EDUCATIVE COMMENTARY ON

JEE 2012 MATHEMATICS PAPERS

Last revised on 14/12/2022

Contents

Paper 1 2

Paper 2 36

Concluding Remarks 60

The pattern of JEE 2012 is the same as that of the previous year. The
number of questions in Paper 1 has been reduced to a more reasonable 20 for
each subject in each of the two papers. Further, negative marking is restricted
only to those questions where only one of the alternatives is correct. Negative
marking for questions where more than one alternatives are correct often
resulted in a draconian penalty when a candidate could identify some but
not all correct answers. An even more welcome change is that the ‘match the
pairs’ types of questions have been dropped. In the past such questions often
artificially clubbed together several unrelated problems the only commonality
being that their numerical answers were the same.
As in the commentaries in the last few years, the questions in every section
in each paper are listed here in a random order except for those based on
paragraphs. Because of staggering of the questions so as to prepare different
versions of the same question paper (to curb copying), no particular order
of the questions is sacrosanct. As in the past, unless otherwise stated, all
the references made are to the author’s book Educative JEE (Mathematics)
published by Universities Press, Hyderabad.
At the time of writing the first version of this commentary, the fate of
the JEE was uncertain. Despite stiff opposition from certain corners, there
is a move to abolish it totally and base the admissions to all engineering
institutes on a single common examination. But till 2023 the proposal has
not been implemented.
 PAPER 1

Contents

Section I (Single Correct Choice Type) 2

Section II (Multiple Correct Choice Type) 18

Section III (Integer Type)

SECTION I
Single Correct Choice Type

This section contains ten multiple choice questions. Each question has 4
choices out of which ONLY ONE is correct. A correct answer gets 3 points
and an incorrect one −1 point.

x2 + x + 1
!
Q.1 If lim − ax − b = 4, then
x→∞ x+1

(A) a = 1, b = 4 (B) a = 1, b = −4 (C) a = 2, b = −3 (D) a = 2, b = 3

Answer and Comments: (B). The given function, say f (x) equals

(1 − a)x2 + (1 − a − b)x + (1 − b)
f (x) = (1)
x+1

which is a rational function of x, i.e. a ratio of two polynomials in x.

The behaviour of such a function as x → ∞ depends on the degrees of
the numerator and the denominator, say p and q respectively. If p > q
the limit does not exist; if p < q the limit is 0 and when p = q the limit
exists and equals the ratio of the leading coefficients of the numerator
and the denominator. In the present problem, the third possibility is
given to hold. So the numerator must be of degree 1 (which is the
degree of the denominator) and moreover the coefficient of x in the
numerator must equal 4 times that of x in the denominator. This gives

2
 a simple system of two equations in two unknowns, viz.,
1 − a = 0 and 1 − a − b = 4 (2)
solving which we get a = 1 and b = 1 − a − 4 = −4.
This is a good problem, requiring very little computation once you
get the key idea. In working out the problem mentally, it is hardly
necessary to write down the steps (1) and (2) elaborately. Taking the
expression as it stands, we can split the numerator and write the ratio
x2
as + 1 (a commonly adopted trick while finding antiderivatives of
x+1
rational functions). The second term is a constant and does not affect
x2
the finiteness of the limit. The first factor, viz. is of the order
x+1
of x as x → ∞ and the limit of the whole expression cannot be finite
unless this troublesome part is cancelled by some term which is also of
the same order. This gives a = 1 almost instinctively. And once a is
determined, the equation for b is obtained by equating the limit with
4. And this much can be easily managed in a minute.
The essential idea in this problem is that of asymptotic estimation
x2
(see Exercise (6.53). Instead of saying that is of the order of x
x+1
x2
as x → ∞, we also say that asymptotically equals x.
x+1
Q.2 The point P is the intersection of the straight line joining the points
Q(2, 3, 5) and R(1, −1, 4) and the plane 5x − 4y − z = 1. If S is the
foot of the perpendicular drawn from the point T (2, 1, 4) to QR, then
the length of the line segment P S is
1 √ √
(A) √ (B) 2 (C) 2 (D) 2 2
2

Answer and Comments:(A). This is a straightforward problem ask-

ing for the distance between two points P and S. Neither point is given
directly. Instead, it is defined by some property. So the first task is
to determine these two points P and S. As both these points lie on
the line QR, we can write them parametrically, since we are given the
point Q and also the vector QR, viz.
−→
Q = (2, 3, 5) and QR= −~i − 4~j − ~k (1)

