EDUCATIVE COMMENTARY ON

JEE 2010 MATHEMATICS PAPERS

(Last updated on March 20, 2021)

Contents

Paper 1 3

Paper 2 40

Concluding Remarks 76

The pattern of JEE 2010 resembles closely that of the previous year. The
proportion of questions with single digit answers has gone up. These ques-
tions are intended to make up for the inherent limitations of multiple choice
questions. But in a way these are multiple choice questions too except that
for each question there are 10 possible choices. Also the requirement that
the answer has to be a single digit sometimes entails some clumsiness, such
as having to multiply some function in the question by a weird factor.
Negative credit for wrong answers was dropped this year. Another novel
feature was that in those questions where more than one answers is correct,
a candidate who chooses some of these (but no wrong ones) would get partial
credit proportional to the number of correct choices he has made. But this
instruction was not properly worded and, in absence of any negative credit,
could mean that a candidate could score full marks merely by darkening all
bubbles!
There was some confusion about the numbering of the questions in the
question papers and that in the Objective Response Sheet. In this commen-
tary, questions in Paper I will be numbered serially from 1 to 28 and those
in Paper II from 1 to 19.
In the commentaries on the JEE mathematics papers for the past several
years, frequent references were made to the book Educative JEE Mathematics
by the author. At the time of writing the present commentary, this book was
out of print. The second edition has appeared subsequently but has a slightly
different pagination. As a result, in this year’s commentary, references to the
book are given in terms of the numbers of the chapter, the comment or the
exercise involved, rather than by page numbers for the convenience of those
having either of the two editions.
In the earlier vesrion of this Commentary, in the solution to Q.28, the sum
S1 was wrongly taken as 1 instead of 0 (and hence the answer, too, was given
as 4 instead of 3). The error was pointed out by Amit Singhal.

2
 PAPER 1

Contents

Section I (Single Correct Choice Type) 3

Section II (Multiple Correct Choice Type) 9

Section III (Paragraph Type) 16

Section IV (Integer Type) 22

SECTION I
Single Correct Choice Type

This section contains 8 multiple choice questions. Each question has 4

choices out of which ONLY ONE is correct.

Q.1 Let p and q be real numbers such that p 6= 0 and p3 6= ±q. Let α and β
be non-zero complex numbers satisfying α + β = −p and α3 + β 3 = q.
Then a quadratic equation having α/β and β/α as its roots is

(A) (p3 + q)x2 − (p3 + 2q)x + (p3 + q) = 0

(B) (p3 + q)x2 − (p3 − 2q)x + (p3 + q) = 0
(C) (p3 − q)x2 − (5p3 − 2q)x + (p3 − q) = 0
(D) (p3 − q)x2 − (5p3 + 2q)x + (p3 − q) = 0

Answer and Comments: (B). A straightforward question about the

roots of a quadratic equation. The product of the two roots of the
α2 + β 2
desired quadratic is 1 while their sum is . So the values of
αβ
α2 + β 2 and αβ have to be determined from those of α + β and α3 + β 3
which are given to be −p and q respectively. Since α3 + β 3 = (α +
β)(α2 + αβ + β 2 ) = (α + β)((α + β)2 − 3αβ), we get q = −p(p2 − 3αβ)
p3 + q
from which it follows that αβ = . From this we also get α2 +β 2 =
3p

3
 p3 + q p3 − 2q
(α+β)2 −2αβ = p2 −2( )= . Hence the desired quadratic
3p 3p
is
p3 − 2q
x2 − x+1 =0
p3 + q
which gives (B). Although the calculations are easy, it is doubtful if
one can do them in 2 minutes which is the time allowed.
x y z
Q.2 Equation of the plane containing the straight line = = and per-
2 3 4
x y z
pendicular to the plane containing the straight lines = = and
3 4 2
x y z
= = is
4 2 3

(A) x + 2y − 2z = 0 (B) 3x + 2y − 2z = 0
(C) x − 2y + z = 0 (B) 5x + 2y − 4z = 0

Answer and Comments: (C). An extremely straightforward ques-

tion. As the desired plane passes through the origin (a point on the
first line) its equation is of the form ax + by + cz = 0. To determine
the coefficients a, b, c upto a constant multiple, we need two equations.
The first one comes from the fact that this plane contains the first line
with direction numbers 2, 3, 4. This gives

2a + 3b + 4c = 0 (1)

The plane containing the other two lines will have its normal along the
vector (3i + 4j + 2k) × (4i + 2j + 3k) which comes out to be 8i − j − 10k.
The normal to the desired plane lies along the vector ai + bj + ck.
Perpendicularity of the two planes gives the perpendicularity of their
normals and hence

8a − b − 10c = 0 (2)

(1) and (2) together give a : b : c = −26 : 52 : −26. Canceling the

factor −26, we can take a = 1, b = −2 and c = 1. Hence the desired
plane is x − 2y + z = 0. Note that we only need the direction of the
normal to the second plane. So, it is a waste of time to actually identify
the equation of the second plane. This is perhaps the only alertness
needed. Otherwise the question is extremely routine. The portion of

4
 solid geometry in the JEE syllabus is so limited and elementary that it
is difficult for the examiners to come up with a really novel question.

Q.3 Let ω be a complex root of unity with ω 6= 1. A fair die is thrown three
times. If r1 , r2 , r3 are the numbers obtained on the die, the probability
that ω r1 + ω r2 + ω r3 = 0 is

(A) 1/18 (B) 1/9 (C) 2/9 (D) 1/36

Answer and Comments: (C) The problem is a combination of com-

plex numbers and elementary probability. The possible values of each
ri are from 1 to 6. Hence there are in all 63 possibilities, each corre-
sponding to an ordered triple (r1 , r2 , r3 ) of integers from 1 to 6. We
have to first see for which of these the equation ω r1 + ω r2 + ω r3 = 0
holds. The basic quadratic satisfied by ω is ω 2 + ω + 1 = 0. We can
multiply this throughout by any powers of ω. Moreover since ω 3 = 1,
we can multiply any of the terms by ω 3 or any power of it to get another
sum of three powers of ω that add to 0. Thus within the permissible
values from 1 to 6, we see that one ri has to be 1 or 4, another 2 or 5
and the third one either 3 or 6. Thus each ri has two possible values
and since they are all distinct, they can be permuted among themselves
in 6 ways. Thus the number of favourable cases is 23 × 3!. Therefore
8×6 8 2
the desired probability is 3
= = .
6 36 9
The problem is a good combination of two essentially unrelated
branches of mathematics. Earlier in 1997, there was a problem which
asked for the probability that the roots of a certain quadratic equation
be real (see Exercise (22.16) (b)).

Q.4 If the angles A, B, C of a triangle are in an arithmetic progression and

if a, b, c denote the lengths of the sides opposite to A, B, C respectively,
a c
then the value of the expression sin 2C + sin 2A is
c a
√ √
(A) 1/2 (B) 3/2 (C) 1 (D) 3

Answer and Comments: (D). By the sine rule, we can replace the
a c
ratios and by the ratios of the sines of the opposite angles. Then the
c a
given expression becomes simply 2(sin A cos C+cos A sin C) = 2 sin(A+

5
 π
C) = 2 sin B. But as the angles are in an A.P., 6 B = . So the given
√ 3
π
expression equals 2 sin( ) = 3.
3
Questions about triangles with angles in an A.P. have been asked
several times. This particular one is very easy the moment the key
idea, viz. the sine rule, strikes.

Q.5 Let P, Q, R, S be the points on the plane with position vectors −2i −
j, 4i, 3i + 3j and −3i + 2j. Then the quadrilateral P QRS must be a

(A) parallelogram, which is neither a rhombus nor a rectangle

(B) square
(C) rectangle, but not a square
(D) rhombus, but not a square

Answer and Comments: (A). The points P, Q, R, S could have as

well been specified by coordinates. The very fact that their position
vectors are given suggests that vector methods may be handy. By
direct calculations, the sides P Q and SR are represented by the vectors
4i − (−2i − j) = 6i + j and 3i + 3j − (−3i + 2j) = 6i + j. Hence the
opposite sides are parallel and equal. So P QRS is a parallelogram. To
see if it is a rhombus or a rectangle we need to calculate the other sides
(actually either one would do). So, QR is represented by the vector
3i + 3j − 4i = −i + 3j. As this vector has a different length than and is
not perpendicular to the vector represented by the other pair of sides
(as can be seen from their dot product), we see that the quadrilateral
is neither a rhombus nor a a rectangle.
Had the problem been stated in terms of the coordinates of the points
P, Q, R, S, the work would have been essentially the same. But in that
case an alternate approach would have been more tempting. That is,
first see if the diagonals have the same mid-point. In the present case,
the point ( 12 , 1) is indeed the midpoint of both the diagonals. So the
quadrilateral is a parallelogram. Further, the lengths of the diagonals
are different and so it cannot be a rectangle. Finally, the diagonals are
not perpendicular to each other and so P QRS is not a rhombus either.
A very simple problem, no matter which approach is taken.

6
 1 x t ln t
Z
Q.6 The value of lim dt is
x→0 x3 0 t4 + 1
(A) 0 (B) 1/12 (C) 1/24 (D) 1/64

Answer and Comments: (B). Problems of this type have been asked
several times. The key idea is not to evaluate the integral. Our interest
is only in the limit and that too the limit of a ratio, for which other
methods are often available. In fact, in the present problem, both
t ln(1 + t)
Z x
the numerator (viz. the integral dt) and the denominator
0 t4 + 4
0
(viz. x3 ) tend to 0 as x → 0. So this is a limit of the form and
0
therefore L’hôpital’s rule is a popular tool. To take the derivative of
the numerator, we need the second form of the Fundamental Theorem
of Calculus, which gives

d x t ln(1 + t) x ln(1 + x)
Z
4
dt = (1)
dx 0 t +4 x4 + 4
x ln(1 + x) 1 ln(1 + x)
Hence the given limit equals lim 2 4
= lim . The
3x (x + 4)
x→0 12 x→0 x
fans of L’ôpital’s rule can apply it again to evaluate the last limit. But
ln(1 + x) − ln(1 + 0)
that is hardly necessary because if we write it as lim
x→0 x−0
we see that it is simply the derivative of ln(1 + x) at x = 0 and hence
1 1
equals = 1. Whichever way we go, the given limit is .
1+0 12
The problem resembles Exercise (17.18), especially Part (ii) of it.

Q.7 Let f (x), g(x), h(x) be real-valued functions defined on the interval [0, 1]
2 2 2 2 2 2
by f (x) = ex + e−x , g(x) = xex + e−x and h(x) = x2 ex + e−x . If
a, b and c denote, respectively, the absolute maxima of f, g and h on
[0, 1], then

(A) a = b and c 6= b (B) a = c and a 6= b

6 b and c 6= b
(C) a = (D) a = b = c

Answer and Comments: (D). All the three functions are differen-
tiable everywhere. Therefore the easiest way to find their extrema on

7
 [0, 1] is by studying the signs of their derivatives. A direct calculation
gives
2
′ x2 2x(e2x − 1)
−x2
f (x) = 2xe − 2xe = (1)
ex2
2
2 2 2 (2x2 + 1)e2x − 2x
g ′ (x) = 2x2 ex + ex − 2xe−x = (2)
ex2
2
′ x2 3 x2 −x2 (2x + 2x3 )e2x − 2x
and h (x) = 2xe + 2x e − 2xe = (3)
ex2
The denominators of the last expressions in each formula are all pos-
itive for all x and so their signs are determined from those of their
2
numerators. In each case, we have e2x > 1 for all x > 0. This immedi-
ately implies that f ′ (x) > 0 for all x > 0. Hence by Lagrange’s Mean
Value Theorem, f (x) is strictly increasing on [0, 1]. So, its maximum
on [0, 1] occurs at the end point 1. In (2), for x > 0, the numerator is
at least 2x2 + 1 − 2x = x2 + (x − 1)2 which is always positive. So, g(x)
is also strictly increasing on [0, 1]. In (3) too, the numerator is at least
(2x+2x3)−2x = 2x3 and hence is positive for x > 0. Hence all the three
functions attain their maxima on [0, 1] at 1. Since f (1) = g(1) = h(1)
we see that a = b = c.
The problem is a combination of maxima/minima and inequalities.
Although all the three functions are strictly increasing, the arguments
needed to show this are slightly different. Since no reasoning is to
be given, an unscrupulous student who simply assumes this saves a
lot of time as compared to a sincere student who carefully analyses the
expressions as above and shows that they are all positive. To reward the
sincere student it would have been better if one of the three functions
was so designed that its maximum did not occur at the end.
Q.8 The number of 3 × 3 matrices
 A whose
  entries are either 0 or 1 and for
x 1
which the system A  y  =  0  has exactly two distinct solutions
   

z 0
is

(A) 0 (B) 29 − 1 (C) 168 (D) 2

Answer and Comments: (A). This is one of those questions where

8
 some irrelevant pieces of data are deliberately inserted. The given sys-
tem of equations is a non-homogeneous system of three linear equations
in three unknowns. Such a system has either a unique solution, or no
solution or infinitely many solutions, depending on the rank of the ma-
trix A and the rank of the augmented 3 × 4 matrix obtained by adding
one more column to A. It can never have exactly two solutions. The
fact that the entries of the matrix A are either 0 or 1 has no bearing
on the question. It is just meant to fool the undiscerning candidates
into counting the number of such matrices (which comes out to be 29 ).
Those not familiar with properties of matrices can do the problem ge-
ometrically. Each equation represents a plane in the three dimensional
space. The solutions of the system correspond to points of intersection
of these three planes. But no three planes can meet in only two points.

SECTION II
Multiple Correct Choice Type

This section contains five multiple choice questions. Each question has
four choices out of which ONE OR MORE may be correct.

Q.9 Let A and B be two distinct points on the parabola y 2 = 4x. If the axis
of the parabola touches a circle of radius r having AB as a diameter,
then the slope of the line joining A and B can be

(A) −1/r (B) 1/r (C) 2/r (D) −2/r

Answer and Comments: (C), (D). In such problems it is convenient

to take the parametric representation of the parabola, viz. x = t2 , y =
2t. Let the points A and B be (t21 , 2t1 ) and (t22 , 2t2 ). Then the slope,
say m, of AB is given by

2(t1 − t2 ) 2
m= 2 2
= (1)
(t1 − t2 ) t1 + t2

We are given that the circle with AB as a diameter touches the axis
of the parabola, viz. the x-axis. This means that the midpoint of AB

9
 is at a distance r from the x-axis. Hence its y-coordinate is ±r. This
gives
(t1 + t2 ) = ±r (2)
2
(1) and (2) together imply that m = ± . So (C) as well as (D) can be
r
correct.
A very simple problem once the key idea, viz. using parametric repre-
sentation strikes. Even if A, B are taken as (x1 , y1 ) and (x2 , y2), the
calculations are not so prohibitive since in that case we would get
y1 − y2 y12 − y22 4
m= = = (3)
x1 − x2 (x1 − x2 )(y1 + y2 ) y1 + y2
y1 + y2
Moreover the y coordinate of the midpoint of AB is . And
2
equating this with ±r and using (3) will give the same answer.
1 x4 (1 − x)4
Z
Q.10 The value(s) of dx is (are)
0 1 + x2
22 71 3π
(A) −π (B) 2/105 (C) 0 (D) −
7 15 2

Answer and Comments: (A). The substitution x = tan θ is tempt-

ing because the denominator of the integrand is 1 + x2 . But that leads
to a complicated integrand involving high powers of cos θ in the de-
nominator. The best method to evaluate the integral is by applying
long division to the integrand. A direct expansion using the binomial
theorem gives
x4 (1 − x)4 x8 − 4x7 + 6x6 − 4x5 + x4
=
1 + x2 1 + x2
(x + x ) − 4(x7 + x5 ) + 5(x6 + x4 ) − 4(x4 + x2 ) + 4(x2 + 1) − 4
8 6
=
1 + x2
4
= x6 − 4x5 + 5x4 − 4x2 + 4 − (1)
1 + x2
Term by term integration gives
1 x4 (1 − x)4 1 4 4
Z
2
dx = − + 1 − + 4 − 4 tan−1 1
0 1+x 7 6 3

