UPSC Civil Services Main 1990 - Mathematics

Calculus
Sunder Lal
Retired Professor of Mathematics
Panjab University
Chandigarh

January 14, 2010

Question 1(a) If a function f (x) of the real variable x has the first five derivatives 0 at a
given value x = a, show that it has a maximum or a minimum at x = a according as the
sixth derivative is negative or positive. What happens if only the first four derivatives are 0,
but not the fifth?

Solution. We shall take up the general case, which would cover both the cases
Let f (n) (a) exist and f (n) (a) 6= 0. Let f (r) (a) = 0, 1 r n 1. Then f (a) is not an
extreme value when n is odd, and if n is even, then f has a maximum at x = a if f (n) (a) < 0
and a minimum if f (n) (a) > 0.
Thus the above result is proved when n = 6. When n = 5, f (x) has neither maximum
nor minimum at x = a.
Proof: The existence of f (n) (a) imlies that f 0 (x), f 00 (x), . . . , f (n1) (x) all exist in a certain
neighborhood of a, say (a 1 , a + 1 ).
Case (1): If f (n) (a) > 0, then f (n1) (x) is increasing at a i.e. there exists 0 < < 1
such that f (n1) (x) < f (n1) (a) = 0 for a < x < a, and f (n1) (x) > f (n1) (a) = 0 for
a < x < a + .
Case (2): If f (n) (a) < 0, then f (n1) (x) is decreasing at a i.e. there exists 0 < < 1
such that f (n1) (x) > f (n1) (a) = 0 for a < x < a, and f (n1) (x) < f (n1) (a) = 0 for
a < x < a + .
Let h be such that |h| < , then by Taylors theorem

hn1 (n1)
f (a + h) = f (a) + hf 0 (a) + . . . + f (a + h), 0 < < 1
(n 1)!

hn1 (n1)
Thus f (a + h) f (a) = f (a + h) as f (r) (a) = 0, 1 r n 2. Since
(n 1)!
a + h (a , a + ), we have the following conclusions:

1
 1. n even and f (n) (a) > 0: If h < 0, then hn1 < 0 and f (n1) (a + h) < 0 (by Case
(1) above, as a < a + h < a. If h > 0, then hn1 > 0 and f (n1) (a + h) > 0 (as
a < a + h < a + .
hn1 (n1)
In either case f (a + h) > 0 for all h with |h| < , i.e. f (a + h) > f (a)
(n 1)!
for |h| < , thus f (x) has a minimum at x = a.
hn1 (n1)
2. n even and f (n) (a) < 0: Using Case (2), we get f (a + h) < 0 for all h
(n 1)!
with |h| < , i.e. f (a + h) < f (a) for |h| < , thus f (x) has a maximum at x = a.

3. n odd and f (n) (a) > 0: If h < 0, then hn1 > 0 and f (n1) (a + h) < 0, so f (a +
h) < f (a) for h < 0, |h| < . If h > 0, then hn1 > 0 and f (n1) (a + h) > 0, so
f (a + h) > f (a) for 0 < h < . Thus f has neither minimum nor maximum at x = a.
The case for n odd and f (n) (a) < 0 is similar h < 0 f (a + h) > f (a), h > 0
f (a + h) < f (a), so there is no extreme value at x = a.

This completes the proof.

Question 2(a) Show that

f (x) = (x 2)2 (x2 + 2bx + c)(x + 3)3

has a critical point at x = 1 2b + 5c = 7.

Solution.

f 0 (x) = 2(x 2)(x2 + 2bx + c)(x + 3)3 + (x 2)2 (2x + 2b)(x + 3)3
+ 3(x 2)2 (x2 + 2bx + c)(x + 3)2
f 0 (1) = 2(3)(1 2b + c)8 + 72(2 + 2b) + 108(1 2b + c)
= 48 + 96b 48c 144 + 144b + 108 216b + 108c = 84 + 24b + 60c

For 1 to be a critical point, f 0 (1) = 0 84 + 24b + 60c = 0 2b + 5c = 7.

