1.

14 Engineering Mathematics

NOW,4 (x) > 0 for x< h/2 and 4'() < 0 for x> hlv

nerefore, A(x) is maximum area is A(h/N2)

=
h/4,
when x =h/V2 and the maxinnu
Leibniz formula Let fand g be two differentiable functions. Then. the nth order derivative of the
product g is given by the Leibniz formula as

Cor) ga) + "C, ste-V) ga)+... +nC. r a r ) g " x ) + . + "CaJa) g")1.9

This formula can be proved by induction.

CAmple 1.15 Find the fourth order derivative of e" sin bx at the point x 0 .

SOUton Letfa) =e", gt) sin bx and Fx) fx) gt). Using the Leibniz formula,
=
we
=
obtain

F) =

(sin bx)
dx*
=

*C,(ea*ya sin bx "C,(")°"

+ (Sin bx)
*
"Ca(*Y" sin bx)" + *c,(Y (sin bx)) + "C2 e"(sin bx)
e a sin bx + 4a'b cos bx 6ab sin bx 4ab' cos bx + b* sin br]
-

Hence, FO(0) =
4a'b 4ab 4ab(a? b).
-

=
-

1.2.6 Taylor's Theorem and Taylor's Series

A very useful
technique in the analysis of real valued functions is the
functions by polynomials.
Taylor's theorem (Taylor's approximation of continuous
such an
approximation by polynomials. Taylor's theoremformula)
can be
is an
important tool which provides
value theorems to
higher order derivatives. Mean value regarded as an extension of the mean
theorems
its first order
derivative, whereas the Taylor's theorem relates the relate the value of the function and
order derivatives. value of the function and
its. higher
Theorem 1.7 (Taylor's theorem
with
derivatives upto (n + 1)th order in some remainder) Let f(x) be defined and have
of the function interval 7,
containing a point a. Then, continueous
f(x) about the x a is point =

given by Taylor's expansion

f) -f(a) +
a)+ s(a)
f"(a) + s(a) +R,)
... +
+
R,()
n (1.fo)
where, (x-a)
R,(x) = -a)"*1
rut(c),
(n+1)! a < c<X
is the
(1.11)
remainder or the error term of the
expansion.
Proof We first find a
polynomial P,r), of
degree n, which satisfies the corditious
P(a) =f(a), P (a),
In
*S°(a), k
=1,2, . ,
a certain sense, P(x) is a polynomial
n.

approximation to Jr). Write the

required polynomialas
required polynomial as
 Functions of a Real Variable 1.15

P,x) = co + C1x- a) + c2(r - a + .. + c,(r - ay'.

conditions, we obtain
1 Ising
the given
P,() =fla) = Co, Pa(a) =fla) = e, P,(a) =f"la) = 2 c2

P ( a ) =f (a) = (n!) e

Hence, we have

a ) , k=0,1.2,.
Therefore, fx) =
P,(x)= f(a) + fla)+ Ma)+.+ a pn(a)
1! 2! n!

The error of approximation is given by R, (x) =f(x) -

P, (x). Therefore,

ft)= P,(x)+R,(x) = 2 ( a ) +R, Ct).

k=0

Now. we derive a form of R, (x). Write R, (x) as

R,
( x )= - a ) " + 1 |

-h(x)
(n +1)!
where h(x) is to be determined.
Consider the auxiliary function
+1

t)+ n+1! h)|.a<t<x.

F)=f(x)-ft)+(x - t)ft)+..+
we have t as a variable and x is fixed. The function F(t) has the following properties:
in <t<x,
1) F(1) is continuous in a Stsx and differentiable a

ii) Flx) =0,

a)= f(r)-Sla) + (x-a)f'la) + . .+ s ( a ) +( r - a ) * -h(x)|

n! (n +1)!

=fx) -fx) =0.