3
Therefore we have

P = (2 − α, 3 − 4α, 5 − α) (2)
and S = (2 − β, 3 − 4β, 5 − β) (3)

for some values of α and β. To determine α we use the fact that P lies
on the plane 5x − 4y − z = 1. A direct substitution gives

10 − 5α − 12 + 16α − 5 + α = 1 (4)

i.e. 12α − 7 = 1, which gives

2
α= (5)
3
and hence
4 1 13
P =( , , ) (6)
3 3 3
Determination of β (and hence of the point S) requires slightly more
work. We are given that the line T S is perpendicular to the line QR.
−→ −→
Hence the vectors T S and QR are perpendicular to each other. The
second vector is already known to us as −~i − 4~j − ~k. The first vector is
given by
−→
T S= −β~i + (2 − 4β)~j + (1 − β)~k (7)
−→
Taking dot product with QR, we get

β + 16β − 8 + β − 1 = 0 (8)

i.e. 18β = 9 which gives

1
β= (9)
2
and hence
3 9
S = ( , 1, ) (10)
2 2

4
 Having determined both P and S, a straightforward calculation now
gives
s
4 3 1 13 9
PS = ( − )2 + ( − 1)2 + ( − )2
3 2 3 2 2
s
1 4 1
= (+ + )
36 9 36
s
18 1
= =√ (11)
36 2

An intelligent candidate would notice that the distance P S could

also have been expressed from (2) and (3) as
√ √
P S = |α − β| 1 + 16 + 1 = |α − β|3 2 (12)

From (5) and (9), we have

2 1 1
|α − β| = | − | = (13)
3 2 6
1
Putting this into (12) we get P S = √ with a lot less calculation than
2
before. The explanation is simple. The work done in calculating P and
S from the values of α and β respectively gets undone when we take
the differences of the coordinates of P and S later.
So, there is some room to apply a little brain in this problem. Still,
it is a very routine computational problem in solid geometry. The time
saved by taking the short cut is not significant as compared to the time
spent on the main body of the problem. The real advantage of the
short cut is that it reduces the chances of numerical mistakes.
sec2 x
Z
Q.3 The integral dx equals (for some arbitrary constant
(sec x + tan x)9/2
K)

−1 1 1
 
(A) 11/2
− (sec x + tan x)2 + K
(sec x + tan x) 11 7
1 1 1
 
2
(B) − (sec x + tan x) +K
(sec x + tan x)11/2 11 7

5
 −1 1 1
 
(C) 11/2
+ (sec x + tan x)2 + K
(sec x + tan x) 11 7
1 1 1
 
2
(D) + (sec x + tan x) + K
(sec x + tan x)11/2 11 7

Answer and Comments: (C). The expression sec2 x dx in the inte-

grand makes it tempting to try the substitution t = tan x. In that case
the integral, say I, would become
dt
Z
I= √ (1)
(t + 1 + t2 )9/2
Rationalising the denominator would convert this to
Z √
I = ( t2 + 1 − t)9/2 dt (2)

As the exponent is not an integer, we cannot expand the integrand (as

a finite sum) by the binomial theorem. So we have to abandon this
line. The given integrand involves a fractional power of a sum of two
functions. In such cases, the right substitution is often to replace the
sum by a single variable. Let us try this approach. So we let

u = sec x + tan x (3)

We now have

du = (sec x tan x + sec2 x)dx = sec x udx (4)

and hence
du
sec xdx = (5)
u
We are still left with a factor sec x which we must express in terms of
u. This can be done from the identity
1 1
sec x − tan x = = (6)
sec x + tan x u
which, coupled with sec x + tan x = u gives
1 1
sec x = (u + ) (7)
2 u
6
 1 1
(we also get tan x = (u − ) but that is not needed here.) We can
2 u
now transform the given integral completely in terms of an integral of
a function of u, viz.
Z
u + u1
I= du (8)
2uu9/2
The integrand can now be expressed as a sum of various powers of u,
each of which can be integrated separately. Thus
1
Z
I = u−9/2 + u−13/2 du
2
1 u−7/2 u−11/2
" #
= − +
2 7/2 11/2
u2
" #
1 1
= − 11/2 + (9)
u 11 7