Question 2(b) Assuming that the above condition is satisfied, examine the nature of f (x)
at its three critical points.

Solution.

f 0 (x) = 2(x2)(x2 +2bx+c)(x+3)3 +(x2)2 (2x+2b)(x+3)3 +3(x2)2 (x2 +2bx+c)(x+3)2

so it follows that the three critical points are x = 2, x = 3 and x = 1 when 2b + 5c = 7.

2
1. x = 2.
(x 2)2 (x2 + 2bx + c)(x + 3)3 = (x 2)2 g(x)
 
f (x) =
f 0 (x) = 2(x 2)g(x) + (x 2)2 g 0 (x)
f 00 (x) = 2g(x) + 4(x 2)g 0 (x) + (x 2)2 g 00 (x)
f 00 (2) = 2g(2) = 250(4 + 4b + c)
= 250(4 + 14 10c + c) = 250 9(2 c)
There is a maximum at x = 2 if 2 < c, and a minimum at x = 2 if 2 > c. The test
fails when c = 2. Considering f 000 (x) when x = 2, c = 2, 2b = 3,
f 000 (x) = 2g 0 (x) + 4g 0 (x) + 4(x 2)g 00 (x) + (x 2)2 g 000 (x) + 2(x 2)g 00 (x)
f 000 (2) = 6 (4 + 2b)(2 + 3)3 + 3(4 + 4b + c)(2 + 3)2
 
 
= 6 125 + 75(4 6 + 2) 6= 0
so there is no maximum or minimum at x = 2 when c = 2.
2. x = 3
(x + 3)3 (x 2)2 (x2 + 2bx + c) = (x + 3)3 h(x)
 
f (x) =
f 0 (x) = 3(x + 3)2 h(x) + (x + 3)3 h0 (x)
f 00 (x) = 6(x + 3)h(x) + 3(x + 3)2 h0 (x) + 3(x + 3)2 h0 (x) + (x + 3)3 h00 (x)
f 00 (3) = 0
f 000 (x) = 6h(x) + 12(x + 3)h0 (x) + (x + 3)2 [some polynomial in x]
f 000 (3) = 6h(3) = 150(9 6b + c) = 150(9 21 + 15c + c) = 600(4c 3)
So if c 6= 34 , f (x) does not have a maximum or minimum at x = 3. If c = 34 , then

f (4) (x) = 18h0 (x) + (x + 3)[some polynomial in x]

 
so f (4) (3) = 18 2(5)(9 6b + c) + 25(6 + 2b) = 18(240 + 110b 10c) < 0. Hence
if c = 43 , b = 13
8
, f (x) has a maximum at x = 3.
3. x = 1. Let y = x + 3. Then we need to consider the value y = 2.
f (y) = y 3 (y 5)2 (y 2 + (2b 6)y + 9 6b + c)
= y 3 (y 2 10y + 25)(y 2 + (1 5c)y 12 + 16c)
= y 7 + (1 5c 10)y 6 + (12 + 16c + 25 10 + 50c)y 5
+(120 160c + 25 125c)y 4 + (400c 300)y 3
= y 7 (9 + 5c)y 6 + (3 + 66c)y 5 + (145 285c)y 4 + (400c 300)y 3
f 00 (y) = 42y 5 30(9 + 5c)y 4 + 20(3 + 66c)y 3 + 12(145 285c)y 2 + 6(400c 300)y
f 00 (2) = 1344 480(9 + 5c) + 160(3 + 66c) + 48(145 285c) + 12(400c 300)
= 864 720c

3
 So if c > 65 , then f 00 (1) < 0 and f (x) has a maximum at x = 1. If c < 65 , f (x) has
a minimum at x = 1. If c = 56 , then this test fails, we consider higher derivatives at
y = 2 with c = 65 :

f 000 (y) = 210y 4 1800y 3 + 4932y 2 4728y + 1080

Now it is clear that f 000 (2) 6= 0, hence there is no maximum or minimum at x = 1

when c = 65 .