Therefore, there exists a value
the hypothesis of the Rolle's theorem
satisfies on la, x].
Ca )
C, a<c<x
such that F'(c) = 0. Now,

F't)=0-s')-s'O) +(x-1)s"()--22! - p")+

+ n+l)(o)- (t+Dr-1)n-=h(r)- f*D0)]

n! n!
(n+1)!
 1.16 Engineering Mathematics

and
F)= 0 =
C[h() fu*l(c)]. -

n!
We obtain
h(x) =
f (n*D(c). Therefore,
R,r) = r - a ) " * 1
c), a <c<X.
n +1)!
The error term can also be written as

R,(x) = -a)"*1 n*)

f+) (a + e(x - a), 0 < 0<1
(n +1)! (L.12
which is called the
Lagrange form of the remainder.
If a =
0, we get

o)-0) )+")++ ()+ (n+1)!- f*"(c), 0

n! <e<x (1.1)
which is called the
Maclaurin's theorem with
Writing x + =
remainder.
a h in Eq.
(1.10), we obtain

fa +h
fla) h2

The error of
+
sa)+"la)+ 2!
h
n! f(a). (1.14
approximation simplifies as

h+1
R,x)=
(n +1)! f c), a <c<a+h. (1.15)
If we
neglect the error term in
Eq. (1.10), we obtain

fo) P,t) = m! fmal

m=0
(1.16
which is called the nth
degree 1aylor's poiynomial
Since c or e in the remainder term
(see
approximation to fx.
aiven x in the interval i.
Eqs.(1.11), (1.12)) is not
HOwever, a
bound on
known,
the error
we cannot
car evaluate R.
can be obtained as
R,cr) = (x-a)* rn+(smax - a .
(n+1)!
(1.17
 Functions ofa Real Variable 1.17

Eor a given error bound E, we can use Eq. (1.17) to determine

(i)n for a
given x and a,
i) x = ** for a given n and a such that IR,(*)] <e

Cauchy form of remainder

Consider a function ox) defined on [a, a + h] as

ox)=f(x) + (a +h-x)f)+...+ (a +h- x) f (x)+ A(a +h- x)

n!

where A is a constant to be determined such that o(a + h) = o(a).

The function r ) satisfies all conditions of the Rolle's theorem. Therefore,

oa + eh) = 0, 0<0< 1.

Now, ola)=a +h-x"fa+)(x) -A

and h
o'(a + eh)="1-0"f+(a + 0h) - A = 0.
n!

Hence, A=1-0)"fn-(a + eh).

n!

From o(a+ h) =
¢(a),we get

fla+h)=fla) +hf'(a) ++f(a)+ hA.

n!

Therefore, the remainder in the Taylor's throrem is

R,(x)=hA =(1-0)" f*(a + 0h), 0< 6<1 (1.18)

Integral form of remainder

Consider the result

xdr =fl,1)-Sx,),
wnere h. Write the transformation t. Then, have
x+1 =
X, + x =
Xy+l- we

'x)dr=- f'-1ndt= Jsx,+l -)dt

 1.18 Engineering Mathematics

Integrating by parts, we get

di
1) di =[ur'a,u1 D1%+ -

1,1-)

hf'C,)+ 1f"1
-

) d.
Integrating by parts again. we get

f ( + 1 - 1) dt
1 ) dt =hf'6,)+

hf)+ 2! )+ s"a,+1 -

) d.

Integrating by parts repeatedly, we get

fx1 ) dt =hf'x,) +fx,) +... + a,) + f,+1-d

f+1)-f%,).
Therefore.

fo-)=/x): 2 k! )+
k=l fa*G) d
where s =Xn+1 -I.

Hence, the remainder in the Taylor's theorem is

R, ,1-s¥ fa*"(o) ds.

Setting x, = a, We may also write

R, a+h-sfo*") ds, (1.19

Erample 1.16 The function J) =
sin xX IS
three about the point x =
0. Find approximated by Tavlor's
such that the
interval [0, c].
c
error
satisfies |R,(x) nolvnomial of de
s 0.001
Solution We have

f)= S0) +xf'0) +

"0)+3T"0)
 Functions ofa Real Variable 1.19

Sin x, we obtain
For fx)
=

)= Cos
x, f"(a) = -

sin x, f"(x) = -cos x andf" (x) = Sin x.

Hence. f0) =
1.f(0) 0, f"(0)
= = -I and fE) = sin

The required approximation is f(t) = sin x =x-

6
The maximum error in the interval [0, c] is given by

R-sins max | sinxS24

0Sxse 24 |0Srsc

Now, c is to be determined such that

0.001 or c0.024

criterion is satisfied.
We obtainc> 0.3936. Hence, for all x in the interval [0, 0.3936], this error

Taylor's Series
then
(Eqs.(1.10), (1.11)), if the remainder R,(x) ->0
as n > co,
In the Taylor's formula with remainder
we obtain

Sx)= f(a)+ a )+ rl)+..- a m(a)+.. (1.20)

2! n!