10
 1 π
= +3−4×
7 4
22
= −π (2)
7
22
So, out of the given options, only (A) is correct. If we accept as
7
an approximate value of π then (C) is also (approximately) correct.
Apparently, the intention behind giving the choice (C) is to test if the
22
candidate knows that π does not equal . (In fact, π is not a rational
7
number.) The trouble is that a candidate has no way to know this.
22
Outside mathematics, π is often taken to equal . Even in mathe-
7
matics, in numerical problems such as those involving mensuration it
22
is not uncommon to replace π by to get the answer in a numerical
7
form. (This is also the reason why in numerical problems the radii are
often multiples of 7.) Since the possibility of more than one answer be-
ing correct is not ruled out, even a good student is likely to mark both
(A) and (C) correct, because he may think that the paper-setters want
22
to test if he knows that π is approximately . In that case, he will
7
be unduly penalised. It would have been better to put this question
in Section I where there is only one correct answer. That would not
confuse anybody. (Apparently, the problem is a replica of a Putnam
1968 problem.)
Q.11 Let ABC be a triangle such that 6 ACB = π/6 and let a, b, c denote
the lengths of the sides opposite to A, B, C respectively. The value(s)
of x for which a = x2 + x + 1, b = x2 − 1 and c = 2x + 1 is (are)
√ √ √ √
(A) −(2 + 3) (B) 1 + 3 (C) 2 + 3 (D) 4 3

Answer and Comments: (B). Using the cosine formula we get

a2 + b2 − c2
cos C =
2ab
(x2 + x + 1)2 + (x2 − 1)2 − (2x + 1)2
= (1)
2(x2 + x + 1)(x2 − 1)
√
Since 6 C is given to be π/6 and cos(π/6) = 3/2, (1) gives us an

11
equation for x, viz.
√
(x2 + x + 1)2 + (x2 − 1)2 − (2x + 1)2 = 3(x2 + x + 1)(x2 − 1) (2)

This is an equation of degree 4 and in general not easy to solve. But

if we combine the first and the third term on the L.H.S. it factors as
(x2 + 3x + 2)(x2 − x) and further as (x + 1)(x + 2)x(x − 1). Thus each
side of (2) has x2 − 1 as a factor. Moreover this factor is non-zero since
x2 = 1 would mean that the side b is 0. So, canceling this factor, (2)
reduces to a quadratic
√
x2 + 2x + x2 − 1 = 3(x2 + x + 1) (3)

i.e.
√ √ √
(2 − 3)x2 + (2 − 3) − (1 +
3) = 0 (4)
√
Before solving this, we
√ divide it throughout
√ by 2 − 3 and note that
the reciprocal of 2 − 3 is simply 2 + 3. So, (5) becomes
√
x2 + x − (5 + 3 3) = 0 (5)
q √
−1 ± 21 + 12 3
whose roots are . The negative sign is ruled out
2
as that would make the side c of the triangle negative. The problem
is thusqreduced to checking which of the four given numbers equals
√
−1 + 21 + 12 3
. This can be done by checking these numbers one-
2 √ √
by-one. Butqif we observe that 21 + 12 3 is simply (3 + 2 3)2 , we see
√
−1 + 21 + 12 3 √
that equals 1 + 3. Hence (B) is the only correct
2
answer.
Although superficially a problem in trigonometry, the trigonomet-
rical part of the problem is very elementary. The real problem lies in
simplification of polynomials and surds. It is not immediately obvious
that x2 − 1 is a factor of both the sides of (2). Nowadays questions on
surds are also√ not very common and therefore it is not easy to strike
that 21 + 12 3 is the square of some simple surd. Both these factors
make the problem very tricky. Even if a candidate gets these tricks

12
 quickly, the rest of the work is also considerable and can hardly be
expected in the proportionate time for the question. It may be argued
that the other trigonometric question about triangles, viz. Q.4, was
exceptionally simple and hence the two questions balance each other.
Another possible solution is based on the identity

(x2 − 1)2 + (2x + 1)2 − (x2 + x + 1)2 = −(x2 − 1)(2x + 1) (6)

1
which immediately implies that cos A = − and hence that 6 A = 2π/3.
2
As 6 C is given to equal π/6, we see that the remaining angle 6 B also
equals π/6. Therefore the triangle ABC is isosceles with b = c. This
gives us a much simpler quadratic
√ for x, viz. x2 − 2x − 2 = 0, solving
which we get that x = 1 + 3. It is not difficult to prove (6), the
key idea once gain being that x2 − 1 is a factor of both the sides. But
this is certainly not one of the standard identities and it is doubtful
if anybody would think of it as a first step to the solution. There are
some identities such as

(x2 − 1)2 + (2x)2 = (x2 + 1)2 (7)

which implies that a triangle with sides x2 − 1, 2x and x2 + 1 is a right

angled one. This is quite well-known and is indeed used in determining
all Pythagorean triples of integers. But (6) is far from well-known. In
a conventional examination, the question could have asked a candidate
to first show that under the hypothesis, the triangle is isosceles and
then to determine the value of x. The candidate would then begin
by showing that the remaining angles are either π/6 and 2π/3 or else
5π/12 each. The first possibility is simpler and that would lead him to
come up with (6). But as the question stands, the approach we have
given is the only natural one and even then the problem is very tricky.

Q.12 Let z1 and z2 be two distinct complex numbers and let z = (1−t)z1 +tz2
for some real number t with 0 < t < 1. If Arg (w) denotes the principal
argument of a non-zero complex number w, then

(A) |z − z1 | + |z − z2 | = |z1 − z2 | (B) Arg(z − z1 ) = Arg(z − z2 )

z − z1 z − z1
(C) =0 (D) Arg(z − z1 ) = Arg(z2 − z1 )
z2 − z1 z2 − z1

13
 Answer and Comments: (A), (C), (D). A complex number of the
form (1 − t)z1 + tz2 (with t real) represents a point on the line passing
through z1 and z2 . The points z1 and z2 correspond to t = 0 and t = 1,
while for 0 < t < 1, the corresponding point lies on the segment from
z1 to z2 . (See Comment No. 16 of Chapter 8 and also Comment No.
15 for what happens if t is complex.) If this is understood then (A) is
obvious. The direction of the ray from z to z1 is opposite to that of
the ray from z to z2 but is the same as the ray from z2 to z1 . Even
without explicit calculations, this shows that (B) is false while (D) is
true. This also implies that z − z1 = λ(z2 − z1 ) for some real λ. But
then we also have z − z1 = λ(z2 − z1 ) = λ(z2 − z1 ). Hence the first row
of the determinant in (C) is λ times the second. So the determinant
vanishes. (Note that for (C) we only need collinearity and not that
0 < t < 1.)
This is a good problem because once the essential idea, viz. collinear-
ity of z1 , z2 and z (with z lying in between z1 and z2 ), is understood,
no computations are needed. The parameter t has the significance that
the point z divides the segment in the ratio t : 1 − t. This is known
as the Section Formula. But that is nowhere needed in the present
problem.
Q.13 Let f be a real valued function defined on the interval (0, ∞) by
Z x √
f (x) = ln x + 1 + sin t dt
0

Then which of the following statements is (are) true?

(A) f ′′ (x) exists for all x ∈ (0, ∞)
(B) f ′ (x) exists for all x ∈ (0, ∞) and f ′ is continuous on (0, ∞), but
not differentiable on (0, ∞)
(C) there exists α > 1 such that |f ′ (x)| < |f (x)| for all x ∈ (α, ∞)
(D) there exists β > 0 such that |f (x)| + |f ′(x)| ≤ β for all x ∈ (0, ∞).

Answer and Comments: (B), (C). By the second fundamental the-

orem of calculus, we have
1 √
f ′ (x) = + 1 + sin x (1)
x
14
for all x ∈ (0, ∞). Note that the square root function is differentiable
when the argument is positive but not (right) differentiable at 0, even
though it is continuous there. As a result the second term on the R.H.S.
of (1) is not differentiable at points x for which sin x = −1. But it is
continuous everywhere as is also the first term. So, (A) is false but√
(B) holds. For (C) note that for x ≥ 1, |f ′(x)| = f ′ (x) ≤ 1 + 2
which is a fixed number. But f (x) is strictly increasing and tends to
∞ as x → ∞. √ So, there will be some α such that for every x > α,
f (x) > 1 + 2. For any such α, (C) holds. We also have that f (x)
is unbounded and so (D) is false as otherwise both f and f ′ would be
bounded.

15
 SECTION III
Paragraph Type

This section contains two paragraphs. There are two multiple choice ques-
tions based on the first paragraph and three on the second. All questions
have ONLY ONE correct answer.

Paragraph for Q. 14 and 15

2 2 x2 y2
The circle x + y − 8x = 0 and the hyperbola − = 1 intersect at
9 4
the points A and B.

Q.14 Equation of a common tangent with positive slope to the circle as well
as to the hyperbola is
√ √
(A) 2x − 5y − 20 = 0 (B) 2x − 5y + 4 = 0
(C) 3x − 4y + 8 = 0 (D) 4x − 3y + 4 = 0

Answer and Comments: (B). The easiest way is to begin by assum-

ing that the common tangent has an equation of the form

y = mx + c (1)

The condition for tangency to each curve will give an equation in m

and c. Eliminating c from these two equations, we shall get an equation
for m which we then have to solve.
To begin, (1) is a tangent to the hyperbola if and only if

c2 = 9m2 − 4 (2)

(This is a standard result. It is obtained by putting y = mx+c into the

equation of the hyperbola and then writing down the condition that
the resulting quadratic in x has coinciding roots.)
We can similarly derive a condition that y = mx + c touch the circle
x2 + y 2 − 8x = 0. But it is easier to do this geometrically. Completing
squares, the centre of the circle is at (4, 0) while its radius is 4. So

16
equating the perpendicular distance of the centre from the line y =
mx + c with the radius we get
4m + c
±√ 2 =4 (3)
m +1
which upon squaring gives

c2 + 8mc = 16 (4)

To eliminate c between (2) and (4), we first put (2) into (4) to get

8mc = 20 − 9m2 (5)

Squaring and again substituting from (2) gives 64m2 (9m2 − 4) = (20 −
9m2 )2 which, upon simplification becomes

495m4 + 104m2 − 400 = 0 (6)

Although superficially this is a fourth degree equation in m, there are

no odd degree terms. So we put u = m2 to get

495u2 + 104u − 400 = 0 (7)

which is a quadraticqin u. We want only the positive root of this which

√
−52 + (52)2 + 495 × 400 −52 + 4 169 + 495 × 25
comes out as , i.e. as .
495 495
The arithmetic involved is rather prohibitive. But if carried out, we get
4
u = . Recalling that u = m2 and we want only the positive value of
5
2
m, we get that the slope of the common tangent is √ . As we already
5
4
know 8mc = 20 − 9m2 we get the value of c as √ . Therefore the
5
2x + 4 √
common tangent y = mx + c becomes y = √ or 2x − 5y + 4 = 0.
5
The complicated arithmetic involved in solving (7) can be bypassed
if we factor the L.H.S. of (7) as (5u − 4)(99u + 100). But this is more
like an after thought. If one tries to solve (7) honestly, the time taken
is enormous. Here an alert student can take advantage of the fact that
the question is designed as a multiple choice question. The slopes of

17
 2 2 3
the lines in the options (A) to (D) are, respectively √ , √ , and
5 5 4
4 4 4 9 16
. The corresponding values of u are , , and . By a direct
3 5 5 16 9
substitution, the last two are ruled out as roots of (7).
Probably the paper-setters intended to allow this sneak path. Of
course a student who wants to rely solely on sneak paths can as well
do the sneaking even further and directly verify that the condition for
tangency to the hyperbola, viz. c2 = 9m2 − 4 is satisfied only in (B).
So, without even checking if this line touches the circle, he can safely
tick (B) as the answer since exactly one of the answers is given to be
correct.

Q.15 Equation of the circle with AB as its diameter is

(A) x2 + y 2 − 12x + 24 = 0 (B) x2 + y 2 + 12x + 24 = 0

(C) x2 + y 2 + 24x − 12 = 0 (D) x2 + y 2 − 24x − 12 = 0

Answer and Comments: (A). The straightforward way is to begin

by actually identifying the points A and B by solving the equations

x2 + y 2 − 8x = 0 (8)
x2 y 2
and − = 1 (9)
9 4
simultaneously. Eliminating y and simplifying we get a quadratic in x,
viz.

13x2 − 72x − 36 = 0 (10)

which has 6 and√−6/13 as its √ roots. When we put x = 6 in (9) we get

y 2 = 12. So (6, 12) and 6, − 12) are two points of intersection. The
other value of x, viz. −6/13 would make y 2 negative
√ and hence has to
be discarded. So the points A and B are (6, ± 12). The equation of
the circle with diameter AB can now be written as
√ √
(x − 6)(x − 6) + (y − 12)(y + 12) = 0 (11)

which, upon simplification coincides with (A). Actually, an alert candi-

date can bypass a lot of work involved. In general two conics intersect

18
 in four points. But in the present problem they are given to intersect
only in two points as otherwise the question would not have a unique
meaning. So he need not bother to check what happens when the other
root of (10) is put into (9). Actually, the sneaking can go even further.
The moment we know that the x-coordinates of the points of intersec-
tion of both A and B are 6, we know that in the equation of the circle
with AB as a diameter we will have x2 − 12x and some other terms.
In the given options, (A) is the only one where this holds. So, without
any further work it is the right answer. Had the paper-setters been a
little careful, among the fake options they would have included one in
which the x terms are x− 12x.
As it often happens with the so called paragraph type questions,
the two questions are totally unrelated. Even though they deal with
the same pair of curves, the work done in any one of them does not
help in the solution of the other.

Paragraph for Q.16 to Q.18

Let p be an odd prime number and Tp be the following set of 2 × 2

matrices
( " # )
a b
Tp = A = : a, b, c ∈ {0, 1, . . . , p − 1}
c a

Q.16 The number of A in Tp such that A is either symmetric, or skew sym-

metric or both and det (A) is divisible by p is

(A) (p − 1)2 (B) 2(p − 1) (C) (p − 1)2 + 1 (D) 2p − 1

Answer and Comments: (D). Let us first assume that A is symmet-

ric. Then c = b and the determinant of A is a2 − b2 which factors as
(a + b)(a − b). Since p is a prime, if it divides a2 − b2 , it has to divide
either a + b or a − b. Since a and b lie between 0 and p − 1, the second
possibility means a = b while the first one can hold only if a + b = 0
or p. a = b has p solutions. As for a + b = 0, the only solution is
a = 0, b = 0 which is already counted. For a + b = p, a can take any
value from 1 to p − 1 and then b is uniquely determined. Also note that
a + b = p and a = b cannot hold simultaneously since p is odd. Hence

19
 in all there are p + (p − 1) i.e. 2p − 1 possibilities, each of which gives
exactly one symmetric matrix A ∈ Tp with determinant divisible by p.
Now assume A is skew symmetric. Then a = 0 and c = −b. Then
det(A) = b2 which is divisible by p only when b = 0. But in that case A
is the zero matrix which is already counted as a symmetric matrix. (A
slicker way is to use that the entries are all non-negative.) Finally, the
only matrix which is both symmetric and skew symmetric is the zero
matrix which is already counted. So the net count remains at 2p − 1.
The essential idea in the problem is that when a prime divides
a product of integers, it must divide at least one of the factors. The
material from the matrices needed is little beyond the definitions of
symmetry, skew symmetry and the determinant. The counting involved
is also elementary. So this problem is a good combination of elementary
ideas from three fields. The only disturbing part is that since only
the symmetric matrices contribute to the counting, a candidate who
eliminates the case of skew symmetric ones by reasoning cannot be
distinguished from someone who blissfully ignores them.

Q.17 The number of A in Tp such that the trace of A is not divisible by p

but det(A) is divisible by p is

(A) (p − 1)(p2 − p + 1) (B) p3 − (p − 1)2

(C) (p − 1)2 (D) (p − 1)(p2 − 2)
[ Note: The trace of a matrix is the sum of its diagonal entries.]