Putting back u = sec x + tan x and adding an arbitrary constant gives

the answer.
In the conventional examination this would have been a reason-
able full length question on finding antiderivatives. It would take some
thinking on the part of the candidate to come up with the right substi-
tution, viz. u = sec x + tan x. In the MCQ format all given alternatives
involve the expression (sec x + tan x)2 . And this gives an unwarranted
clue to the correct substitution. But a sneaky solution need not even
bother. One can simply differentiate each of the four alternatives, one
by one, and see whose derivative equals the given integrand. The work
involved in all these four differentiations is essentially the same, except
for the coefficients.
Had the paper-setters been a little careful, they could have given
sec x tan x instead of sec x + tan x in some of the fake alternatives. But
even then the problem is already diluted by asking it as an MCQ. It
would have been better to ask, instead, the definite integral from, say, 0
to π/4 and give the possible answers numerically. In that case the form
of the answers would not give away the right substitution so easily. But
in that case, the work involved would be far more than justified by the
credit allotted.

7
Q.4 The total number of ways in which 5 balls of different colours can be
distributed among three persons so that each person gets at least one
ball is

(A) 75 (B) 150 (C) 210 (D) 243

Answer and Comments: (B). This problem is exactly the same

as the 1981 JEE problem in Chapter 1, Comment No. 3. Even the
numbers are the same! The only change is that instead of giving the
balls to three persons, there they were to be put into three distinct
boxes so that no box was empty. As the numbers involved are small, the
best solution is by decomposing the possibilities into mutually disjoint
cases. Call the three persons as P1 , P2 , P3 . Since everybody is to get
at least one ball, the numbers of balls they get are either 2, 2, 1 or 3,
1, 1. In each case, there are three possibilities depending upon which
of the three persons is the odd man out. The number of ways to give
2 balls
  to
 P1 , 2 (of the remaining three) to P2 and the last one to P3
5 3
is 2 2 = 30. So in all there are 90 distributions of this type. As for
the second kind, the number ofdistributions where P1 gets 3 balls and
the other two one each is 53 21 = 20. So, there are 60 distributions of
this type. In all the answer is 90 + 60 = 150.
There are also more sophisticated ways of getting the answer.
One, based on Stirling numbers of the second kind, is pointed out
in Exercise (1.29). Yet another solution, based on the principle of
inclusion and exclusion (see Exercise (1.43)) can be given as follows.
Note that every distribution of the 5 (distinct) balls to 3 (distinct)
persons amounts to a function from a 5-set (i.e. a set with 5 elements)
to a 3-set. The requirement that every person is to get at least one
ball amounts to saying that the corresponding function is surjective,
i.e. onto. So the problem is equivalent to counting the number of
surjective functions from a 5-set to a 3-set. More generally, we count
the number of surjective functions from a set, say X, with m elements
to a set, say Y , with n elements using the principle of inclusion and
exclusion.
We follow the notation of Exercise (1.43). Denote the set of all
functions from X to Y by S. Clearly,
|S| = nm (1)

8
We want the number of those functions in S which are surjective. This
requirement can be paraphrased as follows. Let the (distinct) elements
of Y be y1 , y2 , . . . , yn . For i = 1, 2, . . . , n, let Ai be the set of those
functions in S whose ranges do not contain the element yi . Obviously
no such function can be onto. In fact, a function f : X −→ Y is
onto if and only if it is not in any of these Ai ’s, or equivalently it is in
n
A′i . We now apply the
\
the intersection of their complements, viz.
i=1
principle of inclusion and exclusion to get
n n
A′i = s0 − s1 + s2 − s3 + . . . + (−1)n sn = (−1)r sr
\ X
(2)
i=1 r=0

where for r = 1, 2, . . . , n, sr is the sum of the cardinalities of the inter-

sections of these subsets taken r at a time. Thus,
n
X
s1 = |Ai |
i=1
X
s2 = |Ai ∩ Aj |
i<j
X
s3 = |Ai ∩ Aj ∩ Ak |
i<j<k

and more generally,

X
sr = |Ai1 ∩ Ai2 ∩ . . . ∩ Air | (3)
1≤i1 <i2 <...<ir ≤n

Note that sn is simply |A1 ∩ A2 ∩ . . . ∩ An |. Also, s0 = |S|.

