at the previous stage.
A TRANSITION matrix consists of the TRANSITION PROBABILITIES (The probabilities of passing
from one stage to the next) Each column of the transition matrux is a PROBABILITY VECTOR
If P is the transition matrix and p a particular probability vector representing the probabilities at one stage
then Pp represents the probabilities at the next stage, P
2
p theprobabilitiesat thenext but one
stageandsoon.
Powers of the transition matrix give conditional probabilities of moving from one outcome to another
over 2,3,4, ... steps.
Ex. A study of the weather over a long period of time suggests that if today is sunny then there is a 70%
chance that tomorrow will also be sunny whilst if today is cloudy then there is an 80% chance that
tomorrow will also be cloudy. It is sunny on a particular Saturday, what is the chance that it will be
(a) sunny on the following Monday
(b) cloudy on the following Tuesday.
(c) sunny on the following Wednesday.
We first form a transition matrix P =
sunny cloudy
sunny 0.7 0.2
cloudy 0.3 0.8
(a) MondayinvolvestwostepssoprobabilitiesaregivenbyP
2
=
0.55 0.3
0.45 0.7
sochancethat it issunnyonMondayis0.55
(b) TuesdayinvolvesthreestepsowerequireP
3
=
0.475 0.35
0.525 0.65
sochancethat it iscloudyonTuesdayis0.525
(c) Finally, Wednesdayinvolvesfour stepssoP
4
=
0.4375 0.375
0.5625 0.625
sochancethat it issunnyonWednesdayis0.4375
If a Markov chain has more that two states at each stage then the transition matyrix will be larger e.g.
3 3for threepossibleoutcomesat eachstageandsoonbut thegeneral principleremainsthesame.
It isfor thisreasonthat youareadvisedthat acalculator that will handlematrixmultiplicationis
essential whenstudyingthistopic.
For certaintransitionmatricesP wefindthat asn , P
n
amatrixwhosecolumnsareidentical.
ThesecolumnsgivetheEQUILIBRIUMPROBABILITIESandshowthat, inthelongrun, the
probabilitiesof eachstateareindependent of what happenedat thebeginning.
Intheexampleabovewefindthat theequilibriummatrixis
0.4 0.4
0.6 0.6
i.e. thismeansthat thereisa40%chanceof asunnydayat anytimeinthedistant future.
This matrix can be found either by iteration i.e. Raising P to higher and higher powers, or as follows
If p=
p
1
p
2
istheequilibriul vector thenwemust havePp= p
Againusingthesameexampleasbeforethisgives0.7p
1
+ 0.2p
2
= p
1
and0.3p
1
+ 0.8p
2
= p
2
i.e. 0.2p
2
0.3p
1
= 0and0.3p
1
0.2p
2
= 0alsoof coursep
1
+ p
2
= 1soeliminatingp
2
wehave
0.2 0.2p
1
0.3p
1
= 0 0.5p
1
= 0.2 p
1
= 0.4andp
2
= 0.6
Run length of a Markov chain
Still usingour sunny/cloudyexample, if it issunnyonMonday, what istheprobabilitythat it issunny
43
for therest of theweek, i.e. uptoandincludingSaturdaybut cloudyonSunday.
Thisiseasilyfoundtobe0.7
5
0.30= 0.050421
Generalising, theprobabilitythat it issunnyfor thenext X daysisP
(
X = r
)
= 0.7
r
0.3
i.e. theprobabilitiesarethetermsof ageometricsequence. Sowecaneasilyfindtheexpectedrunlength
by evaluating
E
1
rP(X = r) = 0.3 0.7+ 2 0.3 0.7