Answer and Comments: (C). The trace is 2a which is divisible by p

if and only if a = 0. So we assume a 6= 0. For each such fixed a we now
need to determine all ordered pairs (b, c) for which a2 −bc is divisible by
p. Note that even though 1 ≤ a ≤ p − 1, a2 may be bigger than p. Let
r be the remainder when a2 is divided by p. Then r 6= 0 (as otherwise
p would divide a) and so 1 ≤ r ≤ p − 1. The condition that a2 − bc
is divisible by p is equivalent to saying that the integer bc also leaves
the remainder r when divided by p. This rules out the possibility that
b = 0. For every b with 1 ≤ b ≤ p − 1, we claim that there is precisely
one c with 1 ≤ c ≤ p − 1 such that bc leaves the remainder r when
divided by p. To see this consider the remainders when the multiples
b, 2b, 3b, . . . , (p − 1)b are divided by p. None of these remainders is 0 as

20
otherwise p would divide b. We claim these remainders are all distinct.
For, suppose ib and jb leave the same remainder for some i, j with
1 ≤ i < j ≤ p − 1. Then p divides (j − i)b which would mean that
either p divides j − i or it divides b, both of which are impossible since
j − i and b both lie between 1 and p − 1. Since the remainders left by
b, 2b, . . . , (p − 1)b are all distinct, and there are p − 1 possible values
for the remainder, we conclude that the remainder r occurs precisely
once in this list. Put differently, for each b ∈ {1, 2, . . . , p − 1}, there
is precisely one c such that bc leaves the same remainder as a2 does,
which is equivalent to saying that a2 − bc is divisible by p. So, for each
fixed a ∈ {1, 2, . . . , p − 1} there are p − 1 matrices of the desired type.
Since a itself takes p − 1 distinct values, the total number of desired
matrices is (p − 1)2 .
For those familiar with the language of congruence modulo an
integer, (introduced in Comment No. 15 of Chapter 4) the argument
above can be paraphrased slightly. Instead of saying that r is the
remainder when a2 is divided by p we say that let a2 ≡ r (mod p). The
crux of the argument above is that given any a 6≡ 0 (mod p) and any b 6≡
0 (mod p), there is precisely one c such at a2 ≡ bc (mod p). The proof,
however, remains essentially the same. The proof is interesting because
it does not actually exhibit such c, but proves its existence by showing
that the congruence classes of the p − 1 integers b, 2b, . . . , (p − 1)b are
all distinct and since there are only p − 1 (non-zero) congruence classes
in all, exactly one of them is the congruency class of a2 . This proof,
therefore, is an application of the well known pigeon hole principle
introduced in Comment No.s 16 and 17 of Chapter 6. In terms of
multiplication modulo p, the same argument can be used to show that
if b is not divisible by p, then b has an inverse modulo p, i.e. an integer
d such that bd ≡ 1 (modulo p). In fact, if we use this result, then the
integer c above can be quickly identified as the unique integer between
1 to p − 1 whose congruence class modulo p is the same as that of a2 d,
because in the modulo p arithmetic, a2 = a2 bd = bc.
As in the last question, the knowledge of matrices needed is minimal.
The question is more on number theory. It is not easy to come up with
good short questions in number theory. The paper-setters deserve to
be commended for achieving this in the present problem.

21
Q.18 The number A in Tp such that det(A) is not divisible by p is

(A) p2 (B) p3 − 5p (C) (p − 1)2 (D) p3 − p2

Answer and Comments: (D). The very format of the question sug-
gests that complementary counting is the right tool. Since every matrix
in Tp is determined by three mutually independent parameters a, b, c
each taking p possible values, in all there are p3 matrices in the set Tp .
We are interested in counting how many of these have a determinant
not divisible by p. So let us count how many matrices in Tp have de-
terminants divisible by p. In the last question we did this count when
the trace was not divisible by p and that count was (p − 1)2 . Let us
now add to this the number of matrices with both the trace and the
determinant being divisible by p. The trace 2a is divisible precisely
when a is divisible by p, i.e. when a = 0. In this case, the determinant
is simply −bc which is divisible by p if and only if b or c (or both) is 0.
There are 2p − 1 ways this can happen. Thus the number of matrices
with determinant divisible by p is (p − 1)2 + 2p − 1 which is simply p2 .
So, the number of desired matrices is p3 − p2 .
As the answer to the second question of the paragraph is crucially
needed to answer the third question, this is a well designed paragraph
of questions unlike the last paragraph where the two questions had
nothing to do with each other.

SECTION IV
Integer Type

This section contains ten questions. The answer to each question is a

single digit integer, ranging from 0 to 9.
Q.19 let f be a real-valued differentiable function defined on IR (the set of
all real numbers) such that f (1) = 1. If the y-intercept of the tangent
at any point P (x, y) on the curve y = f (x) is equal to the cube of the
abscisa of P , then the value of f (−3) is equal to .... .

Answer and Comments: 9. We first have to formulate the given

geometric condition in terms of some equation about the function f

22
and then solve this equation to determine this function. Since the con-
dition involves tangents and hence derivatives, the resulting equation
will be a differential equation. This is therefore a problem on geometric
applications of differential equations.
Let P = (x0 , y0 ) be a typical point on the graph of the function.
(In the statement of the question, this point is taken as (x, y). But in
that case the same variables cannot be used to write down the equation
of the tangent etc. Such an equation is then written with some other
variables such as X and Y instead of x and y. We prefer to reserve x
and y for the coordinates of any point in the plane and so denote the
point P by (x0 , y0).)
The equation of the tangent at P is

y − y0 = f ′ (x0 )(x − x0 ) (1)

The y intercept of this line is y0 − x0 f ′ (x0 ) and the given condition says

y0 − x0 f ′ (x0 ) = x30 (2)

dy
Replacing x0 , y0 by x, y and f ′ by this becomes a differential equa-
dx
tion, viz.
dy y
− = −x2 (3)
dx x
R dx 1
which is a linear differential equation with integrating factor e− x = e− ln x = .
x
Hence the general solution of (3) is
Z
x3
y = f (x) = x[( −xdx)] = − + cx (4)
2
3
where c is a constant. The condition f (1) = 1 gives c = . Hence we
2
have
x3 3
f (x) = − + x (5)
2 2
27 9 18
from which f (−3) = − = = 9.
2 2 2
23
 Applications of differential equations of this sort are very common
and so the problem is straightforward. But the work involved is time
consuming and prone to errors. The only hint is that if the answer is
not an integer from 0 to 9, then something has gone wrong.
Q.20 The number of values of θ in the interval (−π/2, π/2) such that θ 6=
nπ/2, n = 0, ±1, ±2, tan θ = cot 5θ and sin 2θ = cos 4θ is .... .

Answer and Comments: 3. Here we have to solve two trigonometric

equations in θ simultaneously. The excluded values are those where
sin 5θ = 0 and hence cot 5θ is undefined.
We solve the equations separately by converting each to a trigono-
metric equation where only one trigonometric function of θ is involved.
The first equation can be rewritten as tan θ tan 5θ = 1 which gives
cos 5θ cos θ − sin 5θ sin θ = 0, i.e. as
cos 6θ = 0 (1)
π
whose general solution is 6θ = (2n + 1) or
2
π
θ = (2n + 1) (2)
12
where n is an integer. The interval (−π/2, π/2) contains six such values
of θ, viz.
π 3π 5π
θ=± ,± ,± (3)
12 12 12
The second equation, viz. sin 2θ = cos 4θ can be rewritten as
π
sin( − 4θ) = sin(2θ) (4)
2
whose general solution is
π
− 4θ = (−1)n 2θ + nπ (5)
2
where n is an integer. Depending upon whether n = 2k or 2k + 1 this
splits into two sets of solutions, viz.
π
6θ = − 2kπ (6)
2
π
and 2θ = − (2k + 1)π (7)
2
24
 where k is an integer. As we want to compare these values with those
π
in (3) we rewrite them in terms of multiples of , viz.
12
π
θ = (1 − 4k) (8)
12
π
and θ = (3 − 6(2k + 1)) (9)
12
where again k is an integer. Both of these give only odd multiples of
π
and we have to see which of these occur in (3). For (8) this happens
12
π 3π 5π
when k = 0, ±1 which give θ = , − and θ = while for (9) this
12 12 12
3π
happens only for k = 0 which gives θ = − which is already covered
12
by (8). Thus we see that the two given equations have 3 solutions in
π 3π 5π
common, viz. ,− and .
12 12 12
This problem, too, is straightforward. But the numerical work
needed is lengthy and prone to errors.
Q.21 The maximum value of the expression
1
2
sin θ + 3 sin θ cos θ + 5 cos2 θ
is .... .

Answer and Comments: 2. Denote the denominator by f (θ). If we

complete the squares the denominator can be written as
3 11
cos θ)2 +
f (θ) = (sin θ + cos2 θ (1)
2 4
which shows that the denominator is always positive. Hence the given
expression will be maximum when f (θ) is minimum. This can be found
by differentiating f (θ). A simple calculation gives
f ′ (θ) = 2 sin θ cos θ + 3 cos2 θ − 3 sin2 θ − 10 sin θ cos θ
= 3 cos 2θ − 4 sin 2θ (2)
3
so that f ′ (θ) = 0 gives tan 2θ = . This has two possible solutions viz.
4
3 4
sin θ = ± , cos 2θ = ± . Clearly one possibility gives the maximum
5 5
25
 while the other gives the minimum of f (θ). It is not necessary to find
the values of θ for which these equations hold. We are interested only
in the minimum value of f (θ) and not the points where it occurs. And
this becomes possible by expressing f (θ) as a function of 2θ. Indeed,
we have
3
f (θ) = 1 + sin 2θ + 2 + 2 cos 2θ (3)
2
3 4
When sin 2θ = , cos 2θ = , we get the maximum while sin θ =
5 5
3 4
− , cos 2θ = − gives the mimimum of f (θ). The latter equals 3 −
5 5
9 8 1
− = . As observed before, the maximum values of the given
10 5 2
expression is the reciprocal of this, viz. 2.
We could have as well found the minimum of f (θ) directly from (3),
without differentiation. For this we use the result that for any positive
real numbers the maximum and the minimum of a cos 2θ + b sin 2θ are
1
±√ 2 2
. The easiest way to show this is to write the expression
√a + b √
as a2 + b2 (sin α cos 2θ + cos α sin 2θ) = a2 + b2 sin(α + 2θ) where
a
α = sin−1 ( √ 2 ). So, in the present case, the minimum value of
a + b2
q 5 1
f (θ) is 3 − 94 + 4 = 3 − = , the same value as before.
2 2
There are two key ideas in the solution. First, the maximum of the
reciprocal is the reciprocal of the minimum. Secondly, for trigonometric
functions involving only sines and cosines, the extrema can often be
found without differentiation using the fact that the maximum and the
minimum of the sine and the cosine functions are 1 and −1. For the
first assertion to be valid, the expression must be shown to be positive
throughout. Once again, a sincere candidate who spends time proving
this is at a disadvantage as compared to a candidate who simply ignores
this point.
2π 2π
Q.22 Let ω be the complex number cos + i sin . Then the number of
3 3

26
distinct complex numbers z satisfying
z+1 ω ω2
ω z + ω2 1 =0
2
ω 1 z+ω
is

Answer and Comments: 1. We already encountered one problem

(Q.3) on ω, the complex cube root of unity. The solution to it was
based on the equation ω 2 + ω + 1 = 0. In fact this is the case with most
problems involving ω at the JEE level and the present problem is no
exception. The value of the given determinant is a function of z and so
let us denote it by D(z). It is obvious that D(z) is a cubic polynomial
in z and so in general it will have three roots. But they may not be
all distinct. To see if this is so, we need to evaluate D(z). This can
be done by direct expansion. But the relation ω 2 + ω + 1 = 0 suggests
that a row reduction is a better idea. Indeed if we add the second and
the third row to the first, then the first row has a common factor, viz.
z + 1 + ω + ω 2 which is simply z. So we get
1 1 1
2
D(z) = z ω z + ω 1 (1)
ω2 1 z+ω
which is manageable enough for a direct expansion. Those who shun
direct expansions can subtract the first column from each of the re-
maining ones and expand w.r.t. the first row. Whichever method is
applied, we get (keeping in mind that ω 3 = 1)
D(z) = z[z 2 + ωz + ω 2 z − ωz − ω 2 z] (2)
But this is simply z 3 . So, the given equation has a triple root 0. There-
fore there is only one complex number satisfying it.
This is a good problem but somewhat marred by the duplication
of the essential idea which was also used in Q.3.
The given determinant has a special significance. We see that D(−z)
1 ω ω2
is simply the characteristic polynomial of the matrix ω ω 2 1 =
ω2 1 ω

27
 A (say). The roots of this polynomial are called its characteristic roots
or eigenvalues and are very important in applications. The present
problem says that 0 is an eigenvalue of multiplicity 3 of the matrix A.
That 0 is an eigenvalue of A can also be seen from the fact that the
matrix A is singular (which follows by showing that its determinant
is 0, since the sum of the rows is identically 0. Those familiar with
the concept of the rank of a matrix and the nullity of a matrix will
recognise that the rank of A is 2 because its second and third columns
are multiples of the first. Therefore its nullity is 2, which means that
the multiplicity of 0 as an eigenvalue is 2. However, these observations
do not prove that it is 3 as is needed in the present problem.
Some structural features of the matrix A are noteworthy. Of
course it is a symmetric matrix. But it is much more than that. Each
column in it except the first one is obtained from its previous column
by cyclically shifting its entries one row upwards. For this reason, a
matrix of this form is called a circulant matrix.

Q.23 If the distance between the planes Ax − 2y + z = d and the plane

containing the lines
x−1 y−2 z−3 x−2 y−3 z−4
= = and = =
2 3 4 3 4 5
√
is 6, then |d| is .... .

Answer and Comments: 6. There is an implied hint in the prob-

lem that the two planes are parallel to each other. Hence when their
equations are written the coefficients of the linear terms will be propor-
tional. This will enable us to identify the unknown A in the equation of
the first plane, provided we first identify the second plane. We already
know that (1, 2, 3) is a point on the second plane. Also a normal, say
n to it is given by the cross product of the vectors parallel to the given
lines. Thus
i j k
n = 2 3 4 = −i + 2j − k (1)
3 4 5

Hence the equation of the second plane is −(x−1)+2(y−2)−(z−3) = 0

28
 i.e.

−x + 2y − z = 0 (2)

This plane will be parallel to the first plane if and only if A = 1. Once
we know this, the perpendicular distance between the two planes is
|d − 0| |d| √
√ = √ . For this to equal 6, |d| must equal 6.
1+4+1 6
A very straightforward problem. The idea of specifying a plane by
specifying two lines in it was also used in Q.2. This duplication could
have been avoided, for example, by specifying three points in the plane.
x2 y 2
Q.24 The line 2x+ y = 1 is tangent to the hyperbola 2 − 2 = 1. If this line
a b
passes through the point of intersection of the nearest directrix and the
x-axis, then the eccentricity of the hyperbola is ... .

Answer and Comments: 2. Another instance of duplication of ideas,

this time with Q.14). Writing the given line as y = −2x + 1, the
condition for tangency to the hyperbola gives

4a2 − b2 = 1 (1)

If e denotes the eccentricity of the hyperbola then b2 = a2 (e2 − 1) and

putting this into (1) we get

5a2 − e2 a2 = 1 (2)

To determine e we need one more equation in a and e. This is provided

by the data that the line passes through the point of intersection of the
nearest directrix with the x-axis. There are only two directrices of the
a
parabola, viz. the lines x = ± . The statement of the question says
e
that we have to take the nearer of the two. But is is not clear nearer to
what. Presumably, it is the point of contact of the line 2x + y = 1 with
the hyperbola. To find it we substitute y = −2x + 1 into the equation
of the hyperbola and get

x2 (−2x + 1)2
− =1 (3)
a2 b2

29
 which, in view of (1) becomes

(4a2 − 1)x2 − a2 (−2x + 1)2 − a2 (4a2 − 1) = 0 (4)

Further simplification gives

−x2 + 4a2 x − 4a4 = 0 (5)

which has x = 2a2 as a double root. This is hardly surprising because

if there were no double roots then the line will not be a tangent. The
point to note is that the x-coordinate of the point of contact of the given
line is positive. Therefore, even without finding its y-coordinate, we see
a a
that the directrix x = is nearer to it than the directrix x = − .
e e
Thus the data now means that the line 2x + y = 1 passes through
a
the point ( , 0). This gives
e
2a
=1 (6)
e
or a = e/2. Putting this into (2) gives a quartic

e4 − 5e2 + 1 = 0 (7)

which gives e2 = 1 or e2 = 4. The first possibility is ruled out because

the eccentricity of a hyperbola is always bigger than 1. The second
possibility gives e = 2.
Yet another problem which, although straightforward, demands
a lot of computation. Especially disturbing is the fact that the data
is ambiguous. As shown above, if the nearest directrix is with refer-
ence to the point of contact, then some work is needed to make the
choice. A candidate who simply assumes the correct choice stands to
gain undeservedly in terms of the precious time saved.