To complete the solution, we need to calculate each sr . Note that
for every 1 ≤ i ≤ n, Ai is precisely the set of all functions from X to
the set Y − {yi } which has n − 1 elements. So, Ai = (n − 1)m for every
i = 1, 2, . . . , n. This gives

s1 = n(n − 1)m (4)

Next, for i < j, Ai ∩ Aj is the set of all functions whose ranges contain
neither yi nor yj . So each such function is effectively a function from

9
 X to the set Y − {yi , yj } which is a set with n − 2 elements. As this
holds for every pair (i, j) with i < j, we get
!
n
s2 = (n − 2)m (5)
2

The same reasoning gives that for every r = 1, 2, . . . , n,

!
n
sr = (n − r)m (6)
r

Note that this is also valid for r = 0. Substituting these values into (2)
we get the number of surjective functions from X to Y as
n
!
r n
(n − r)m
X
(−1) (7)
r=0 r

There is no closed form expression for this sum. But in any special cases
it can be evaluated by finding each term which is easy to evaluate. In
the present problem m = 5 and n = 3. So the sum involves only four
terms and comes out to be

35 − 3 × 25 + 3 × 15 − 1 × 05 = 243 − 96 + 3 = 150 (8)

which is the same answer as before.

Q.5 Let P = (aij ) and Q = (bij ) be two 3 × 3 matrices with bij = 2i+j aij
for all 1 ≤ i, j ≤ 3. If the determinant of P is 2, then the determinant
of Q is

(A) 210 (B) 211 (C) 212 (D) 213

Answer and Comments: (D). An honest solution would be to begin

by writing Q out as
 
4a11 8a12 16a13
Q =  8a21 16a22 32a23 

 (1)
16a31 32a32 64a33

10
 Taking out the factors 4, 8 and 16 from the first, the second and the
third row respectively, we get

4a11 8a12 16a13

|Q| = 8a21 16a22 32a23
16a31 32a32 64a33
a11 2a12 4a13
= 4 × 8 × 16 × a21 2a22 4a23 (2)
a31 2a32 4a33

We now take out the factors 2 and 4 from the second and the third
column respectively. What is left is simply |P |. So,

|Q| = 4 × 8 × 16 × 2 × 4 × |P |
= 22+3+4+1+2+1 = 213 (3)

This is a good problem with a reasonably straightforward solution.

But there is a sneak solution which makes a mockery of even this simple
solution. Note that it is already implicit in the problem that the ratio
of the determinants of Q and P is a constant. So, we shall get a solution
as soon as this constant is determined. And for this purpose it suffices
to take P to be any matrix with determinant 2. As the simplest choice
take P to be a diagonal matrix with entries 2, 1 and 1. Then Q is
also a diagonal matrix with entries 8, 16 and 64. So its determinant is
213 . In a conventional examination, this solution would get little credit
since it is applicable only in a special case. But when the problem is
posed as a multiple choice question, there is no way to know how the
candidate got the answer. So a candidate who comes up with a sneak
solution not only gets full credit but also some bonus in terms of the
time saved which is a precious commodity.

Q.6 Let z be a complex number such that the imaginary part of z is non-zero
and a = z 2 + z + 1 is real. Then a cannot take the value
1 1 3
(A) −1 (B) (C) (D)
3 2 4

Answer and Comments: (D). Before commenting on the problem

itself, its wording calls for a mild rebuttal. Although the phrase ‘such

11
that’ is used commonly (and often inevitably) in precisely stating some
complicated definitions (a well known instance being the definition of
a limit), for a layman and at the JEE level it often appears clumsy
and should be avoided as far as possible. Thus, the present problem
could have been worded as ‘Let z be a complex number with a non-
zero imaginary part. Let a = z 2 + z + 1. Then ......’ which is a simpler
diction.
Now, coming to the problem itself, assume z = x + iy where x, y
are real. We are given y 6= 0. A direct calculation gives

a = (x + iy)2 + (x + iy) + 1
= (x2 − y 2 + x + 1) + i(2xy + y) (1)