2
+3 0.3 0.7
3
+ ...
whichis0.3 0.7timesthebinomial expansionof
(
1 0.7
)
2
hence, expectednumber of further sunnydaysafter asunnydayis
0.30.7
0.3
2
=
0.21
0.09
= 2.33
Generalising, if c istheprobabilitythat thesystemremainsinthesamestateat thenext stagethenthe
expectedrunlengthfor that stateis
c(1c)
(1c)
2
=
c
1c
Notethat thisistheexpected(mean) number of FURTHER consecutivedaysthat theinitial state
remainsunchangedandisonelessthanthenumber of timesthestateisrepeated.
Classification of Markov Chains
A Markov chain is said to be REGULAR if some power of the transition matrix contains only
In a regular Markov chain it is possible to pass from any state to any other state and positiveentries.
there is a unique limiting probability vector.
Random Walks
If it is possible to assign an order to the various states so that from any one state it is only possible to
move to a limited number of other states then this can be described as a RANDOM WALK.
Periodic chains
If P is a transition matrix and for some value of n P
n+1
= P thentheMarkovchainisPERIODIC with
PERIODn.
Ex. P =
0. 0.5 0
1 0 1
0 0.5 0
P
2
=
0.5 0 0.5
0 1 0
0.5 0 0.5
andP
3
=
0 0.5 0
1 0 1
0 0.5 0
= P
Notethat P
2n
= P
n
andP
2n1
= P
2n+1
for all n soperiodicwithperiod2
Reflecting Barriers
If at somestage, theoutcomeisinevitablethenwehaveaREFLECTINGBARRIER.
Ex. Theexampleaboverepresentsthefollowingsituation. A bagcontains2balls, bothof whichare
black or white.
A ball isdrawnat randomandreplacedwithoneof theoppositecolour. Thetransitionmatrixfor the
number of black ballsinthebagat eachstageisP =
0b 1b 2b
0b 0 0.5 0
1b 1 0 1
2b 0 0.5 0
Aswesawaboveif thereare0blackballsinthebagthenat thenext stagetheremust beoneand
similarlyif therewere2black ballsinthebagthenat thenext stagetheremust beonlyone.
Wethereforehavetworeflectingbarriers. Thestatesof 0blackballsor 2black balls.
A zerointheleadingdiagonal of atransitionmatrixwitha1aboveor belowit indicatesareflecting
. barrier
Absorbing States
If at somestagewearriveat somestatewhereit isimpossibletoleavethat statethenwehavean
44
ABSORBINGBARRIER.
Ex. A scientist is studying a colony of a particular organism, and classifies the individuals as being in one
of three states, well, ill or dead. Which of these is an absorbing state?
If at somestagetheorganismisdeadthenit cannot changethat statesothisistheabsorbingstate.
Ex. A decorative light glows in one of four colours, purple,blue,red and yellow. The colour changes at
theendof afixedtimeinterval andthetransitionmatrixisasfollows:
P B R Y
P 0 0.2 0.4 0
B 1 0 0.2 0
R 0 0.4 0 1
Y 0 0.4 0.4 0
(i) Explain the significance of the zeros in the leading diagonal of the matrix.
(ii) Which colour occurs most often in the long run? What proportion of the time does the light show
that colour?
The light develops a fault. Once it shows blue it gets stuck there and can no longer change to any other
colour.
(iii) Write down the new transition matrix and describe any features that the chain now has.
(iv) When the fault occurs, the light is showing red. What is the probability that seven intervals later it is
still not showingblue?
(i) Thezerosshowthat thelight never showsthesamecolour intwosuccessiveintervals.
P B R Y
P 0 0.2 0.4 0
B 1 0 0.2 0
R 0 0.4 0 1
Y 0 0.4 0.4 0
p
1
p
2
p
3
p
4
=
p
1
p
2
p
3
p
4
0.2p
2
+ 0.4p
3
= p
1
, p
1
+ 0.2p
3
= p
2
, 0.4p
2
+ p
4
= p
3
,
(ii) 0.4p
2
+ 0.4p
3
= p
4
so0.2p
2
+ 0.4p
3
= p
2
0.2p
3
0.6p
3
= 0.8p
2
so3p
3
= 4p
2
hence, p
1
= 0.55p
3
andp
4
= 0.7p
3
sop
1
: p
2
: p
3
: p
4
= 0.55: 0.75: 1;0.7sop
3
isthegreatest
i.e. redoccursmost frequently,withprobability
1
3
(iii) M =
P B R Y
P 0 0 0.4 0
B 1 1 0.2 0
R 0 0 0 1
Y 0 0 0.4 0
Blueisanabsorbingstate.
(iv) so probability of blue after 7 intervals is 0.9488 and so the M
7
=
P B R Y
P 0 0 0.0256 0
B 1 1 0.9488 0.936
R 0 0 0 0.064
Y 0 0 0.0256 0
probabilityof not showingblueis0.0512
45

Revision Notes FP1-3

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Revision Notes FP1-3

Uploaded by

Copyright:

Available Formats

Further Concepts for Advanced Mathematics FP1

betheunit vector inthedirectionof thepositivetangent andn theunit vector inthedirectionof

If C isthecentreof curvaturethenPC = pn andthepositionvector of C isgivenbyc = r +pn

rP(X = r) = 0.3 0.7+ 2 0.3 0.7

You might also like