Question 2(c) Show that f (x) has at least three points of inflection irrespective of the
condition 2b + 5c = 7.

Solution. From the calculation of f 00 (3) = 0 whether 2b + 5c = 7 or not. Since f (3) =

f (2) = 0, there exists a real number b between 3 and 2 such that f 0 (b) = 0. Now f 0 (2) =
f 0 (b) = f 0 (3) = 0, thus we get a1 (3, b), a2 (b, 2) where f 00 (a1 ) = 0, f 00 (a2 ) = 0. Thus
there are least 3 points where f 00 (x) = 0.
Note: This question is not complete: A point of inflection is a point where the curvature
of a curve changes sign (the curve goes from convex to concave or vice versa). Any point a
is a point of inflection if f 00 (a) = 0 and the lowest order non-zero derivative at a is of odd
order, or equivalently, f 00 (a + ) and f 00 (a ) have opposite signs in the neighborhood of a.
Thus f 00 (x) = 0 is a necessary condition for a point to be an inflection point, but is not
sufficient. f 00 (x) is a 5th degree curve. Take two of the roots and equate them, and two other
roots and equate them too. We can thus get two equations in b, c, which can be solved to get
their values. Now f 00 (x) = A(x B)(x C)2 (x D)2 , where B, C, D are distinct. This curve
has only one point of inflection, B. For example, the similar curve g(x) = 35x3 21x5 + 5x7
has only one inflection point, although g 00 (x) has 3 roots.

Question 3(a) Prove that the n-th derivative of f (x)g(x) equals

n  
X n (ni)
f (x)g (i) (x)
i=0
i

n


where f (m) denotes the m-th derivative of f (x) and i
are the binomial coefficients.

Solution. The proof is by induction on n. It is true for n = 1, as (f g)0 = f 0 g + f g 0 =

1
f (1) (x)g (0) (x) + 11 f (0) (x)g (1) (x) where f (0) (x) = f (x).



0
Induction hypothesis: Assume that the result is true for n = m.
m  
(m)
X m (mi)
(f g) = f (x)g (i) (x)
i=0
i

4
Differentiating both sides,
m  
(m+1)
X m 
(f g) (x) = f (mi+1) (x)g (i) (x) + f (mi) (x)g (i+1) (x)
i=0
i
  m h   
m (m+1) (0)
X m m i (mi+1)
= f (x)g (x) + + f (x)g (i) (x)
0 i=1
i i 1
 
m (0)
+ f (x)g (m+1) (x)
m
  m  
m + 1 (m+1) (0)
X m + 1 (mi+1)
= f (x)g (x) + f (x)g (i) (x)
0 i=1
i
 
m + 1 (0)
+ f (x)g (m+1) (x)
m+1
m+1
X m + 1
= f (m+1i) (x)g (i) (x)
i=0
i

We used m0 = 1 = m+1


m
m+1

m
m

m+1


0
, m
= 1 = m+1
, i
+ i1
= i
. The result is now
established for n = m + 1, and hence by induction for all n.

Question 3(b) Show that the m-th derivative gm (x) of g(x) = tan1 x satisfies

(1 + x2 )gm+1 (x) + 2mxgm (x) + m(m 1)gm1 (x) = 0

and hence is of the form

(1)m1 (m 1)!m (x)
gm (x) =
(1 + x2 )m
where m (x) is a polynomial of dergree m given by 1 (x) = 1, 2 (x) = 2x and the recursion
m+1 (x) = 2xm (x) (1 + x2 )m1 (x).