As we are given that a is real, we get 2xy + y = 0. As y 6= 0, we get

1
x = − . With this substitution in (1) we have
2
3
a= − y2 (2)
4
3 3
As y is real and non-zero, y 2 > 0. Hence a < . So, a 6= . The other
4 4
3 1
values given are all less than and can be assumed by a with x = −
4 2
and a suitable choice of y.
The problem can also be tackled by looking at z 2 + z + 1 = a as a
quadratic in z. The two possible solutions are
√
−1 ± 4a − 3
z= (3)
2
3
If a = , then z would be purely real contradicting the hypothesis.
4
This solution is shorter but not so straightforward.
Since the number 1 is real, the reality of z 2 + z + 1 is equivalent
to that of z 2 + z. It is not clear what purpose is served by giving
that z 2 + z + 1 is real instead of giving that z 2 + z is real, which is
more direct. one possibility is that z 2 + z + 1 can be expressed as a
z3 − 1
ratio, viz. . The ratio of two non-zero complex numbers, say z1
z−1

12
 and z2 , is real if and only if their arguments differ by a multiple of π.
Geometrically this means that the points 0, z1 and z2 are collinear. In
the present case this amounts to saying that the points z 3 , 1andz are
collinear. But that does not apparently lead to any elegant solution of
the problem. Anyway, the problem is so simple as it stands that it is
only of an academic interest to see if there is any elegant solution.
Q.7 The locus of the mid-point of the chord of contact of tangents drawn
from points lying on the straight line 4x−5y = 20 to the circle x2 +y 2 =
9 is

(A) 20(x2 + y 2 ) − 36x + 45y = 0 (B) 20(x2 + y 2) + 36x − 45y = 0

(C) 36(x2 + y 2) − 20x + 45y = 0 (AD 36(x2 + y 2 ) + 20x − 45y = 0

Answer and Comments: (A). Let P = (x1 , y1 ) be a point on the

given line, say L, and suppose the tangents from to P to the given
circle, say C, touch it at the points A and B. Let Q = (h, k) be the
midpoint of the chord AB. We have to find the locus of the point Q.
For the formula lovers, there are two formulas which are almost tailor
made for this problem. One is the equation of the chord of contact of
the point P w.r.t. the circle C. Popularly, it is remembered by T = 0.
In the present case, since

C = x2 + y 2 = 9 (1)

and P = (x1 , y1), the equation of the chord AB comes out to be

xx1 + yy1 = 9 (2)

The second formula, which is less frequently used, is for the equation
of a chord in terms of its middle point M. In a compact form it is
T = S1 . In the present case, since Q = (h, k), this becomes

hx + ky = h2 + k 2 (3)

As (2) and (3) are equations of the same line AB, we get
x1 y1 9
= = 2 (4)
h k h + k2

13
These are two equations. In addition we are given that (x1 , y1 ) lies on
the line L. So,
4x1 − 5y1 = 20 (5)
The locus of (h, k) can be obtained by eliminating x0 and y0 in these
three equations. The most straightforward way to do this is by getting
the values of x1 and y1 from (4) and substitute them in (5). Thus we
get
36h 45k
− 2 = 20 (6)
h2 +k 2 h + k2
i.e.
20(h2 + k 2 ) − 36h + 45k = 0 (7)
To get the locus of Q, replace its current coordinates h and k by x and
y respectively. This gives (A) as the answer.
Formulas analogous to (2) and (3) are valid for any conic. So the
same method would have worked if instead of the circle C we were given
some other conic in the plane. In the present problem as the conic is
a circle which is heavily studied in pure geometry, instead of using the
formulas (2) and (3), a solution based on some standard properties of
tangents to a circle from a point outside it can be given as follows.
We follow the notation in the solution above. Then it can be seen by
symmetry that the points P , Q and O (the centre of the circle) are
collinear and moreover that OP.OQ = 9, the radius square. (An easy
way to see this is the similarity of the triangles OQB and OBP .)
y
P (x1 , y1 )

A
Q (h,k)

O B
x

14
 With this approach, we have, by collinearity of O, Q and B,

x1 = λh and y1 = λk (8)

for some λ. Also the equality OQ.OP = 9 gives

√ q
h2 + k 2 x21 + y12 = 9 (9)

Substituting the values from (8) into (9) we get

λ(h2 + k 2 ) = 9 (10)

Hence we have
9h 9k
x1 = and y 1 = (11)
h2 + k 2 h2 + k 2
which is exactly the same as (4). The rest of the work is the same as
before.
The problem is straightforward once you know which formulas or
properties (of circles) to apply. But again, the work involved is more
than can be justified on the basis of the credit allotted.
(
x2 cos πx , x 6= 0
Q.8 Let f (x) = for x ∈ IR. Then f is
0, x=0