Q.25 For any real number x, let [x] denote the largest integer less than or
equal to x. Let f be a real valued function defined on the interval
[−10, 10] by
(
x − [x] if [x] is odd
f (x) =
1 + [x] − x if [x] is even

30
 π 2 Z 10
Then the value of f (x) cos πx dx is .... .
10 −10

Answer and Comments: 4. The function [x] is popularly called

the greatest integer function or the integral part function and it is an
unwritten rule that every JEE paper must contain at least one question
based on it ! The paper-setters have obeyed the rule. The reason for
the popularity of this function is that it is a standard example of a
step function. So this function and some other functions derived from
it such as the fractional part function {x} defined as x − [x] figure
in many problems about continuity and differentiability. In the present
problem, however, we need that this function is periodic with period
1. Further, as the definitions are slightly different depending upon
whether [x] is even or odd, we see that the given function f (x) has
a period 2 and not 1. Note further that the function cos πx is also
periodic with period 2. As the two periods match, we see that the
integrand f (x) cos πx is a periodic function of x with period 2π.
We now use a crucial property of the integral of a periodic function,
viz. that the integral over any interval whose length equals the period
is the same, i.e. independent of the starting point of the interval. So,
in the present case, we chop the interval [−10, 10] into 10 subintervals
of length 2 each and get
π 2 10 2
Z Z
f (x) cos πx dx = π 2 f (x) cos πx dx (1)
10 −10 0

Since the definition of f (x) differs on [0, 1] and on [1, 2], we split the
integral on the R.H.S. to get
Z 2 Z 1 Z 2
f (x) cos πx dx = f (x) cos πx dx + f (x) cos πx dx
0 0 1
Z 1 Z 2
= (1 − x) cos πx dx + (x − 1) cos πx dx
0 1
Z 1 Z 1
= (1 − x) cos πx dx − x cos πx dx
0 0
Z 1
= (1 − 2x) cos πx dx (2)
0
The last integral can be evaluated by parts as
1 2 1
Z 1
1
Z
(1 − 2x) cos πx dx = = (1 − 2x) sin πx + sin πx dx
0 π 0 π 0

31
 2 1
= 0− 2
cos πx
π 0
4
= (3)
π2
Substituting this into (1), we get the given integral as 4.
The problem is a good combination of the properties of the integrals
of periodic functions and those of the integral part function. But once
again, the computations can hardly be completed within the allocated
time.
i − 2j 2i + j + 3k
Q.26 If a and b are vectors in space given by a = √ and b = √ ,
5 14
then the value of (2a + b) · [(a × b) × (a − 2b)] is .... .

Answer and Comments: 5. Note that both a and b are unit vec-
tors, although this has little bearing on the problem. The question asks
for the value of a scalar triple product of three vectors each of which
is expressed in terms of the vectors a and b. As both these vectors
are given explicitly, the most straightforward approach is to express all
the three vectors in terms of their components and evaluate the deter-
minant of the coefficient matrix. But as it often happens, sometimes
certain vector identities lead to a more elegant solution. In the present
case we use the following identity about the box product.
u · (v × w) = v · (w × u) (1)
(which is a reflection of a certain property of determinants). Applying
this to the given dot product we see that
(2a + b) · [(a × b) × (a − 2b)] = (a × b) · [(a − 2b) × (2a + b)] (2)
The advantage gained is that the cross product inside the brackets on
the R.H.S. can be expanded to get
(a − 2b) × (2a + b) = a × b − 4(b × a) = 5a × b (3)
If we put this into the R.H.S. of (2) it simply becomes 5|a × b|2 which
is best evaluated by directly calculating a × b as
i j k
1 1
a×b= √ 1 −2 0 = √ (−6i − 3j + 5k) (4)
70 2 1 3 70

32
 1
which gives |a × b| = (36 + 9 + 25) = 1. Hence the R.H.S. of (2) is
70
simply 5.
Instead of resorting to (2), we could have simplified the given
expression using an identity about the vector triple product, viz.
u × (v × w) = (u · w)v − (u · v)w (5)
Using this property and the anti-commutativity of the cross product,
we get
(a × b) × (a − 2b) = −(boa − 2b) × (a × b)
= (2b − a) × (a × b)
= [(2b − a) · b]a − [(2b − a) · a]b
= (2b · b − a · b)a − (2b · a − a · a)b (6)
The vectors a, b are given explicitly in terms of the components. So we
can easily calculate the various dot products and express the R.H.S. of
(6) as a linear combination of a and b. Substituting this into the desired
scalar triple product and again putting in the values of the various dot
product, we get the numerical value of the scalar triple product. The
work involved here is more than in the earlier approach. But that is
largely because in the earlier approach, the bracketed expression in the
R.H.S. of (2) luckily came out to be a scalar multiple of the vector
a × b. If instead of a × b we had some other vector the computation
would not have been so easy. There is therefore really no way to tell
beforehand which approach is better in a particular numerical problem.
Q.27 The number of all possible values of θ where 0 < θ < π, for which the
system of equations
(y + z) cos 3θ = xyz sin 3θ (1)
2 cos 3θ 2 sin 3θ
x sin 3θ = + (2)
y z
(xyz) sin 3θ = (y + 2z) cos 3θ + y sin 3θ (3)
has a solution (x0 , y0, z0 ) with y0 z0 6= 0, is .... .

Answer and Comments: 3. This is a combination of systems of

equations and trigonometric equations. Here if we take the unknowns

33
as x, y, z (a natural choice) the given system is not a system of linear
equations. But if we multiply (2) throughout it becomes

xyz sin 3θ = 2y sin 3θ + 2z cos 3θ (4)

So, if we take the unknowns as xyz, y and z, then (1), (3) and (4)
forms a homogeneous linear system which in the matrix notation can
be written as
    
sin 3θ − cos 3θ − cos 3θ xyz 0
 sin 3θ

−2 sin 3θ −2 cos 3θ   y  =  0 
  
 (5)
sin 3θ − sin 3θ − cos 3θ −2 cos 3θ z 0
This system will have a non-trivial solution if and only if the determi-
nant, say D, of the coefficient matrix vanishes. If we take out common
factors sin 3θ, −1 and − cos 3θ from the first, the second and the last
column respectively, we get
1 cos 3θ 1
D = sin 3θ cos 3θ 1 2 sin 3θ 2 (6)
1 sin 3θ + cos 3θ 2
Subtracting the first row from the other two we get
1 cos 3θ 1
D = sin 3θ cos 3θ 0 2 sin 3θ − cos 3θ 1
0 sin 3θ 1
= sin 3θ cos 3θ(sin 3θ − cos 3θ) (7)

Thus D will vanish when any one of the factors vanishes. This gives
us three trigonometric equations in θ and the union of their solution
sets is the set of values of θ for which the system (5) has a non-trivial
solution. But our requirement is more specific. A non-trivial solution
simply means a triple (x0 , y0 , z0 ) for which at least one of x0 y0 z0 , y0 and
z0 is non-zero. But we want only those solutions in which both y0 , z0
are non-zero. So the method above is not applicable.
We therefore tackle the problem directly. We see that the term
xyz sin 3θ appears in all the three equations (1), (3) and (4). If we
subtract (4) from (3), both x and z get eliminated and we get

y(cos 3θ − sin 3θ) = 0 (8)

34
 For this to have a non-zero solution, we must have sin 3θ = cos 3θ, or
equivalently, tan 3θ = 0. In the interval (0, π) this happens only for
π 5π 9π
θ= , and . (1) and (3) together imply that
12 12 12
z cos 3θ + y sin 3θ = 0 (9)

which, in presence of sin 3θ = cos 3θ, becomes (y + z) cos 3θ = 0 and

hence z = −y since cos 3θ cannot vanish when it also equals sin 3θ.
Thus we have z = −y. Therefore whenever y is non-zero, so is z.
We are not yet done. We have shown that when tan 3θ = 1, (8)
and (9) have a solution in which both y, z are non-zero and y + z = 0.
We must now check whether the original system (1), (2), (3) has a
solution for these values of θ. The very first equation gives x = 0.
This is also consistent with (2) and (3). Thus in all there are three
values of θ (already listed above) for which the original system has a
solution in which y, z are both non-zero. Indeed we know that the
complete solution set is the set of all triples of the form (0, y0 , −y0 )
where y0 6= 0.
This is a very good problem where a wrong technique (viz. the
criterion for a homogeneous system of linear equations to have a non-
trivial solution) is very tempting. But once again, a candidate who
arrives at the correct answer after going through the analysis above
cannot be distinguished from one who ignores the factors sin 3θ and
cos 3θ of D and simply solves sin 3θ = cos 3θ.

Q.28 Let Sk , k = 1, 2, . . . , 100 denote the sum of the infinite geometric series
k−1 1
whose first term is and the common ratio is . Then the value
k! k
of
100
1002 X
+ (k 2 − 3k + 1)Sk is .... .
100! k=1

Answer and Comments: 3. This is a problem about getting a closed

form expression for a finite sum whose k-th term is specified as a func-
tion of k, viz. |(k 2 − 3k + 1)Sk | where Sk is itself specified indirectly,
viz. the sum of an infinite geometric series. The formula for this sum
a
is standard, viz. where a is the first term and r is the common
1−r

35
ratio. But this formula is valid only when |r| < 1. In the present case,
the common ratio being 1/k, the sum S1 has to be calculated directly
without this formula. When k = 1, the first and hence all the terms of
the G.P. are zero and so we have S1 = 0. (It is a little misleading to
use the term ‘common ratio’ for such a geometric progression, because
when we take the ratio of two numbers, there is a tacit assumption that
the denominator is non-zero. But if we interpret the term ‘geometric
progression’ to mean simply a sequence whose terms are of the form
a, ar, ar 2 , . . . , ar n , . . ., where a and r are some numbers, then S1 can
pass as a geometric progression. Note however, that it does not have a
uniquely defined common ratio. Such degeneracies occur elsewhere too.
For example, the argument of the complex number 0 is not uniquely
defined. A zero vector has no unique direction. And so on.)
100
(k 2 − 3k + 1)Sk .
X
Since S1 = 0, we might as well replace the given sum by
k=2
For k > 1, the formula is applicable and we have
(k − 1)/k! k(k − 1) 1
Sk = = = (1)
1 − (1/k) k!(k − 1) (k − 1)!
for all k > 1. Let us now analyse the k-th term of the given summation,
viz. |(k 2 − 3k + 1)Sk |. From (1) we have
k 2 − 3k + 1
|(k 2 − 3k + 1)Sk | = (2)
(k − 1)!
n n
k2 .
X X
There are well-known formulas for sums of the form k and
k=1 k=1
But because of the term (k − 1)! in the denominator, they are of little
use here. The only way to sum a series whose general term is of the
type in (2) is to rewrite it as a telescopic series, i.e. to express the k-th
term as the difference of two expressions, say as ak − bk in such a way
that bk cancels with ak+1 . In the present case, we write the numerator
k 2 − 3k + 1 as (k − 1)2 − k. Then we have
k 2 − 3k + 1 (k − 1)2 − k
=
(k − 1)! (k − 1)!
k−1 k
= − (3)
(k − 2)! (k − 1)!

36
the first fraction is greater than the second whenever (k −1)2 > k which
holds for all k ≥ 3. So,
k−1 k
k 2 − 3k + 1 − for k ≥ 3
(
(k−2)! (k−1)!
= k k−1 (4)
(k − 1)! (k−1)!
− (k−2)!
for k = 2

As a result, the first term of the (reduced) summation is exceptional.

Isolating it, we write
100 k=2 100
X k 2 − 3k + 1 X k 2 − 3k + 1 X k 2 − 3k + 1
= +
k=2 (k − 1)! k=2 (k − 1)! k=3 (k − 1)!
100
!
1 X k−1 k
= + − (5)
1! k=3 (k − 2)! (k − 1)!

3−1 100
The last series is a telescopic series which adds up to −
(3 − 2)! (99)!
1002
i.e. to 2 − . So, the reduced sum and hence also the given sum
(100)!
1
equals + 2 = 3.
1!
The essential idea in the problem is that of a telescopic series.
But to make the answer come out as an integer from 0 to 9, the term
1002
is added. That makes the problem clumsy. The exceptionality
100!
of the first two terms of the summation adds to its clumsiness. And
as if all this was not enough, instead of giving Sk directly, it is given
as the sum of an infinite geometric series (where again, S1 has to be
obtained differently). Such a thing would have been appropriate in the
past for a full length question carrying sufficient weightage allowing the
candidates to spend about 6 to 7 minutes for the solution. It is true
that in the present format, the candidate does not have to show all his
work. But he still has to do it as rough work and that takes time. It
is high time that the paper-setters themselves do a realistic assessment
of the time it takes to put all the pieces together. What is worse still is
that a candidate who does the harder part of the problem (viz. writing
the series as a telescopic series) but makes a mistake in some relatively
minor part (such as realising that S1 = 0) gets the wrong answer and
hence the time he has spent is absolutely wasted.

37
 The term S1 deserves some comment. Note that the first two ratios
in (1) make no sense for k = 1, because they are of the indeterminate
0 1
form . But the last term, viz. , is perfectly well defined even
0 (k − 1)!
for k = 1 and has value 1. So, it is very tempting to take S1 as 1 rather
than as 0, in which case the answer to the question would be 4 and
not 3. Since the sum of an infinite series is defined as the limit of its
partial sums, for the series S1 , all the partial sums are 0 and so the
correct value of S1 is 0. But there is some plausible justification for
taking S1 as 1. First of all, S1 is the sum of an infinite number of terms,
each of which is 0. Hence it is of the form 0 × ∞, which is another
0
indeterminate form (closely related to the form). The values of such
0
indeterminate forms can be non-zero if they are expressed as limits of
some other type. For example, the expressions x2 cot4 x, x4 cot4 x and
x6 cot4 x are all of the form 0 × ∞ at x = 0. But the three functions
approach the limits ∞, 1 and 0 respectively as x → 0.
In (1), the variable k is a discrete variable. Can we make it into
a continuous variable and then regrad S1 as lim+ Sk ? The immediate
k→1
difficulty in doing so is that the factorials are defined only for integral
values of k. But this can be rectified by considering a new function,
called the gamma function defined by
Z ∞
Γ(a) = xa−1 e−x dx (6)
0
It is easy to show that this integral is finite for all a > 0 and further,
integration by parts gives the functional equation
Γ(a + 1) = aΓ(a) (7)
for all a > 0. A direct calculation gives Γ(1) = 1 and then repeated
applications of (7) give Γ(k + 1) = k! for every positive integer k. So,
in (1), replacing the factorials by the gamma function and using (6),
we can write
(k − 1)/Γ(k + 1) k(k − 1) 1
Sk = = = (8)
1 − (1/k) Γ(k + 1)(k − 1) Γ(k)
1
which is valid for all real k > 1. So, as k → 1+ , Sk tends to which
Γ(1)
equals 1 because the gamma function can be shown to be continuous
at all a > 0. Therefore it is not entirely unreasonable to take S1 as 1.

38
 Similar ambiguities arise with other indeterminate forms. For
example, what meaning do we give to the power 00 ? On one hand, for
every non-zero x, the power x0 equals 1 and so by taking the limit as
x → 0, we should set 00 = 1. But we might as well keep the base fixed
and let the exponent vary. In that case for every y 6= 0, 0y equals 0 and
so 00 should be given the value 0. The first interpretation is generally
taken to prevail.
The term S1 has nothing to do with the main theme of the problem,
viz. the telescopic series. By including it, the paper-setters probably
wanted to test if a candidate is careful enough to realise that it has to
be treated differently from Sk for k > 1. But then they ought to have
exercised equal care by avoiding the phrase ‘common ratio’ which is not
applicable to S1 . Their failure to do so gives some credence to treat S1
on par with the remaining Sk ’s and makes the problem controversial.
It would have been better to drop this term which anyway has nothing
to do with the main theme of the problem, viz. the telescopic series.