1
Solution. g1 (x) = or (1 + x2 )g1 (x) = 1.
1 + x2
Using Leibnitzs theorem proved above and differentiating the above equation m times,
we get      
m 2 m m
(1 + x )gm+1 (x) + 2xgm (x) + 2gm1 (x) = 0
0 1 2
or
(1 + x2 )gm+1 (x) + 2mxgm (x) + m(m 1)gm1 (x) = 0
which is the required relation.
1 (1)11 (1 1)!1 (x)
Now g1 (x) = = as 1 (x) = 1.
1 + x2 (1 + x2 )1

5
 2x (1)21 (2 1)!2 (x)
g2 (x) = = as 2 (x) = 2x.
(1 + x2 )2 (1 + x2 )2
Assume that the result is true for all n m. From the relation proved above:

(1)m1 (m 1)!m (x) (1)m2 (m 2)!m1 (x)

(1 + x2 )gm+1 (x) + 2mx + m(m 1) =0
(1 + x2 )m (1 + x2 )m1
or
(1)m (m 1)!m (x) (1)m1 (m 2)!m1 (x)
gm+1 (x) = 2mx + m(m 1)
(1 + x2 )m+1 (1 + x2 )m
m
(1) m! h i
= 2 m+1
2xm (x) (1 + x2 )m1 (x)
(1 + x )
(1)m m!
= m+1 (x)
(1 + x2 )m+1

as m+1 (x) = 2xm (x) (1 + x2 )m1 (x). This completes the proof.

Question 3(c) Find m (x) for x 6. Can you find a general formula?

Solution.

1 (x) = 1
2 (x) = 2x
3 (x) = 2x 2x (1 + x2 ) = 3x2 1
4 (x) = 2x(3x2 1) (1 + x2 )(2x) = 4x3 4x
5 (x) = 2x(4x3 4x) (1 + x2 )(3x2 1) = 5x4 10x2 + 1
6 (x) = 2x(5x4 10x2 + 1) (1 + x2 )(4x3 4x)
= 6x5 20x3 + 6x

Writing these down using binomial coefficients:

 
1 0
1 (x) = x
0
 
2 1
2 (x) = x
1
   
3 2 3 0
3 (x) = x x
2 0
   
4 3 4 1
4 (x) = x x
3 1

6
      
5 4 5 2 5 0
5 (x) = x x + x
4 2 0
     
6 5 6 3 6 1
6 (x) = x x + x
5 3 1
     
n n1 n n3 n
n (x) = x x + xn5 . . .
n1 n3 n5
2r+1n   2r+1n  
X
r n n2r1
X
r n
= (1) x = (1) xn2r1
r=0
n 2r 1 r=0
2r + 1

Question 3(d) Show that the n-th derivative of ex tan1 x at x = 0 equals

     
n n n
n 2! + 4! 6! + ...
3 5 7

Solution. Let g(x) = tan1 (x). Using Leibnitzs theorem, we get

n   n  
dn x 1
X n dnr ex X n
(e tan x) = gr (x) nr = gr (x)ex
dxn r=0
r dx r=0
r

Thus n   n  
dn x

1
X n X n
(e tan x) = g r (0) = (1)r1 (r 1)!r (0)
dxn x=0 r=0
r r=1
r
as g(0) = 0 and gr (x) was computed above in terms of r (x). Now 2r (0) = 0, and 2r+1 =
(1)r . Thus
2r+1n
dn x
 X  n 
1
(e tan x) = (2r)!(1)r
dxn x=0 r=0
2r + 1
or
dn x
        
1 n n n n
n
(e tan x) = 0! 2! + 4! 6! + ...
dx x=0 1 3 5 7
as required.

Paper II
Z
2
Question 4(a) Discuss the convergence of the integral log(sin x) dx and if convergent,
0
evaluate it.