(A) differentiable both at x = 0 and x = 2

(B) differentiable at x = 0 but not differentiable at x = 2
(C) not differentiable at x = 0 but differentiable at x = 2
(D) differentiable neither at x = 0 nor at x = 2

Answer and Comments: (B). Here we have two separate problems

about checking differentiability at the given points. A detailed analysis
would be to consider the right as well as the left handed derivatives of
f (x) at the two points given. But instead, we give an argument which
is based on the known differentiability of two very commonly studied
functions, viz. g(x) = x2 sin x1 for x 6= 0 (with g(0) = 0) and h(x) = |x|,
at the point 0. The first is differentiable at 0 while the second one is

15
not. We count on this familiarity and tackle the problem using the
same reasoning.
The reason that the function g is differentiable at 0 is that it
g(x) − g(0)
is of the form x2 times a bounded function. So the ratio
x−0
becomes x times a bounded factor and hence tends to 0. Exactly the
same reason applies to show that f is differentiable at 0. As for its
differentiability at 2, note that the first factor x2 of f (x) is differentiable
at 2 and non-zero at 2. As f is continuous at all x 6= 0 and hence in a
neighbourhood of x = 2, differentiability of f at 2 depends entirely on
the differentiability of its second factor, viz. | cos πx |. (To see this more
f (x)
formally, write | cos πx| as 2 in a neighbourhood of x = 2 and apply
x
the quotient rule for derivatives.)
Thus the problem is reduced to checking the differentiability of the
function, say |k(x)| (where k(x) = cos πx ) at x = 2. Here the crucial
point is that the absolute value function is differentiable everywhere
except where it takes the value 0. In the present case, in a small
neighbourhood of x = 2, the function k(x) vanishes only at x = 2.
However, from this we must not hastily conclude that |k(x)| is not
differentiable at x = 2. What matters is not just that |k(x)| → 0 as
x → 2 but how rapidly. If it tends to 0 more rapidly than |x − 2|
then |k(x)| would indeed be differentiable at x = 2 with a vanishing
derivative. (For example, the functions |(x − 2)3 | or | sin3/2 (x − 2)| are
differentiable at x = 2.) On the other hand, if |k(x)| tends to 0 at a
rate slower than or comparable to that of |x − 2|, then |k(x)| is not
differentiable at x = 2. (In the former case, the right and left handed
derivatives don’t exist while in the latter they
q exist but are unequal.
The two cases are illustrated, for example, by |x − 2| and | sin(x − 2)|
respectively.)
So, now we have to see how fast cos πx tends to 0 as x → 2. For this
cos πx
we consider and simplify the ratio .
x−2

 
π π
cos π sin 2
− x
x
=
x−2 x−2

16
  
π(x−2)
sin 2x
= (1)
x−2
Now we use the well-known fact that as θ → 0, sin θ tends to 0 at
a rate comparable to that of θ. So, we replace the numerator in the
π(x − 2)
expression above by . So, we get that
2x
cos πx π(x − 2) π
lim = lim = (2)
x→2 x − 2 x→2 2x(x − 2) 4
which shows that as x → 2, | cos πx | tends to 0 at a rate comparable
to that of |x − 2|. We are now justified in saying that cos πx and hence
f (x) is not differentiable at x = 2.
This is a good problem in which brute force computations can
be avoided with some fine thinking. Unfortunately, a candidate who
argues that | cos πx | is not differentiable at x = 2 on the superficial
ground that the absolute value function is not differentiable at 0 also
gets full credit undeservedly.
Q.9 The function f : [0, 3] −→ [1, 29] , defined by f (x) = 2x3 −15x2 +36x+1

(A) one-one and onto (B) onto but not one-one

(C) one-one but not onto (D) neither one-one nor onto

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

behaviour of a polynomial function on an interval. The only way a con-
tinuous function defined on an interval, say [a, b], can be one-one is that
it is either strictly increasing or decreasing. If the function is further
differentiable, then the sign of its derivative provides a handy crite-
rion. (So handy, if fact, that many people wrongly believe that a posi-
tive derivative throughout an interval is the very meaning of its being
strictly increasing, forgetting that the concept of increasing/decreasing
is based purely on inequalities and has nothing to do with derivatives!)
Returning to the problem, a straight calculation gives
f ′ (x) = 6x2 − 30x + 36
= 6(x2 − 5x + 6)
= 6(x − 2)(x − 3) (1)