39
 PAPER 2

Contents

Section I (Single Correct Choice Type) 40

Section II (Integer Type) 48

Section III (Paragraph Type) 58

Section IV (Matrix Type) 66

SECTION I
Single Correct Choice Type

This section contains 6 multiple choice questions. Each question has 4

choices out of which ONLY ONE is correct.

Q.1 Let S = {1, 2, 3, 4}. The total number of unordered pairs of disjoint
subsets of S is equal to

(A) 25 (B) 34 (C) 42 (D) 41

Answer and Comments: (D). Let us first consider all ordered pairs
(A, B)) of mutually disjoint subsets of S. The number of elements in
the subset A can beany integer, say r from 0 to 4. For each such r,
4
A can be chosen in r ways. Once the subset A is fixed is fixed, the
subset B can be any subset of S − A, the complement of A in S. Since
S − A has 4 − r elements, the number of its subsets (including the
empty set) is 24−r . Hence the total number, say T , of ordered pairs of
mutually disjoint subsets of S is
4
!
X 4 4−r
T = 2 (1)
r=0 r

As there are only five terms in the sum, it is possible to evaluate it

40
simply by computing each term and adding. The count comes out as
T = 24 + 4 × 23 + 6 × 22 + 4 × 21 + 20 = 16 + 32 + 24 + 8 + 1 = 81 (2)
An alert reader will hardly fail to notice that the coefficients are pre-
cisely those that would occur in a binomial expansion and so there has
to be an easier way to evaluate this sum. This guess is correct. If we
multiply the r-th term in (1) by the power 1r which equals 1 for every
r, we see that
4
!
4 4−r r
2 1 = (2 + 1)4 = 34 = 81
X
T = (3)
r=0 r
In fact, this shows that if the set S had n elements to begin with then
the number, say Tn , of all ordered pairs (A, B) of mutually disjoint
subsets of S would be 3n .
Now, coming to the problem, it asks for unordered pairs {A, B}
instead of ordered pairs (A, B) of mutually disjoint subsets. In other
words we now have to identify the pair (A, B) with (B, A). These
two pairs are distinct since two mutually disjoint subsets can never
be the same, except when both of them are the empty sets. Hence
out of the 81 ordered pairs, 80 pairs pair off into 40 pairs, each giving
rise to an unordered pair, while the pair (∅, ∅) remains the same after
interchanging its entries. Thus in all there are 41 unordered pairs.
3n − 1
More generally, if S had n elements the answer would be + 1.
2
There is a far more elegant way to count the number Tn of all mutually
disjoint pairs of subsets of a set S with n elements. (In our problem,
n = 4.) The idea is to get a one-to-one correspondence between the
set of all ordered pairs (A, B) of mutually disjoint subsets of S and the
set of all functions from the set S to a set with three elements. Given
any such ordered pair, for every x ∈ S, exactly one possibility holds,
either x is in A, or it is in B or it is in neither. We let f (x) = a, b, c
depending upon which possibility holds. This gives us a function from
S to the three element set {a, b, c}. Conversely every such function
determines an ordered pair (A, B) of mutually disjoint subsets of S, if
we let A = f −1 (a) and B = f −1 (b). There are in all 3n functions from
S to {a, b, c}. Hence we have Tn = 3n . If we want unordered pairs, the
3n − 1
answer is + 1.
2
41
 The problem is a good one, based on elementary counting. The
cardinality of the set S is kept small enough to enable those who can’t
think of either the solution using the binomial theorem or the elegant
solution at the end. Problems based on this idea have been asked before
in the JEE. (See the 1990 problem in Comment No. 17 of Chapter 5,
and also its alternate solution in Chapter 24.)
Q.2 For r = 0, 1, . . . , 10, let Ar , B2 , Cr denote respectively, the coefficient of
xr in the expansions of (1 + x)10 , (1 + x)20 and (1 + x)30 . Then
10
X
Ar (B10 Br − C10 Ar )
r=1

equals
2
(A) B10 − C10 (B) A10 (B10 − C10 A10 ) (C) 0 (D) C10 − B10

Answer and Comments: (D). This is a binomial sum. The terms

A10 , B10 and C10 are independent of r. Hence the given sum, say S,
equals
10
X 10
X
S = B10 Ar Br − C10 Ar Ar (1)
r=1 r=1

There is in general no formula for the sums of products of the binomial

coefficients. But in the present case because of the symmetry relations
for the binomial coefficients, we have
Br = B20−r (2)
and Ar = A10−r (3)
for every r = 1, 2, . . . , 10. So, we can rewrite S as
10
X 10
X
S = B10 Ar B20−r − C10 Ar A10−r (4)
r=1 r=1

The reason for these replacements is that there is a way to sum the
products of binomial coefficients when their suffixes add to a fixed num-
ber. Specifically, we have the following identity for any non-negative
integers p, q, m.
m
! ! !
X p q p+q
= (5)
r=0 r m−r m

42
 which is proved by writing (1 + x)p (1 + x)q as (1 + x)p+q and collecting
the coefficient of xm from both the sides. (If the value of some r exceeds
the exponent p or q, then the corresponding binomial coefficient is to
be taken as 0. That way we do not have to worry about how big m is
as compared to p and q. Indeed, it could even be bigger than both p
and q.)
We apply (5) with m = 10 to each of the sums in (4). Note that in
both these sums the index variable only varies from 1 to 10 and not
from 0 to 10. So allowing for the missing terms, we have

S = B10 (C10 − A0 B20 − C10 (B10 − A0 A10

= B10 (C10 − 1) − C10 (B10 − 1)
= C10 − B10 (6)

This is a problem based on two identities about the binomial coeffi-

cients. Both are standard and problems based on them have appeared
many times (see Chapter 5 for examples). In the multiple choice for-
mat, it is very difficult to ask any radically new problem.

Q.3 Let f be a real valued function defined on the interval (−1, 1) such that

−x
Z x √
e f (x) = 2 + t4 + 1 dt
0

for all x ∈ (−1, 1) and let f −1 be the inverse function of f . Then

(f −1 )′ (2) is equal to

(A) 1 (B) 1/3 (C) 1/2 (D) 1/e

Answer and Comments: (B). Note that the question does not ask
us to show that the inverse function exists. To see this, rewrite the
given equation as
x x
Z x √
f (x) = 2e + e t4 + 1 dt (1)
0
x
The
√ function e is strictly increasing everywhere. As the integrand
4
t + 1 is positive everywhere, the function defined by its integral is
also strictly increasing as x increases. Hence f (x) is strictly increasing
on (−1, 1) (in fact on the entire real line) and therefore the inverse

43
 function definitely exists (as a consequence of the Intermediate Value
property of continuous functions), its domain being the range of the
function f (x). For notational simplicity, denote the inverse function
f −1 by g. We are asked to find g ′ (2).
The theorem about the derivative of an inverse function says that
if f (x0 ) = y0 , then
1
g ′ (y0 ) = (2)
f ′ (x0 )
In the present problem y0 is given as 2. So first we have to find x0 such
that f (x0 ) = y0 . Had we known a formula for g we would simply let
x0 = g(x0 ). But in the present problem, it is not easy to express g(x)
by an explicit formula. Nevertheless, by inspection we see from (1) that
f (0) = 2. (Probably the reason to specify the data in a different form
than (1), which is the most natural way to express a function, was to
make it a little more difficult to see that f −1 (2) = 0.)
Our problem is now reduced to finding f ′ (0). For this we differentiate
(1) using the second form of the fundamental theorem of calculus to
get
Z x √ √
f ′ (x) = 2ex + ex t4 + 1 dt + ex x4 + 1 (3)
0

which directly gives f ′ (0) = 2 + 1 = 3. Hence (f −1 )′ (2) = 1/3.

The crux of the problem is that even though the inverse function f −1
cannot be described by an explicit formula, its derivative at a particular
point y0 can be obtained if we are able to find (or rather guess) the
inverse image of this particular point y0 . The problem per se is very
simple.

Q.4 If the distance of the point P (1, −2, 1) from the plane x + 2y − 2z = α
where α > 0, is 5, then the foot of the perpendicular to the plane is

(A) (8/3, 4/3, −7/3) (B) (4/3, 4−/3, 1/3)

(C) (1/3, 2/3, 10/3) (D) (2/3, −1/3, 5/2)

Answer and Comments: (A). A straightforward problem in solid

coordinate geometry. The formula for the perpendicular distance of a

44
 point from a plane gives
|1 − 4 − 2 − α| |5 + α|
5= √ = (1)
1+4+4 3
which implies that α + 5 = ±15 and hence α = 10 or α = −20. As α
is given to be positive, we have α = 10. So the equation of the plane
now is

x + 2y − 2z = 10 (2)

Let Q = (x0 , y0, z0 ) be the foot of the perpendicular from P to this

plane. We express the coordinates of Q in terms of a parameter. For
this we note that the line joining P to Q is perpendicular to the plane
and hence parallel to the normal to the plane. So we must have
x0 − 1 y0 + 2 z0 − 1
= = (3)
1 2 −2
Let each ratio equal t. Then we have Q = (t + 1, 2t − 2, −2t + 1). As
this point lies on the plane (2), we have t + 1 + 4t − 4 + 4t − 2 = 10
which yields t = 5/3. Hence from (3), x0 = 5/3 + 1 = 8/3. We can sim-
ilarly determine y0 and z0 . But an alert student will not waste his time
in doing so, since among the given options (A) is the only one where
the first coordinate is 8/3. At least for such an extremely straightfor-
ward problem like this the paper-setters should have designed the fake
answers so as to preclude such short cuts.
Q.5 A signal which can be green or red with probability 4/5 or 1/5 re-
spectively, is received by station A and then transmitted to station B.
The probability of each station receiving the signal correctly is 3/4. If
the signal received at station B is green, then the probability that the
original signal was green is

(A) 3/5 (B) 6/7 (C) 20/23 (D) 9/20

Answer and Comments: An interesting problem on probability.

Also the setting is different from the usual tossing of coins or dice
or drawing balls from an urn or cards from a pack. The estimation
of accuracy in signal transmission is an important problem in the the-
ory of communication. So apart from its mathematical content, the

45
paper-setters deserve some appreciation for giving a very inviting (and
colourful!) garb to the problem.
Now, coming to the problem itself, we are given that the signal re-
ceived by B is green. This can happen in any of the following mutually
exclusive ways:
E1 : the original signal was green and both A and B received their
respective signals correctly
E2 : the original signal was green and both the stations received their
respective signals wrongly
E3 : the original signal was red and A received it correctly but B re-
ceived the retransmitted signal wrongly, and
E4 : the original signal was red, A received it wrongly but B received
the transmitted signal correctly.
Denote the disjunction of these four events by E and the disjunction
of the first two events by F . The problem amounts to finding the
conditional probability of F given E. The easiest way to do this is to
calculate the probability P (Ei ) for i = 1, 2, 3, 4 and express P (E) and
P (F ) in terms of these.
Let us begin by P (E1 ). This is a conjunction of three mutually
independent events, viz. the original signal is green, the reception
at station A is correct and thirdly, that at B is also correct. The
probabilities of these three are given to be 54 , 43 and 34 respectively. So,
we have
4 3 3 36
P (E1 ) = × × = (1)
5 4 4 80
By a similar reasoning, we have
4 1 1 4
P (E2 ) = × × = (2)
5 4 4 80
1 3 1 3
P (E3 ) = × × = (3)
5 4 4 80
1 1 3 3
P (E4 ) = × × = (4)
5 4 4 80
Adding all four,
46
P (E) = P (E1 ) + P (E2 ) + P (E3 ) + P (E4 ) = (5)
80
46
 while, adding only the first two,
40
P (F ) = P (E1 ) + P (E2 ) = (6)
80
The desired probability is the ratio P (F )/P (E) which comes out as
40/46 i.e. 20/23.
It is easy to generalise this to the case where instead of two stations
we have a series of stations, say, A1 , A2 , . . . , An , each (except the last)
transmitting the received signal to the next one. Assume that the prob-
ability of correct reception at each station Ai is p. (In a more realistic
situation, these probabilities could change from station to station. But
we do only the simple case.) Let α be the probability that the original
signal is green. Now suppose that the signal received by the last station
is green. Then there are two possibilities: (i) the original signal was
green and the reception at an even number of the stations was wrong
or (ii) the original signal was red and the reception at an odd number
of stations was wrong. Call these events as A and B respectively. Then
we have
n
!
X n n−r r
P (A) = α p q
r even r
n
!
X n n−r r
and P (B) = (1 − α) p q (7)
r
r odd

where, as usual, q = 1 − p is the probability of a wrong reception. The

P (A)
desired probability now is . In some cases (e.g. when
P (A) + P (B)
1
p = q = ) it is possible to express these sums in a closed form.
2
overall, this is a good problem on conditional probability.
−→
Q.6 Two adjacent sides of a parallelogram ABCD are given by AB= 2~i +
−→
10~j + 11~k and AD= −~i + 2~j + 2~k. The side AD is rotated by an acute
angle α in the plane of the parallelogram so that AD becomes AD ′ . If
AD ′ makes a right angle with the side AB, then cosine of the angle α
is given by
√ √
(A) 8/9 (B) 17/9 (C) 1/9 (D) 4 5/9

47
 Answer and Comments: (B). The problem is apparently designed
to test a candidate’s ability to weed out the irrelevant portions of the
data and focus only on the essential part. It is not clear what role is
−→
played by the parallelogram. In essence, we are given three vectors AB
−→ −→
, AD and AD ′ all in the same plane along with the information that
−→ −→
AD ′ is obtained from AD by a rotation through an acute angle α and
−→
further that it is perpendicular to the vector AB and we are asked to
find cos α. Under these circumstances it is very obvious that the angle
−→ −→ π
between AB and AD is − α and therefore
2
cos α = sin θ (1)
−→ −→
where θ is the angle between the vectors AB and AD. We are given
both these vectors explicitly. So, we have
−→ −→
AB · AD
cos θ = −→ −→
| AB | | AD |
−2 + 20 + 22
= √ √
4 + 100 + 121 1 + 4 + 4
40 8
= = (2)
15 × 3 9
√ q
It is now immediate that cos α = sin θ = 1 − cos2 θ = 1 − (8/9)2 =
√
17/9.

The essence of the problem is merely that if three lines in the same
plane are concurrent and two of them are mutually perpendicular, then the
acute angles made by the third line with these two lines are complementary.

SECTION II
Integer Type

This section contains five questions. The answer to each question is a

single digit integer, ranging from 0 to 9.

48
Q.7 Consider a triangle ABC and let a, b, c denote the lengths of the sides
opposite to the vertices A, B, C √ respectively. Suppose a = 6, b = 10
and the area of the triangle is 15 3. If 6 ACB is obtuse and r denotes
the radius of the incircle of the triangle, then r 2 equals .... .