7
 Solution. The integrand log sin x has a discontinuity only at x = 0. We consider the
Z
2
integral log(sin x) dx so that the integrand is positive.
0
Let g(x) = xm , 0 < m < 1, then

log sin x cos x 1 xm cos x x

lim = lim+ = lim =0
x0+ xm x0 sin x mxm1 x0+ m sin x
as LHospitals rule applies.
Z Z
2 2
But g(x) dx is convergent as 0 < m < 1, therefore log(sin x) dx converges. Now
0 0
Z Z Z Z
2 2 2 2
log(sin 2x) dx = log 2 dx + log(cos x) dx + log(sin x) dx
0 0 0 0
Z Z
2 2
But log(cos x) dx = log(sin x) dx, so
0 0
Z Z
2 2
2 log(sin x) dx = log 2 + log(sin 2x) dx
0 2 0

By putting 2x = t.
Z Z
2 1
log(sin 2x) dx = log(sin t) dt
0 2 0
Now substituting t = y, we have

Z Z 0 Z
2
log(sin t) dt = log(sin y) (dy) = log(sin y) dy

2 2
0
Z Z
2 2
log(sin 2x) dx = log(sin y) dy
0 0
Z
2
log(sin x) dx = log 2
0 2

2
Question 4(b) Find the point on the parabola
y = 2x, z = 0 which is nearest to the plane
z = x + 2y + 8. Show that this distance is 6.

Solution. Let the point (x, y, 0) by on the parabola. Then the distance d from the plane
z = x + 2y + 8 is given by

|x + 2y 0 + 8| |x + 2y + 8|
d= =
1+4+1 6

8
 6d2 = x2 + 4xy + 4y 2 + 16x + 32y + 64.
Put 2x = y 2 , and let F = 6d2 to get

y4 2
F = 6d = + 2y 3 + 4y 2 + 8y 2 + 32y + 64
4
We need to minimize F . The critical points are given by
dF
= y 3 + 6y 2 + 24y + 32 = 0
dy
(y + 2)(y 2 + 4y + 16) = 0

Thus y = 2 is the only real value.

d2 F
= 3y 2 + 12y + 24 = 12 24 + 24 > 0
dy 2
at y = 2. Hence y = 2 is a minimum.
When y = 2, x = 2, and d = |24+8|

6
= 6 as required.

Question 4(c) Show that

ZZZ
1
(a2 b2 c2 b2 c2 x2 c2 a2 y 2 a2 b2 z 2 ) 2 dx dy dz

x2 y 2 z 2 2 a2 b 2 c 2
over the volume bounded by 2 + 2 + 2 = 1 is equal to .
a b c 4
Solution. Let
ZZZ
1
I = 2
(a2 b2 c2 b2 c2 x2 c2 a2 y 2 a2 b2 z 2 ) 2 dx dy dz
x2 2
{(x,y,z) | + y2 + z2 1}
a2 b c

x2 y 2 z 2  21
ZZZ 
= 8abc 1 2 2 2 dx dy dz
{(x,y,z) | x2 2 2
+ y2 + z2 1,x0,y0,z0} a b c
a2 b c

Let x = aX, y = bY, z = cZ, so that dx dy dz = abc dX dY dZ

ZZZ
2 2 2 2 2 2 21
I = 8a b c 2 2 2
(1 X Y Z ) dX dY dZ
X +Y +Z 1
X0,Y 0,Z0

du dv dw
Let X 2 = u, Y 2 = v, Z 2 = w, so that dX = , dY
2 u
= , dZ
2 v
=
2 w
and
ZZZ
1 1 1 1
2 2 2
I=a b c (1 u v w) 2 u 2 v 2 w 2 du dv dw
u+v+w1
u0,v0,w0

9
 We now convert this to a Dirichlet integral let u + v + w = , v + w = , w = ,
(u, v, w)
so that = 2 and
(, , )
Z 1Z 1Z 1
1 1 1 1
I = abc 2 2 2
(1 ) 2 ( ) 2 ( ) 2 () 2 2 d d d
Z0 1 Z0 1 Z0 1
1 1 1 1 1
= a2 b 2 c 2 (1 ) 2 2 (1 ) 2 (1 ) 2 2 d d d
0 0 0
3 3
2 2 2 ( 2 )( 2 ) ( 21 )(1) ( 21 )( 21 )
= abc
(3) ( 32 ) (1)
1 1 4
2 2 2 2 ( 2 ) 2 a2 b 2 c 2
= abc =
2 4

as required, as ( 12 ) = .

10