When 0, we obtain the Maclaurin's series

which is called the Taylor's series. a =

S)=S0)+SO)+ 0)++ n!
"(0)+ (1.21)
2!
*

has continuous derivatives upto (n + 1)th order, f" '(x) is

Since it is assumed that f(x)
bounded in the interval (a, x). Hence, to establish that lim IR,(x)| 0, it is sufficient to show that =

-a 1 = 0 for any fixed numbers x and a. Now, for any fixed numbers x and a, we can always
lim
n-o +1)!
find a finite positive integer N such that lx - al < N. Denote q = lr - al IN. Then,

(n+1)!

N-1|
(-a)4.4 ** (x-a N*2
(N-1! 4N-1)! |
 1.20 Engineering Mathematics

Now, -a)-1| n. Also q<1.

Hence,
is independcnt of
(N-1)! t c quantily and

lim-a)"-| 1R,(r)|
= 0.
=
0 for and lim
no(n+1)! any fixed x and a,

Example 1.17 Oblain the Taylor's polynomial expansion of the functionrealf(x) sinx about
=

fOr any real x. Hene x.

point x T/4. Show that the
=
term tends to
error
Hence, write
zero as n
o

Taylor's series expansion of f(a).

Solution For flx) sinx, =
we have
(-1y" cosr

(-1)" sinr and "(x)

=

fx) =

for any
integer n. Therefore,

f(z/4) =
(-1)"WZ and (t/4) (-1)N2. =

Hence, the Taylor's expansion of fa) = sinx about x = t/4 is given dy

fo---4) )R.
Now,
2+1
4

since.fn+"(c) =
|-1y" cosc|<1. Hence, R)>0 as n oo,

Similarly, we find that R2 +1)>0 as n o,

Therefore, the required Taylor's series expansion is given by

Example 1.18 Obtain the Maclaurin's series expansion of flx) = sin (m sin x), where m is a conN
Maclaurin's series is given by
Solution The

flx)= f0) +xf0) +s"0)+f"(0)+.. (

2!
 Functions ofa Real Variable 1.21

= sin (n sin'x). we find

For y(x)f)

c o s (1m sin 'x). (1.23)

y'(x)= f'(x)=
as follows:
We proceed
From (1.23),
Let v.(a) =y"x) = ( ) . we get

=*(1 - y).
(1-)i =m* coss(m sin'x) nm'(1 sin?(m sin'x)
= -

Defferentiating, we get
2 ( 1 - ) y 2 - 2x yf =-2m"yy1

( - x ) y - *y + mty = 0, >1*0 (1.24)

or

times using Leibniz theorem, we obtain

Differentiating (1.24) n

=0
(1-x)y,+2 +
n(-2x)yn+ + 2
-2)y,-(xyn+ +ny,)+ m*y,

or (1-yn+2 -

(2n + 1)xyn+1 -

(n-m)y, =0.
For x = 0, we get

n+2 0)
=
(n* -

m)y, (0), n =
0, 1, 2, ...

Therefore, we find

0,f(0) y(0) =
m, s"(0) =
y2(0) =
0,
SO) =
Yo(0) = =

Ys,(0) m(13- m), f"(0) y4(0) =0,

=

(0) = =

f0) =
ys (0) m(12- m) (3- m),
= . .

Substituting in (1.22), we obtain

m(1-m)
+
3 m{1-m)(3n)5
+
f(x) =mx 3! 5!

obtain the value of cos 31° correct to four decimal places.

Example 1.19 Using Taylor's series,
Solution Let fx) = coSx. We have

sinx and so on. Using Taylor's series, we obtain

S) = -sin x, f"x) = -

cos.x, f"(x) =

cos(x+ h)=flx + h) = f(x) +hf'(x) + 2! "x)+*s"(x)+.