Answer and Comments: 3. Yet another problem on solving a tri-

angle. There were already two questions in Paper I (Q.4 and Q.11)
where it was given that a, b, c denote the lengths of the sides opposite
to the vertices A, B, C respectively. This is such a standard practice
that many times in the past, JEE questions have been asked without
this elaboration. One wonders why this has been done now and that
too several times. Perhaps the idea is to help those who have studied
in other media. The notation r for the inradius is also standard.
Coming back to the problems, there is a huge number of identities
involving the sides, the angles, the inradius, the circumradius, the area,
the perimeter and so on of a triangle. It is therefore often possible to
arrive at the answer in a variety of ways (see Chapter 11 for many
illustrations of this.) Usually, it is a good idea to spend some time to
see which approach will be better for a given problem. In the present
problem, we are given the area, say ∆ of the triangle and asked to find
the inradius r. There is a simple formula which connects these two.,
viz.
∆
r= (1)
s
where s = 12 (a + b + c) is the semi-perimeter of the triangle. We are
already given a and b. So we would be done if we can find the third
side c. This can be done by yet another formula which expresses the
area directly in terms of the sides, viz.
16∆2 = (a + b + c)(a + b − c)(a − b + c)(b + c − a) (2)
Here everything except c is known and so substituting their values, we
shall get an equation in c. superficially this will be a fourth degree
equation in c. But this need not deter us. If we carefully group the
factors, we see that there are no odd degree terms in c. (This is similar
to the equation
√ in m we encountered in Q.14 of Paper I.) So, if we put
∆2 = (15 3)2 = 675, a = 6 and b = 10 in (2) we get
10800 = (16 + c)(16 − c)(c − 4)(c + 4) = (256 − c2 )(c2 − 16) (3)

49
 If we put u = c2 , this is a quadratic in u. Unfortunately, the arithmetric
involved would be rather complicated. (This was also the case in Q.14
of Paper I. But there we could sneak by looking at the given answers.
In the present problem this is not easy since the question does not ask
for c but something else.) So, let us try some other formula for the
area. There is a formula which gives the area in terms of any two sides
and the included angle, viz.
1
∆ = ab sin C (4)
2
√
√ 3
which gives 15 3 = 30 sin C and hence sin C = . Hence cos C =
2
1 1
± . But as 6 C is given to be obtuse, we have cos C = − . The cosine
2 2
formula
a2 + b2 − c2
cos C = (5)
2ab
now gives a2 + b2 − c2 = −ab i.e. c2 = a2 + b2 + ab = 36 + 100 + 60 =
196. Hence c = 14. (We can now easily verify that this value of c
does satisfy (3). But that is an afterthought.) Therefore the semi-
perimeter
√ s equals 15 and hence using (1) we get the inradius r equals
3. Therefore r 2 = 3.
The problem is easy provided you identify the easy approach cor-
rectly.
Q.8 Let a1 , a2 , . . . , a11 be real numbers satisfying a1 = 15, 27 − a2 > 0 and
ak = 2ak−1 − ak−2
a2 + a22 + . . . + a211
for k = 3, 4, . . . , 11. If 1 = 90, then the value of
11
a1 + a2 + . . . + a11
is .... .
11

Answer and Comments: 0. If we rewrite the formula for ak as

ak − ak−1 = ak−1 − ak−2 (1)
for k ≥ 3, then it is clear that the numbers are in an A.P. Once this
idea strikes the rest of the solution is routine. We are given that the

50
 first term of this A.P. is 15 but we are not given the common difference.
Instead, we are given the sum of the squares of the terms. From this we
first have to find the common difference, say d. We have the formula
11 10
a2k (15 + kd)2
X X
=
k=1 k=0
10 10
2
k2
X X
= 11 × 225 + 30d k+d
k=0 k=0
10 × 11 × 21 2
= 11 × 225 + 30d × 55 + d (2)
6
We are given that this sum equals 990. Canceling the factors 11 and 5
from all the terms this gives us a quadratic in d, viz.

7d2 + 30d + 27 = 0 (3)

√
−15 ± 225 − 189 −15 ± 6 9
whose roots are = , i.e. −3 and − . With
7 7 7
these values a2 becomes 15 − 3 and 15 − (9/7). The latter possibility
contradicts a2 < 13.5. Hence we must have d = −3. To get the answer
a1 + a2 + . . . + a11
we can evaluate by a summation formula. But that
11
is not necessary. The desired number is the arithmetic mean of an odd
number of terms in an A.P. So it simply equals the middle term, i.e.
the 6-th term in the present case. Since a6 = a1 + 5d = 15 − 15 = 0,
the given expression is 0.
This is a straightforward problem on arithmetic progressions. But
some twists are given. For example, it is not given directly that the
numbers are in an A.P. Secondly, the common difference d has to be
obtained after solving a quadratic in d and discarding one of the roots
using the condition on the second term. And finally, the candidate
also has to realise that the arithmetic mean of the terms of an A.P. is
the middle term (or the average of the middle two terms in case their
number is even). The last twist does test some alertness on the part of
the candidate. The other twists only serve to complicate the problem.

Q.9 For a square matrix M, let adjM denote its adjoint and for a real
number k, let [k] denote the largest integer less than or equal to k.
Now for a positive real number k, let

51
 √ √  √ 
−1 2 k 2 k
2k√ 0 2k − 1 √k
 

A =  2 √k 1 −2k  and B =  1 −√2k 0√ 2 k .

   

−2 k 2k −1 − k −2 k 0
6
If det(adjA) + det(adjB) = 10 , then [k] equals .... .

Answer and Comments: 4. Yet another problem where the main

idea has been given several twists. The main idea of this problem is
the determinant of the adjoint of a square matrix. If M is an n × n
matrix then we have the following result about its relationship with its
adjoint.
Madj(M) = DIn (1)
where D is the determinant of M and In is the identity matrix of order
n. Note that the matrix on the R.H.S. is a diagonal matrix of order n
whose diagonal entries all equal D. Obviously, its determinant is D n .
On the other hand the L.H.S. is a product of two matrices and so its
determinant is the product of the determinants of the two matrices.
Keeping in mind that det(M) = D, (1) implies that
Ddet(adjM) = D n (2)
If M is non-singular, we can divide by D to get
det(adjM) = D n−1 (3)
which remains valid even when M is singular, for in that case D = 0
and the R.H.S. of (1) is the zero matrix, whence adj(M) is also singular
and hence has a vanishing determinant. This is the basic formula we
need in the present problem. We have included its derivation because
at the JEE level this formula is not so well-known. We apply this
formula with n = 3 to the 3 × 3 matrices A and B. A direct calculation
gives
√ √
2k√−1 2 k 2 k
det(A) = 2 √k 1 −2k
−2 k 2k −1
√ √ √ √ √ √
= (2k − 1)(4k 2 − 1) + 2 k(4k k + 2 k) + 2 k(4k k + 2 k)
= 8k 3 − 4k 2 − 2k + 1 + 8k 2 + 4k + 8k 2 + 4k
= 8k 3 + 12k 2 + 6k + 1
= (2k + 1)3 (4)

52
We can similarly calculate det(B). But there is a better way. Note that
B is skew-symmetric. So, by definition, B t = −B where the exponent
t stands for the transpose and not for the power. Since a matrix has
the same determinant as its transpose, we have det(B) = det(B t ) =
det(−B) = (−1)3 det(B) = −det(B) which shows that det(B) = 0.
(More generally, the same argument shows that the determinant of
every skew-symmetric matrix of odd order is 0.)
Now that we know the determinants of the matrices A and B, (3)
gives us the determinants of their adjoints as ((2k + 1)3 )2 = (2k + 1)6
and 0 respectively. We are given that these add up to 106 . Thus we
have 2k + 1 = ±10. The negative sign is excluded since k is given to
be positive. So we have 2k + 1 = 10 and hence k = 9/2. Therefore the
integral part [k] equals 4.
The problem is a good application of the formula for the adjoint of
a determinant. But this formula is not so well known at the JEE level.
Those who do not know it will foolishly try to calculate det(adjA)
by actually writing down all the nine entries of the adjoint matrix.
Even those who know the formula are likely to commit a numerical
slip in the calculation of det(A). The addition of the matrix B to the
problem only serves to complicate it. The work needed to show that
its determinant is 0 is quite independent of that needed to calculate
the determinant of A, which itself is quite substantial. So, those who
calculate det(A) correctly but miss on det(B) get heavily penalised as
there is no provision for partial credit. Finally, it is not clear what
is achieved by designing the problem so as to make k come out to
be fractional. The knowledge of the definition of the integral part of
a number has been tested several times, and in any case, it is very
elementary as compared to the knowledge of a formula like (3) or the
knowledge of the fact that every skew-symmetric matrix of an odd order
is singular. Why unnecessarily add to the work of a candidate when
it is totally irrelevant to the main theme of the problem? Perhaps the
idea was to make the answer come out as an integer. But that could
have as well been served by asking the value of 2k rather than k. Or,
still better, if the sum of the determinants of the adjoints of the two
matrices were given as 116 instead of 106 , then k itself would come out
as an integer, viz. 5 and this ugly addition of the integral part could
have been avoided.

53
 Problems testing several (possibly independent) ideas can be asked
as full length questions where a candidate has to show his work and
therefore gets some partial credit. In a multiple choice test with a large
number of questions, where a candidate can barely spend a minute or
two on each question, designing such questions is not only unfair to the
candidates, but it also distorts the selection. A good candidate may
get rejected despite despite doing nearly all the hard work correctly
but then committing some trivial slip at the end (e.g. writing [9/2] as
5 instead of 4 in the present problem). This is rather like rejecting an
otherwise excellent essay on the biography of Mahatma Gandhi simply
because it gives his birth date wrong. (Another example occurred in
Q.28 of Paper I.) The present problem would have been an excellent
question in a conventional examination. But the present multiple choice
format of JEE has killed it.
√
Q.10 Two parallel chords of a circle of radius 2 are at a distance 3 + 1
apart. If the chords subtend at the centre angles of π/k and 2π/k,
where k > 0, then the value of [k] is .... .

Answer and Comments: 3. Yet another problem where the integral

part had to be introduced to make the answer come out as an integer.
The problem itself is a simple trigonometric problem. As the distance
between the two chords exceeds 2, they must lie on opposite sides of
the centre. Let d1 , d2 be their perpendicular distances from the centre.
Then we have
√
d1 + d2 = 3 + 1 (1)

If θ is the angle which a chord of a circle of radius r subtends at the

centre, then its distance from the centre is r cos(θ/2). In our problem,
r = 2 and the angles subtended by the chords are given to be π/k and
2π/k. Substituting into (1) we get
√
2 cos(π/2k) + 2 cos(π/k) = 3 + 1 (2)

Let us call π/2k as θ. Then (2) becomes a trigonometric equation in

θ, viz.
√
3 1
cos θ + cos 2θ = + (3)
2 2
54
To solve this we reduce it to a quadratic in cos θ, viz.
√
2 3 3
2 cos θ + cos θ − − =0 (4)
2 2
q √
−1 ± 13 + 4 3
whose roots are cos θ = . The negative square root is
4
to be discarded since it would
√ make cos θ√ < −1. Further simplification
hinges on recognising 13 + 4 3 as (1 + 2 3)2 . We now have
√ √
−1 + 1 + 2 3 3
cos θ = = (5)
4 2
which gives θ = π/6. So, we have π/2k = π/6 which gives k = 3 and
hence [k] = 3 too. As k itself comes out to be an integer, asking to find
out its integral part is only a dirty trick played with the candidates by
inducing them to think that k would come out to be fractional.
We have solved the trigonometric equation (3) systematically and
as in Q.11 of Paper I, the success depended on recognising a simple
surd as the square of another simple surd. (It is significant that that
question was also on trigonometry.) But when you merely have to guess
π π
a solution of (3), we rewrite the R.H.S. as cos + cos . In other words
6 3
it is the sum of the cosines of two angles one of which is the double
of the other. But so is the L.H.S. So equating the smaller angles of
π
both the sides we get θ = . This approach is time saving but not
6
mathematically sound. It will be so if we show that if α, β are two
angles in (0, π/2), then cos α + cos 2α = cos β + cos 2β can hold only
when α = β. This can be done algebraically. But it follows more slickly
if we observe that if α < β (say), then we would also have 2α < 2β and
hence cos α > cos β and also cos 2α > cos 2β since the cosine function is
strictly decreasing on the interval [0, π] which contains α, β, 2α and 2β.
But then we would have cos α+cos 2α > cos β +cos 2β, a contradiction.
This problem is of the same spirit as the Main Problem of Chapter 10.
In both the problems, a geometric data is reduced to a trigonometric
equation and the solution to this equation leads to a solution of the
original problem. In other words, it is a problem where trigonometric
equations are applied to solve a problem in geometry. Such problems
are not very common and the paper-setters have come up with a good

55
one. It is tempting to bypass trigonometric equations and try to solve
the problem by getting a system of algebraic equations for d1 and d2 .
We already have one such equation, viz. (1). The second one can be
obtained by using the fact that the angle subtended at the centre by one
of the chords is double that subtended by the other. Let these angles
be α and 2α. Without loss of generality, we assume that d1 ≥ d2 . Then
we have
q
4 − d21
sin(α/2) = (6)
2
d1
cos(α/2) = (7)
2
q
4 − d22
and sin α = (8)
2
Eliminating α from these three equations and squaring gives
4 − d22 = d21 (4 − d21 ) (9)
We now have to solve (1) and √ (9) simultaneously. Once again, we can
guess a solution in which d1 = 3 and d2 = 1. But arriving at it is not
easy. We can easily eliminate one of d1 and d2 . But no matter which one
we eliminate, we shall be left with a fourth degree equation in the other
variable which will not be easy to solve, because it will involve terms of
degree one too in that variable. So, in this case trigonometric equations
is the only way. A similar situation occurred in the solution to the Main
Problem of Chapter 10. As elaborated in Comment No. 1 there, if we
try to solve that problem by reducing the data to a cubic equation, we
are in trouble because the cubic we get has no obvious root and hence
is not easy to solve. In fact, as shown there, sometimes a cubic can be
solved by converting it to a suitable trigonometric equation.
The Main Problem of Chapter 10 was a full length question in 1994
JEE. It would have been better if the present problem were also a full
length question because in that case it would be possible to see if the
candidate actually solves (3) or guesses a solution and substantiates
it with reasoning (given above) or merely guesses a solution and runs
away. In the new JEE format, the runner is the winner as his scrupulous
competitors foolishly waste their time in giving justification to their
own conscience. A telling example of the gentleman finishing the last.

56
Q.11 Let f be a function defined on IR (the set of real numbers) such that

f ′ (x) = 2010(x − 2009)(x − 2010)2 (x − 2011)3(x − 2012)4

for all x ∈ IR. If g is a function defined on IR with values in (0, ∞)

such that f (x) = ln g(x) for all x ∈∈ IR, then the number of points in
IR at which g has a local maximum is .... .

Answer and Comments: 1. The number 2010 and its neighbours

have absolutely no role in the problem. The number 2010 is indicative
of the fact that the question appears in an examination held in the year
2010. This practice is consistently followed in the International Math-
ematics Olympiads. In JEE 2005, Q.13 of the screening paper had the
figure 2005 thrown for no mathematical reason. (See the Commentary
for that year.)
Now, coming to the problem itself, the function g(x) is specified
by f (x) = ln g(x). This is equivalent to defining g(x) as ef (x) . In
fact, with this definition there would be no need to specify that g(x)
takes values in the interval (0, ∞). One really wonders why g(x) has
been specified so indirectly. Just to give an unwarranted twist to the
problem perhaps. We are asked how many local maxima g has. For
this we can differentiate and get

g ′ (x) = f ′ (x)ef (x) (1)

Since the exponential function is always positive, the critical points of

g are the same as those of f and the problem of finding the maxima
of g is the same as that of finding those of f . We could have also seen
this without differentiation. Since the exponential function is strictly
increasing, the function ef (x) increases or decreases according as f (x).
Hence the two have the same points of local maxima/minima.
So, we now focus entirely on f (x). We are not given f (x). But we
are given f ′ (x) and that is what we really need. So even though we can
find f (x) by integrating f ′ (x), it is a waste of time to do that. (See
Exercise (17.16)(a) for a similar situation.) We are given

f ′ (x) = 2010(x − 2009)(x − 2010)2 (x − 2011)3(x − 2012)4 (2)

57
 The constant 2010 is positive and hence plays no role in the sign de-
termination of f ′ (x). We simply ignore it and concentrate on the re-
maining factors of f ′ (x). f ′ has four zeros, 2009, 2010, 2011 and 2012
of multiplicities 1,2,3 and 4 respectively. A change of sign occurs only
at zeros of odd multiplicities, i.e. at 2009 and 2011. For x < 2009,
all the ten non-constant factors of f ′ (x) are negative and so f ′ (x) is
positive. If x lies between 2009 to 2010 f ′ (x) is negative since nine of
its factors are negative. So f ′ changes sign from positive to negative at
2009. Hence the behaviour of f changes from increasing to decreasing
at 2009. Therefore there is a local maximum of f at 2009. This is also
a local maximum of g(x) as noted above.
The only other zero of odd multiplicity of f ′ (x) is 2011. For
x ∈ (2010, 2011), f ′ (x) < 0 since three factors of f ′ are positive and
the remaining 7 are negative. But for x ∈ (2011, 2012), 6 factors are
positive and 4 factors are negative.So f ′ changes sign from negative
to positive at 2011 and correspondingly the behaviour of f (x) changes
from decreasing to increasing at 2012. So there is a local minimum
of f (x) and hence of g(x) at 2011. All told, f (x) and hence g(x) has
precisely one local maximum.
This is a simple problem where the extrema of a function have
to be found knowing only its derivative. Probably to make up for its
simplicity, the function g(x) has been added. In the printed version
of the original question paper, the dash in f ′ (x) is alarmingly close to
the left parenthesis. It appears as f ′(x) and actually almost as f(x).
′

So, unless one reads it very carefully, it is likely to be read as f (x).

This completely changes the problem and makes it considerably more
difficult. Possible sources of confusion like this need to be weeded out
at the time of proof reading. A mis-spelt word generally does not cause
so much confusion since it can often be corrected from the context. But
confusing f ′ (x) with f (x) can be killing.

SECTION III
Paragraph Type

This section contains two paragraphs. Based on each of the paragraphs, there
are three multiple choice questions. Each of these questions has four choices

58
out of which ONLY ONE is correct.

Paragraph for Questions 12 to 14

Consider the polynomial

f (x) = 1 + 2x + 3x2 + 4x3 .

Let s be the sum of all distinct real roots of f (x) and let t = |s|.

Q.12 The real number s lies in the interval

(A) (−1/4, 0) (B) (−11, −3/4) (C) (−3/4, −1/2) (D) (0, 1/4)

Answer and Comments: (C). As f (x) is a cubic polynomial with real

coefficients, it has either one or three real roots. If it had three distinct
real roots, then by Rolle’s theorem its derivative f ′ (x) = 12x2 + 6x + 2
would have at least two zeros. But this quadratic polynomial has a
negative discriminant 36 − 96 = −60. So, f (x) has only one real root,
say α and s equals this lone real root. It is not easy to identify it
explicitly. But the Intermediate Value Property (IVP) of continuous
functions can be applied to locate α within an interval. Clearly α < 0
and so (D) is ruled out as an answer. To find out which of the remaining
is the correct answer, we can evaluate f at the endpoints of the given
intervals till we come across an interval (a, b) such that f (a) and f (b)
are of opposite signs. But this is essentially a trial and error method. To
carry out the search for α more systematically, we imitate the standard
proof of the IVP given in Comment No. 3 of Chapter 16. We first
identify some reasonably small interval which contains α. Clearly we
have f (0) = 1 > 0 while f (−1) = 1 − 2 + 3 − 4 = −2 < 0. Hence α lies
in (−1, 0). To narrow the search further, we evaluate f at the midpoint,
3 4
−1/2. Again a direct computation gives f (−1/2) = 1 − 1 + − =
4 8
1
> 0. So α lies somewhere in (−1, −1/2). None of the given intervals
4
contains this interval. So we need to carry out the search further. So
again we evaluate f at the mid-point −3/4 of this interval. This time
3 27 27 1
we have f (−3/4) = 1 − + − = − < 0. Thus we see that
2 16 16 2
the root α lies in the interval (−3/4, −1/2). Luckily this happens to

59
 coincide with the interval given in (C). But it would have sufficed if it
were merely contained in it.
The method adopted here is called successive bisection or binary
search because at each stage, we are narrowing down the search to one
of the two halves of the previous interval. If we carry out the binary
search a sufficient number of times, we can approximate the (unknown)
root as closely as we want. Of course, to begin with the initial interval
has to be identified by trial. But thereafter the method proceeds like
an algorithm. The initial interval should not be extravagantly large.
But even if it is, the binary search method is fairly efficient because the
powers of 2 increase very fast (an exponential growth in the language
of Exercise (6.51).
The problem is a good but standard application of two important
results in theoretical calculus, viz. Rolle’s theorem and the IVP.
Q.13 The area bounded by the curve y = f (x) and the lines x = 0, y = 0
and x = t, lies in the interval

(A) (3/4, 3) (B) (21/64, 11/16) (C) (9, 10) (D) (0, 21/64)

Answer and Comments: (A). Denote the area by A. We know that

Z t
A= 1 + 2x + 3x2 + 4x3 dx = t + t2 + t3 + t4 (1)
0
Here t = |s| = −α where α is the only real root of f (x). Let g(x) =
x + x2 + x3 + x4 . Then A = g(t). From the last question we know that
t ∈ (1/2, 3/4). As the function g(x) is obviously strictly increasing for
all x > 0 we definitely have the following estimate on A = g(t)
1 3
g( ) < A < g( ) (2)
2 4
To calculate the functional values of g, we simplify g(x) to x(1+x+x2 +
x(x4 − 1)
x3 ) = which is valid for all x 6= 1. For x < 1 we rewrite this
x−1
x(1 − x4 ) 15 81 3 × 175
as . Then g(1/2) = and g(3/4) = 3(1 − )= =
1−x 16 256 256
525
. Hence from (2), we get
256
15 525
A∈( , ) (3)
16 256
60
 If we can show that this interval is contained in any one of the four given
intervals, then that interval would be the answer. The intervals in (C)
and (D) are clearly ruled out because they are disjoint from the interval
15 525
( , ). For easy comparison, we bring all fractions involved to a
16 256
240 525
common denominator 256. Then A lies in ( , ). The intervals in
256 256
192 768 84 176
(A) and (B) are, respectively, ( , ) and ( ). Out of these
256 256 256 256
two only the first one contains the interval in (3). So the correct choice
is (A).
The problem is straightforward. But the arithmetic involved is
a little too much. The problem would have been more interesting if
none of the given choices contained the interval in (3). That does not
necessarily mean that the question is wrong. It only means that the
estimate in (3) is too crude and needs to be improved. For this, we
need a sharper estimate on the root α of f (x). We already know that
α lies in the interval (−3/4, −1/2). To get a better estimate we divide
5 111
this interval at its midpoint −5/8. Then we have f (− = > 0. So
8 256
3 5
the root α lies in the interval (− , − ) and, as a result we now have
4 8
5 3 5 3 3
< t < and further g( ) < A < g( ). We already know g( ) as
8 4 8 4 4
525 5
. The calculation of g( ) will introduce fractions with denominators
256 8
84 = 4096. We can then see if this new interval is contained in any one
of the given ones. If not, we refine the estimate for α still further
and so on. Obviously the computations involved would be horrendous
if attempted by hand. But the problem would have really tested the
thinking ability of the candidate. As it stands, the given problem
involves less thinking and more computation. The paper-setters are
apparently aware of this and in order to simplify the calculations they
have given the function f (x) in such a way that its antiderivative g(x)
will come in a very simple form, viz. as a polynomial in which all
coefficients are 1.

Q.14 The function f ′ (x) is

(A) increasing in (−t, −1/4) and decreasing in (−1/4, t)

61
(B) decreasing in (−t, −1/4) and decreasing in (−1/4, t)
(C) increasing in (−t, t)
(D) decreasing in (−t, t)

Answer and Comments: (B). We have f ′ (x) = 2 + 6x + 12x2 and

the question asks for the increasing/decreasing behaviour of f ′ (x) on
certain intervals. As f ′ (x) is differentiable everywhere, the problem
reduces to checking the sign of f ′′ (x) = 6 + 24x on these intervals. As
the factor 6 is positive, we might as well check the sign of 4x + 1. Let
us call this function h(x). It has only one zero, viz. −1/4. And h(x)
is positive for x > −1/4 and negative for x < −1/4. Now, all the
given intervals involve t whose exact value is not known to us. But we
already have estimates on its value and, if need arises, we know how to
improve these estimates. Let us first see what information we can get
from the estimate we already have, viz.
1 3
< t < (4)
2 4
which also means
3 1
− < −t < − (5)
4 2
(which is the same as the estimate for α we had obtained since α = −t.)
Put together this means
1 1 3 3
(− , ) ⊂ (−t, t) ⊂ (− , ) (6)
2 2 4 4
1
In particular, the point − at which h changes sign lies in (−t, t).
4
So there is a change of increasing/decreasing behaviour of f ′ (x) in
these intervals. This rules out (C) and (D) as possible answers. In
the remaining two options, the interval (−t, t) has been split into two
parts at the point −1/4, which is the point at which h(x) changes sign
from negative to positive. So f ′ (x) is decreasing on (−t, −1/4) and
increasing on (−1/4, t). Therefore (B) is the correct answer.
This bunch of questions fits for a paragraph because the answer
to the first question is crucially needed in the other two. But there is

62
 considerable duplication of ideas. Moreover, the essential idea, viz. the
increasing/decreasing behaviour of a function has already occurred in
Q.7 of Paper I and again in Q.11 of Section II.

Paragraph for Questions 15 to 17

x2 y2
Tangents are drawn from the point P (3, 4) to the ellipse + = 1
9 4
touching the ellipse at the points A and B.

Q.15 The coordinates of A and B are

√ !
8 2 161 9 8
 
(A) (3,0) and (0,2) (B) − , and ( − ,
√ !
5 15 5 5
8 2 161 9 8
 
(C) − , and (0,2) (D) (3,0) and − ,
5 15 5 5

Answer and Comments: (D). We are asked to find the points of

contact of the tangents from the given point P , i.e. points on the
ellipse such that the tangents at them pass through P . For this it is
convenient to take parametric coordinates of the given ellipse, viz.

x = 3 cos θ, y = 2 sin θ (1)

Now suppose (3 cos θ, 2 sin θ) is a point on the ellipse. Then the slope
2 cos θ
of the tangent at this point is − and so its equation is
3 sin θ
2 cos θ
y − 2 sin θ = − (x − 3 cos θ) (2)
3 sin θ
which simplifies to

3 sin θ y + 2 cos θ x = 6 (3)

(which can be derived in other ways too). As the tangent passes

through P (3, 4), we get

2 sin θ + cos θ = 1 (4)

63
 which is a trigonometric equation in θ. To solve it we introduce an
angle α by
2 1
cos α = √ and sin α = √ (5)
5 5
(We could also have introduced α as the acute angle for which tan α =
1
.) Then (4) becomes
2
1
sin(α + θ) = √ (6)
5
The R.H.S. equals sin α. Hence the two solutions are given by α +
θ = α and 2α + θ = π, which gives θ = 0 or π − 2α. The first
value gives (3 cos θ, 2 sin θ) as (3, 0). The second value gives it as
(−3 cos 2α, 2 sin 2α), which from the values of sin α and cos α comes
9 8
out as (− , ). Therefore these are the points of contact of the two
5 5
tangents from P .

Q.16 The orthocentre of the triangle P AB is

(A) (5, 8/7) (B) (7/5, 25/8) (C) (11/5, 8/5) (D) (8/25, 7/5)

Answer and Comments: (C). There is a formula (given in Comment

No. 3 of Chapter 8) which gives the orthocentre of a triangle directly in
terms of the coordinates of its vertices. But it is far too complicated.
Here we have a special triangle, viz. a triangle formed by a point
outside a conic and the chord of contact of this point w.r.t. that conic.
But even in this special case, there is no handy readymade formula. So
we proceed from scratch by finding two of the altitudes of the triangle
P AB and identifying their point of intersection. In the present case,
from the last question, we can take A as (3, 0) and B as (−9/5, 8/5)
(without loss of generality). Since P = (3, 4). So, obviously the side
P A is along the line x = 3. Hence the altitude through B is the line
8
y = . Among the given four choices (C) is the only one where the
5
8
y-coordinate is . So without any further work, if at all one of the
5
answers has to be correct, it must be (C).

64
 But if we do not want to take an unfair advantage of the carelessness
on the part of the paper-setters, then we need one more altitude of the
triangle P AB. It is preferable to take the altitude through P because to
find it we would first need to find the equation of the line AB and this
equation will also be needed in the next question. (This kind of a non-
mathematical alertness pays off sometimes.) Since A = (3, 0) and B =
8/5
(−9/5, 8/5), we get the equation of the line AB as y = (x−3)
−9/5 − 3
which, upon simplification becomes
1
y = − x+1 (7)
3
Hence the slope of the altitude through P is 3 and therefore its equation
is

y − 4 = 3(x − 3) (8)
8
Solving this simultaneously with y = gives x = 11/3. Therefore the
5
11 8
orthocentre of of the triangle P AB is at ( , ).
5 5
Q.17 The equation of the locus of the point whose distances from the point
P and the line AB are equal, is

(A) 9x2 + y 2 − 6xy − 54x − 62y + 241 = 0

(B) x2 + 9y 2 + 6xy − 54x − 62y − 241 = 0
(C) 9x2 + 9y 2 − 6xy − 54x − 62y − 241 = 0
(D) x2 + y 2 − 2xy + 27x + 31y − 120 = 0

Answer and Comments: (A). By definition, the locus is a parabola

with focus at P and the line AB as its directrix. But this is not much
of a help in finding the equation. That is best done by equating the
two distances. Let (h, k) be the moving point. Its distance from P is
q
(h − 3)2 + (k − 4)2 . We already know the equation of AB from (7).
|k + 1 h − 1|
So the distance of (h, k) from it is q 3 . Equating the distances
1 + 1/9

65
 and squaring gives

(h + 3k − 3)2
(h − 3)2 + (k − 4)2 = (9)
10
Replacing h, k by x, y and simplifying, the locus is

10(x2 + y 2 − 6x − 8y + 25) = x2 + 9y 2 + 6xy − 6x − 18y + 9 (10)

i.e. 9x2 + y 2 − 6xy − 54x − 62y + 241 = 0.

The entire paragraph is routine and computational.

SECTION IV

Matrix Type

This section contains 2 questions. Each question has four statements (A,
B, C and D) given in Column I and five statements (p, q, r, s and t) in
Column II. Any given statement in Column I can have correct matching
with one or more statement(s) given in Column II.

Q.18 Match the statements in Column I with those in Column II.

[Note : Here z takes values in the complex plane and Im z and Re z
denote, respectively, the imaginary part and the real part of z respec-
tively.]

66
 Column I Column II
(A) The set of points z satisfying (p) an ellipse with eccen-
4
tricity
|z − i|z|| = |z + i|z| 5

is contained in or equal to
(q) the set of points z sat-
(B) The set of points z satisfying isfying Im z = 0

|z + 4| + |z − 4|
(r) the set of points z sat-
is contained in or equal to isfying | Im z| ≤ 1
(C) If |w| = 2, then the set of
points z satisfying z = w − (s) the set of points z sat-
1
isfying | Re z| ≤ 2
w
is contained in or equal to

(D) If |w| = 1, then the set of (t) the set of points z sat-
points z satisfying z = w + isfying |z| ≤ 3
1
w
is contained in or equal to

Answer and Comments: (A) → (q, r) (B) → (p) (C) → (p, s, t)

(D) → (q, r, s, t).
If this were a one-to-one matching, the answers would be easier
to arrive at. The sets given in Column II are either given or are easy
to identify geometrically. The entries in Column I are easy to iden-
tify qualitatively. And if we are given beforehand that each entry in
Column I has only one match then it can usually be found from this
qualitative description. But in the present problem several entries may
match with the same entry. And that makes it necessary to identify
the sets in Column I more precisely. This can be done by translating
the descriptions in terms of the real and imaginary parts of z. As usual
we denote these by x and y respectively.
Let us first describe the sets in Column II in terms of x and y. For
simplicity, denote these subsets by P, Q, R, S and T respectively. The

67
set P is not uniquely defined. But for the remaining sets, we have

Q = {(x, y) : y = 0} (1)
R = {(x, y) : |y| ≤ 1} (2)
S = {(x, y) : |x| ≤ 2} (3)
T = {(x, y) : x2 + y 2 ≤ 9} (4)

It is easy to describe these geometrically. Q is the x-axis, R is the

horizontal strip bounded by the lines y = ±1, S is the vertical strip
bounded by the lines x = ±2 and T is the disc of radius 3 centred at
the origin. Clearly Q ⊂ R and so if (q) is a correct answer to some
entry in Column I, then (r) is also correct.
Let us now describe the sets in Column I. For simplicity we denote
them by A, B, C and D respectively. In (A), no matter what z is, the
points i|z| and −i|z| are on the y-axis, symmetrically located w.r.t. the
xaxis. The condition in (A) implies that z is equidistant from these
two points and hence lies on the perpendicular bisector of the segment
joining them. So, z must lie on the x-axis. So it lies in Q and hence in
R too. Thus we have A ⊂ Q and A ⊂ R.
Geometrically, the set B is an ellipse with foci at ±4. Let e be the
eccentricity of the ellipse and let (a, 0), (−a, 0) be its vertices. Then we
have two 2a = 10 whence a = 5. But then ae = 4 implies e = 4/5. So
(p) holds. Also (q) is ruled out because an ellipse cannot be completely
contained in its major axis. To check the remaining options, note first
that the points (±5, 0) lie on the ellipse but are obviously outside the
vertical strip S as well as the disc T . But we must check if the ellipse
lies in the horizontal strip R. For this we need to find the semi-minor
√ 3
axis, say b of the ellipse. This is given by a 1 − e2 = 5 × = 3. So
5
the points (0, ±3) lie on the ellipse but not in R. Hence only (p) holds
for (B).
For C and D, note that w is a complex number which varies over
some subset of the complex plane and z is given as a function of w. In
(C), we can take w = 2 cos θ + 2i sin θ for some θ. Then
1 1 1
= = (cos θ − i sin θ) (5)
w 2(cos θ + i sin θ) 2

68
 This gives
1 1
w− = 2(cos θ + i sin θ) − (cos θ − i sin θ)
w 2
3 5
= cos θ + i sin θ (6)
2 2
So, we get
3 5
C = {(x, y) : x = cos θ, y = sin θ} (7)
2 2
But these are precisely the parametric equations of an ellipse with
major and minor axis 5 and 3 respectively. (Note
√ that the major axis
25 − 9 4
is along the y-axis.) So the eccentricity is = . Hence (p)
5 5
holds. But unlike A, this time the ellipse C is contained in the strip S
as well as the disc T . Hence (s) and (t) also hold true.
Finally, for D the reasoning is analogous to that in (C). This time
1
w is of the form cos θ + i sin θ and so w + comes out to be simply
w
2 cos θ. So the set D is the segment of the x-axis from (−2, 0) to (2, 0).
Clearly (q), (r), (s) and (t) all hold true.
The problem is simple but once the basic ideas are understood,
there is considerable duplication of work. Since every option in each
column has to be tried for every option in the other column, in ef-
fect, 20 questions are asked here. The paper-setters have designed the
numerical data of the problem very carefully. Both (B) and (C) are
ellipse. In fact the second ellipse is similar to the conjugate ellipse of
the first one by a factor of 1/2. As a result, both the ellipses have the
same shape and hence equal eccentricities. But the second ellipse is
smaller and so fits inside a strip and a disc while the first one does not.