= cOS x - hsin x
cosx+sin x +.*
 ngineering Mathematirs
Substituting 30 Tt6 and h
cos 31 1 /180 0.01745. we get

T/180) sin(T/6)(7/180 cos(T/6) +

6 (t/180

(0.01745)
001745 01745
0.86602 0.00872 -

= 0.85717.
0.00013+0.00000+
-0.00013
obtain
cos 31°
1.2.7 Exponential, Logarithmie
0.8572 corre
=

to four decimal places.

Exponential series and Binomial Series
Consider the
function f(x) Taylor's polynomial
mial
e.
approximation of degree S
=

The approximation
Taylor's polyno n about the
nomial
approximation is point x.
given by

For f(x) =
x, we
obtain
S)-s0)+)+0)+ (0)
s)= e, sO) =
1, r =0, 1,
Hence, .., nand
fu(x) =Fe.
Sx) =e
1+x+++
=

Using the
Lagrange 2! n!
form of the
remainder, we get
R,x) =
n+1)
or as +1)!
R,x) =. ,0<8<1.
n+1)!
Therefore, lim
- R,)|= lim |
for all since e
x,
(n+1= lim+
is bounded for a
n n+1)!
Hence, we obtain the given x.
exponential series

e1+ x+
n
 Functions ofa Real Variable 1.23

Example 1.20
Example 1.20 For the Taylor's polynomial approximation of degree n about the point x = 0 for

the function f(x) = e', determine the value o f n such that the error satisfies | R , (x) 0 . 0 0 5 , when

- 1 SxS i.

Solution We have the Taylor's polynomial approximation of e as

Sx) = e = 1 +x + x"
2! n

The maximum error in the interval [1, 1] is given by

R,() = f+) (c) max max

Now, n is to be determined such that

n+1)! -ISxs1| (n +1)!-1Sxs1| n+1)!
e
005 (n
(n +1)!
or +
1)!2 200 e.

We find that n 2 5.
Hence, we will require at least 6 terms in the
toachieve the given Taylor's polynomial approximation
accuracy.
Example 1.21 Obtain the fourth degree Taylor's
x =0. Find the maximum error when 0 Sx< 0.5. polynomial approximation to fx) =
e" about
Solution We have

Sx) -
f0) +xf(0)
For
+
)+ "O)+ O).
fx) =
, obtain
we
() =
2'ea, S0) =
2, r =
0, 1, 2, . and
...

f(C) =
32 e.
CTOre, Sx) =
e =
1 + 2x 4x8x+ 16x 2!
=
1
+2x + 2+
The error
term is
3! 4!
+
given by

RAx)= fO
5! () =

5!
e, 0 < c<.
and
IR,)20
RCx)|S32 max x
120 0sxS0.5 max e
0SxSO.S 20
120
Logarithmic series
Consider the
function fx) Taylor's polynomial approximation of
Taylor's polynomial degree s about the point
In (1 +
=

x). The n

approximation is given by x =
0 for the
 .26 Engineering Mathematics

Therefore. we obtain

S) =
(1 +1)" 10)+1/0)+(0)
=
t. f(o)
r0m!

1 +mx+1m(1)2+... +X
2!

We also have

R,r) =
f+ (c) = 0, n 2 m.
(n+1)!
Case 2 When m is not
orders provided
a
positive integer, flx) =
(1 + x)" possesses continuous
x # -1. derivatives of
Let -1 < x <
1. We have

fax)= m(m - 1) (m n) (1
Using +x"
-
... -

the
Cauchy's form of the
remainder, we get
R,)= (1- "
n! f(0x)
n! (1-0 mm - 1) ..
(m - n) (1 +
0x 7-1-1

m(m-1).. (m-n)| 1-0

n!

Now, for lx|< 1,

0< 6
I+ and
1+0x (1+0x-*0 <0<1 (1.30

1
- +x
0, lim
where a is =0, lim
finite
-1)..(m-n)
a
Eq. (1.30) antity. Since (1 +
8x)" isis n
independent of n
and bounded, we obtain
Therefore, whe
a
positive integer
lim 1R,(c)|= 0.
n-00 boundea fro

and \x\<
1, we
obtain
 Functions of a Real Variable 1.27

f(x) =1+ mx + -)2..mm)...(m-n)4

2! n! (1.31)
Alternative proof

Consider the series 2a, where a, = "m-)...(m-n)

n!
Using the ratio test, we get

n
lim
(m-n-=lxl.
n= no(n+1)
limn
Hence, the series (1.31) converges for 1 <x< 1.
Further, it can be shown that the binomial series (1.31) converges at x =1 when m> -1.
For example, consider the series

(1 1-x+*-x +
For x= 1, the series on the right hand side has two limit points 0 and 1 hence the series is not convergent.
We have the following binomial series

(1+x) =1-2x+ 3x-4x' + -1 <x < 1.