Q.19 Match the statements in Column I with the values in Column II.

69
 Column I Column II

(A) A line from the origin meets the lines

x−2 y−1 z+1 (p) −4
= = and
1 −2 1
x − (8/3) y+3 z−1
= = at points P
2 −1 1
and Q (q) 0
respectively. If length P Q = d, then d2
is ... .
(r) 4
(B) The values of x satisfying
3
tan−1 (x + 3) − tan−1 (x − 3) = sin−1 ( )
5
are .... .
(s) 5
(C) Non-zero vectors ~a, ~b, ~c satisfy ~a · ~b = 0,
(~b − ~a) · (~b + ~c) = 0 and 2|~b + ~c| = |~b − ~a|.
If ~a = µ~b + 4~c, then the possible values of (t) 6
µ
are .... .

(D) Let f be a function on [−π, π] given by

f (0) = 9 and f (x) = sin(9x/2)/ sin(x/2)
2 π
Z
for x 6= 0. The value of f (x) dx is
π −π
... .

Answer and Comments: (A) → (t), (B) → (p, r), (C) → (q, s),
(D) → (r).
Unlike in the last question, here the entries in Column I have
numerical answers. In (A) and (D), they have to be found by systematic
calculations. In n(B) and (C) we have to solve certain equations and
so we can, in theory, try all the alternatives in Column II one by one
and see which of them are correct. But this is hardly practicable. Nor
is it mathematically sound. So we tackle each problem systematically.
In (A), taking the parametric equations of the two lies, we see that
the points P and Q are of the form

P = (2 + s, 1 − 2s, −1 + s) (1)

70
 8
and Q = ( + 2t, −3 − t, 1 + t) (2)
3
for some real numbers s and t. We are further given that O (the origin),
P and Q are collinear. Hence the coordinates are proportional. This
gives a system of two equations in s and t, viz.
2+s 1 − 2s −1 + s
= = (3)
(8/3) + 2t −3 − t 1+t
We can equate any two of these ratios to get two equations in s and t.
But they will involve terms in st and hence will not be linear. Instead
we use a few simple properties of ratios to get simpler equations. Call
each of the three ratios above as r. Then if we take any linear combi-
nation of any two numerators and divide it by the corresponding linear
combination of their respective denominators we still get the same ratio
r. We can choose these linear combinations so as to eliminate one of
the two variables. For example, if we simply add the numerators of the
last two ratios and divide it by the sum of their denominators, we get
1 − 2s − 1 + s s
r= = (4)
−3 − t + 1 + t 2
Similarly, we can eliminate t using the first and the third ratio to get
2 + s − 2(−1 + s) 4−s
r= 8 = 2 (5)
3
+ 2t − 2(1 + t) 3

Equating these two, we get a single equation in s, viz.

s 4−s
= 2 (6)
2 3

which can be easily solved to get s = 3. We can similarly eliminate s

to get a linear equation for t. Alternately we can put s = 3 into any of
the two ratios in(3). Doing so for the last two ratios gives
−5 2
= (7)
−3 − t 1+t
which gives t = 1/3. Putting these values into (1) and (2) gives P as
(5, −5, 2) and Q as (10/3, −10/3, 4/3). A direct calculation now gives
d2 = (5/3)2 + (−5/3)2 + (2/3)2 = 54/9 = 6.

71
 Although straightforward, the calculations involved here are fairly
lengthy and prone to errors. Comparatively, the equation we get for
3 3
x in (B) is much easier. Recognising sin−1 ( ) as tan−1 ( ) and taking
5 4
the tangents of both the sides, we get

(x + 3) − (x − 3) 3
= (8)
1 + (x + 3)(x − 3) 4

which simplifies to a quadratic in x, viz. x2 − 8 = 8. This gives ±4 as

possible values of x.
In (C), it is not immediately obvious how to get an equation for µ
from the data. Although there are three vectors ~a, ~b, ~c in the problem,
the vector ~a is given as a linear combination of ~b and ~c, viz.

~a = µ~b + 4~c (9)

So, we can write each of the three given equations in terms of the
vectors ~b and ~c and the scalar µ. When we expand each of the three
given equations it becomes an equation that involves the three dot
products ~b · ~b, ~b · ~c and ~c · ~c. Specifically, the first equation ~a · ~b = 0
becomes (µ~b + 4~c) · ~b = 0 and gives

µ~b · ~b + 4~b · ~c = 0 (10)

Similarly, the second equation, (~b −~a) · (~b + ~c) = 0 becomes ~b · (~b + ~c) =
(µ~b + 4~c) · (~b + ~c) or equivalently,

(µ − 1)~b · ~b + (µ + 3)~b · ~c + 4~c · ~c = 0 (11)

The third equation, viz. 2|~b + ~c| = |~b − ~a| is ostensibly in terms of the
lengths of the vectors. But if we square both the sides we can write it
as an equation in dot products, viz.

4(~b · ~b + 2~b · ~c + ~c · ~c) = ~b · ~b − 2~b · ~a + ~a · ~a (12)

Putting ~a = µ~b + 4~c and collecting the like terms, we get

(µ2 − 2µ − 3)~b · ~b + (8µ − 16)~b · ~c + 12~c · ~c = 0 (13)

72
We regard (10), (11) and (13) as a system of three homogeneous linear
equations in three unknowns ~b · ~b, ~b · ~c and ~c · ~c. Since the vectors ~b, ~c
are given to be non-zero, so are the dot products ~b · ~b and ~c · ~c. Hence
this system has at least one non-trivial solution. Therefore setting its
determinant equal to 0 we get the desired and much awaited equation
for µ, viz.

µ 4 0
µ−1 µ+3 4 =0 (14)
2
µ − 2µ − 3 8µ − 16 12

Dividing the last column by 4 and then expanding directly, this reduces
to a quadratic in µ, viz.

µ(−5µ + 25) + 4(µ2 − 5µ) = 0 (15)

i.e. −µ2 + 5µ = 0. Hence the two possible values of µ are 0 and 5.

Unlike Part (A) which is most computational, here the thinking
ability is also tested. But the computations involved after getting the
key idea are also substantial. The work can be shortened considerably
by introducing suitable coordinates, taking advantage of the orthogo-
nality of ~a and ~b. So, without loss of generality, assume ~a = a~i and
~b = b~j. Then ~c = 1 (~a − µ~b) = 1 (a~i + (4 − µ)b~j) and the condi-
4 4
tion (~b − ~a) · (~b + ~c) = 0 becomes a2 = (4 − µ)b2 while the condition
2|~b − ~c| = |~b − ~a| reduces to 4(a2 + b2 ) = a2 + (4 − µ)2 b2 . Eliminating
a in these two equaltions gives

4(5 − µ)b2 = [(4 − µ) + (4 − µ)2 ]b2

As b 6= 0, canceling it we get a quadratic in µ which is the same as

before. (Actually, the coefficient of b2 on the R.H.S. can be written as
(4 − µ)(5 − µ), which makes it possible to solve the quadratic simply
by inspection.)
However, even this slicker solution takes time way beyond that
justified by the credit allotted to the question. The problem is more
suitable for a full length question. But the present JEE does not permit
such questions.

73
 Finally, let us denote the integral in (D) by I. Note that the
integrand is not discontinuous at 0, because by L’ôpital’s rule (or sim-
ply by multiplying and dividing by 9x/2), one can easily show that
sin(9x/2)
lim+ equals 9. By a change of variable x = 2θ we get
x→0 sin(x/2)
sin 9θ
Z π/2
I = 2 dθ (16)
−π/2 sin θ
Z π/2
sin 9θ
= 4 dθ (17)
0 sin θ
where the last equality follows since the integrand is an even function
of θ. It is possible to express sin 9θ as a polynomial of degree 9 in sin θ
(analogous to sin 3θ = 3 sin θ − 4 sin3 θ) and then integrate each term.
But that would be very complicated. So we try some other method.
We note first that there is nothing special about the integer 9. We
could just as well replace it by any odd positive integer, say 2n + 1.
Let us do that and call the resulting integral as In . In other words,
Z π/2 sin(2n + 1)θ
In = dθ (18)
0 sin θ
With this notation the integral in (9) is simply I4 . Although our interest
is only in this particular integral, it is sometimes easier to tackle the
general problem of evaluating In for every n ≥ 0, if we are able to
find a reduction formula for it, i.e. a formula which expresses In in
terms of In−1 (or other similar integrals with lower values of the suffix.)
This technique of evaluating certain special definite integrals without
identifying antiderivatives of their integrands is explained in detail in
Comments 5 to 9 of Chapter 18.
In the present problem a reduction formula for In can be obtained
as follows. Note that
sin(2n + 1)θ − sin(2n − 1)θ
Z π/2
In − In−1 = dθ
0 sin θ
Z π/2
2 cos 2nθ sin θ
= dθ
0 sin θ
Z π/2
= 2 cos 2nθ dθ
0
1 π/2
= sin 2nθ =0 (19)
n 0

74
 In other words, we have proved that In = In−1 . Of course, not
all reduction formulas are so simple. In fact, many of them are quite
complicated and for many integrals involving an integer parameter, no
reduction formulas exist. But as far as our integral is concerned, by
repeated applications of the reduction formula, we get that In = In−1 =
In−2 = . . . = I2 = I1 = I0 . The last integral I0 has to be computed
sin θ
directly. But that is easy. As the integrand is the constant 1,
sin θ
π π π
we have I0 = and hence In = for every n. In particular I4 = .
2 2 2
Going back to our original integral I, by (17) it equals 4I4 = 2π. Hence
2 π sin(9x/2)
Z
the expression dx equals 4.
π −π sin(x/2)
The problem is easy once the idea of the reduction formula
strikes. The problem is closely related to the problem of showing that
1 − cos mx
Z π
dx equals mπ for every positive integer m. In fact, the
0 1 − cos x
two problems can be converted to each other. This latter problem was
asked as a full length question in JEE 1995. (See Comment No. 8 of
Chapter 18 for a solution.)
Except for (B), all entries in Column I require considerable thinking
and work. It is very unlikely that even a good candidate will be able
to give the time demanded by them.

75
 CONCLUDING REMARKS
The paper-setters have obviously worked very hard. Even where there is
considerable numerical work in a problem, the data has been so designed
that the final answer would look manageable. In other words, the paper-
setters have actually done the computations down to the last end. This has
paid off. There are no mathematical errors and hardly any ambiguities or
obscurities in both the papers. The two exceptions are Q.24 of Paper 1 in
which a reference is made to the ‘nearest directrix’ which is wrong on two
counts. First, a hyperbola has only two directrices. So it would be more
appropriate to say the ‘nearer directrix’. But more seriously, it is not made
clear how nearness is to be measured. One simply has to prseume that it
is in terms of the distance from the point of contact. Secondly, in Q.28 of
Paper 1, the phrase ‘common ratio’ is confusing for a progression all whose
terms vanish. Another place where confusion may arise is Q.10 in Paper 1
where the correct answer is 0 but since more than one answer are allowed
22
to be correct, a candidate may get confused as to whether − π is also to
7
be taken as a correct answer. As already pointed out, in the text of Q.11 of
Paper 2, as it appeared in the actual question papers given to the candidates,
f ′ (x) looked more like f(x).
′
But that is a fault of the proof-readers than the
paper-setters.
The paper-setters do, however, have a tendency to give unwarranted twists
to a problem, which are irrelevant to the main theme of the problem and only
serve to take up additional time on the part of the candidate without testing
any desirable quality on his part. Examples include Q.28 in Paper 1 (on
a telescopic series), Q. 8 in Paper 2 (on arithmetic progressions) , Q. 9 in
Paper 2 (on adjoint matrix) and Q.11 of Paper 2 (on local maximum of a
function). There are also questions where arriving at the solution honestly is
quite time consuming, but guessing it from the alternatives given is easy, e.g.
Q. 14 of Paper 1 and Q. 10 of Paper 2. One wonders if the idea was to test
some non-mathematical alertness on the part of a candidate. In Q.15 and
16 of Paper 1, a scrupulous answer must eliminate certain possibilities. But
these possibilities do not affect the solutions. So a candidate who is blissfully
unaware of them gets rewarded in terms of the time he saves. There are also
problems where in order that the answer comes out as a single digit integer
the problem has been made clumsy.
There is some duplication of ideas. For example, both Q.3 and Q.22 of

76
Paper 1 are based on the same basic property of ω, the complex cube root of
unity. One of these problems could have been replaced so as to make room
for some untouched area such as even and odd functions or a functional
equation. Numerous problems on solid coordinate geometry are asked. At
the JEE level there is not much variety in these problems. So, no harm would
have been committed if one of them were replaced by differential equations
which have been paid only a lip service.
But despite such deficiencies (some of which are essentially matters of
taste) the paper-setters have undoubtedly come up with some interesting
problems, often covering some rare areas. Examples are arithmetic modulo
a prime (Q.17 of Paper 1), the adjoint of a matrix (Q. 9 of Paper 2) and
a geometric application of a trigonometric equation (Q.10 of Paper 2). The
binary search needed for the location of a zero of a continuous function in
Q.12 of Paper 2 is also unprecedented. Although reduction formulas for defi-
nite integrals were commonly asked in the past, after the JEE became totally
objective type, they did not figure. But the paper-setters have managed to
ask a question based on a reduction formula in Part (D) of Q.19 of Paper
2. Part (C) of the same question is also an unusual problem on vectors.
Although questions on solving a triangle are frequent, the way the triangle
is determined in Q.11 of Paper 1 is quite tricky. But what takes the cake
is Q. 27 of Paper 1 where an indiscriminate application of the determinant
criterion for the solution of a homogeneous system of linear equations is a
very tempting trap.
Unfortunately, the multiple choice format and the huge number of routine
questions do not do justice to these good problems. There is simply no time
to solve them honestly. So if a candidate has answered them correctly that is
not necessarilty an indication of any acumen on his part. He could, in fact,
be a very mediocre candidate, as is often revealed in the B. Tech. courses he
takes after getting into the IITs.
One can only dream that a handful of such well-chosen problems is all
that is asked in a two or three hour examination, in which the candidate has
to show all his work. That is how JEE was a long long time ago. Those were
the days.

77