(1+2=1-x - 2-46
-1<x<l.

(1 + = 1+ -2 ..,-1<xs1.
.6
3.6 3.6.9
Exercise 1.1
Find the approximate values of the following quantities using differentials.

1. (1005) 2. (999)1/3
3. (1.001)+2(1.001)S+5. 4. sin 60° 10'.

5. tan 45° 5 30".

6. State why Rolle's theorem cannot be applied to the following functions.

[0, r). (ii) flr) =

Lx] in the interval -1/2, 3/2].
) fx) = tanx in the interval

0SxSI
(ii) X,
f(x) =12-x, 1sxs2.
* brs + cx, 1 Sxs2 at the
7. It is that the Rolle's theorem
holds for the function ft) = *
given
and c.
point x = 4/3. Find the values of b
continuous on la, b] and differentiable in (a, b) such that
8. The functions flx) and g(x) are
c<b.
1 and g(b) = 3. Then, show that f(c) =
3g'(c), a <
fa) =
4, f(b) = 10, g (a) =

roots of e" sinxX=1,

there exists atleast one root of e"cosx+1=0.
9. Prove that between any two real
 1.28 Engineering Mathematics
on
shov
la, b]. Then, 1Ow that for a <

differentiable
functions
c<h
Le
(). g (v) be continuous and
1th)-fta)-(b-a)/ta)_) s'c) *0.
g(c)
b)-g(a)- (b-a)g (a)
differentiable in (4
-

1, a +
show
1). Show that
that there
continuous on la + 1] and 1, a
exist
De +)-f (a-6).
that I such Sa-1)-2 f(a) +fla +1) =f'(a
12. Using the Lagrange mean value theorem, show that

(ii) In (1 +x) <x,

x> 0;
(1)1+<e<l+xe':

in'sV1-*, 0<x<1; (iv)+ sinx<

3. Suppose
for all
that flv)
is differentiable for all values of x such that (a) =a, f(-a) =
- a andi
x. Show that f(0) 0. =

14. Let Fr) and G(x) be two functions defined on [a, b] satisfying the hypothesis of the mean
with value th
G(x) #0 for any x in [a, b]. Show that there exists a point c in (a, b) such that

Fb)-Fla)Flc_G)
Gb)-G(a) C(c)| Ga)G(6)|
Evaluate the limits in
problems 15 to 28.
15. lim-]
XIx"-1 16. lim-e-2c0S x +e
X0 XSin x
17. limn(sinr)
7/2 (T -2x) 18. lim Sinr-x
x0 2x sin
x
19. lim (1 -x) tan
(Tx/2).
20. lim*
21. lim x-i) x2 X
22. lim (sinxanx
In(x-1)|
23. lim XT/2
0 ) = 0.
x0 X)
24. lim
25. lim [ Xo e
x)]l,
+

lim fx) =
0.
26. lim
27. lim x+ SInx
2 x
In
problems 29 136. find the
28. lim +4
29. n (2 + x) intervals in which Xo r+2
- 2x/(2+ x), f(x) is
31. lan xx,
E
lR
xe IR
increasing or
33. In (sin x), 0 30,
32. sin
xlsl, xe
decreasing.
IR.
35. sin x(l cos X),
+

0xn/2 x+|sin
37. Let a b 00 and 34. xl,
x>0. 0<xS 27.
be a (n x)/x,
posiive inleger 36. x>0.
satisfying n22. Prove
that
a-b< (a-D
 Functions ofa Real Variable 1.29

in problems 38 to 43, find the extreme values of the given function fa).

38. (r-1 (a+ 1)'. 39. sin x + cos X.

40. 41. (sin x)sn X
42. 2 sin x + cos 21, 0 Sxs 27. 43. sin x sin 2x + cos* x cos 2x, 0 <x <T.
44. Show that the funcion f(x) = (ar + b/(cx + d) has no extreme value regardless of the values of
a, b, c, d.

-+[(6 -B +b-1)/(B +3b+ 2)], 0S x<1

45. Let fa)2x-3
,1Sx<3.
Find all possible real values of b such that flr) has minimum value at r = 1.
46. Ify =x e, then find d" yldx" at x = 0.
47. Find the nth order derivative of f ) = Vax +b.
48. Find the nth order derivative of f a) = e sin (bx + c).
49. If y = cos,-(T/2) Srs (/2), then find d" yldr" at x = 0.
50. If y = esna, then find d"yld at x =0.
In problems 51 to 55, obtain the Taylor's polynomial approximation of degree n to the function fix)
about the point x = a. Estimate the error in the interval.
51. fx) = R, n = 3, a = 1, 1 <xs1.5. 52. fx) = e , n = 3, a = 0, -1 <xs1.
53. fr) = x sin x, n = 4, a = 0, - 1sxs1. 54. f(x) =xa,n = 4, a = 1, 0.5 x 1.5.
55. f(x) = 1/(1 - x), n = 3, a = 0, 0 <xs 0.25.

In problems 56 to 59, obtain the Taylor's polynomial approximation of degree n to the function fx)
about the point x = a. Find the error term and show that it tends to zero as n > co, Hence, write its
Taylor's series.
56. fx) = sin 3x, a = 0. 57. fx) = sin x, a = 0.
58. f(x) =Inx, a =1. 59. fx) = 2", a = 1.
60. Show that the number which occurs in the Taylor's formula with Lagrange form of remainder
(given in Eq. 1.12) after n terms approaches the limit 1/n + 1) as h > 0 providedfu -"x) is
continuous and not zero at x = a.

61. Find the number of terms that must be retained in the Taylor's polynomial approximation about the
pointx= 0 for the function cosh x in the interval [O,1] such that I Error I<0.001
62. Find the number of terms that must be retained in the Taylor's polynomial approximation about the
0 for the function sinx in the interval [0, 1] such that | Eror I< 0.0005.
point x = cos x
63. The function In (1 - ) is approximated about x = 0 by an nih degree Taylor's polynomial.

Find n such that | Error I< 0.1 on 0 x S 0.5.

64. The function sin?x is approximated by the first two non-zero terms in the Taylor's polynomial
expansion about the point x = 0. Find c such that| Error I < 0.005, when 0 < x <c.
65, The function tanx is approximatd by the first two non-zero terms in the Taylor's polynomial expansion
about the point x = 0. Find c such that lI Error | <
0.005, when 0 <x <<c.
06. Obtain the Maclaurin's series expansion of y() = e" ** .where m is a constant.

07. Using Taylor's series, find the approximate value of (i) VI5, (ii) sin 29°
Dlain the Taylor's series expansions as given in problem 68 to 75.

68. a =1 +x lna*2!a) (xIn a)

3!
 1.30 Engineering Mathematics

69. In (1
-x) = -x - -1 Sx<1.

-1 <x<1.

71. In x =F
2-11x-1 . x>0.

72 x - - (

73. tan

74. tan ! 1 + - 1
=

V1+x+y1-
75. sin

1.3
Integration and Its
Let fx) be defined and be Appllications
that F'x) = continuous on a closed
that if F(x) flx), SxSb.
a
is an Then, the function interval [a, b]. Let
anti-derivati
anti-derivative of flr). We of fx), F(x) is
called there exist
then F() + the function F(x) Sucn
a
write c, where c anti-derivative
is an of f(x). We observe
arbitrary constant
int, is also an
We note that,
not
every functi
fo) dx F() +c
=

is
integrable, for example
the (1.32)
x)=
fx)= x i s rational function fx)
xis irrational
does not have an defined on [0, 1] as
(a) Every anti-derivative
function which is not
(b) For
fx)
integrability,
may only be
ntinuous on a closed We note
conditio the
on that
integrable. the
and
(c) Let m and M be piecewise
the
f(x)
minimumcontinuous on
is
[a, b.
bounded
continuous interval is following
and
maximum values
on
\a, b}
can be integrable.
relax
laxed. The functi
mlb -a) s of
flx) on
s) dx [a, b].
sM(b -
Then,
a).
(1.33)