17
 Both the roots of f ′ (x) are simple and lie in the interval [0, 3]. So, f ′
is positive on [0, 2) and negative on (2, 3]. Hence f is neither strictly
increasing nor strictly decreasing on [0, 3] and hence not one-one.
Next, to ascertain which values f takes, too the derivatives help.
As f is strictly increasing on [0, 2] the image of [0, 2] is the interval
[f (0), f (2)] i.e. [1, 29]. This alone is enough to conclude that f is onto.
(Just for the sake of completeness, f is strictly decreasing on [2, 3] and
hence maps it onto [f (3), f (2)] = [28, 29]. As this interval is already
included in [0, 29], it does not add to the range. It remains [0, 29].)
x2 y 2
Q.10 The ellipse E1 : + = 1 is inscribed in a rectangle R whose sides
9 4
are parallel to the coordinate axes. Another ellipse E2 , passing through
the point (0, 4) circumscribes the rectangle R. The eccentricity of the
ellipse E2 is
√ √
2 3 1 3
(A) (B) (C) (D)
2 2 2 4

Answer and Comments: (C). A straightforward problem once you

understand the vocabulary involved, specifically, what is meant by ‘in-
scribe’, ‘circumscribe’ and ‘eccentricity’. (Incidentally, these three are
highly technical words and even with an otherwise good command over
English, a candidate is not likely to have come across them unless he
has studied geometry in the English medium. This is no reason to shed
tears if the candidate is weak in geometry. But if he is strong, then
it means that he is being penalised merely for not having studied in
the English medium. Availability of the Hindi version of the question
paper alleviates the problem to some extent but is of little help to a
candidate who has studied in Kannada or Tamil. One solution might
be to make dictionaries of technical terms available to the candidates,
similarly to the log tables).
Now coming to the problem itself, the extreme ties of the major and
the minor axes of the ellipse E1 are (±3, 0) and (0, ±2) respectively.
Hence the rectangle R has its vertices’s at (±3, ±2). The second ellipse
is given to pass these four points. However, this does not determine the
ellipse E2 uniquely. You need five conditions to determine an ellipse
(and more generally, any conic). This happens because the general

18
second degree equation
Ax2 + 2Hy + By 2 + 2G + 2F Y + C = 0 (1)
has five independent unknowns. (Superficially there are six unknowns,
A, B, C, D, E, F , but it is only their relative proportions that matter).
If (1) is to represent an ellipse (and indeed any conic) passing through
the four points (±3, ±2) we must have
9A + 12H + 4B + 6G + 4F +C = 0 (2)
9A − 12H + 4B + 6G − 4F +C = 0 (3)
9A − 12H + 4B − 6G + 4F +C = 0 (4)
and 9A + 12H + 4B − 6G − 4F +C = 0 (5)
Subtracting (3), (4) and (5) from (2), we get three equations in three
unknowns, viz.,
24H + 8F = 0 (6)
24H + 12G = 0 (7)
8F + 12G = 0 (8)
which yield H = F = G = 0. We are now justified in assuming that
the ellipse E2 has its axes parallel to the coordinate axes. Hence we
can represent E2 by an equation of the form
x2 y 2
+ 2 =1 (9)
a2 b
where we need two equations to determine the parameters a and b. One
of them is already given by any of the four points (±3, ±2).
9 4
2
+ 2 =1 (10)
a b
The other equation comes from the data that E2 passes through the
point (0, 4). In fact, this directly gives b = 4. Putting this into (10)
and solving, we have a2 = 12. Note that for this ellipse, the major axis
is along the y-axis rather than on the x-axis, as is usually the case. So
the usual formula for the eccentricity has to be modified by switching
a and b. That is, we have
a2 = b2 (1 − e2 ) (11)

19
 Substituting the values of a and b we get
12 3
1 − e2 = = (12)
16 4
1 1
which gives e2 = and hence e = .
4 2

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. All correct
choices get 4 marks, otherwise, none.
The problem is easy once you get that the ellipse E2 has its axes parallel
to the coordinate axes. This is not given in the statement of the problem
and deriving this from the rest of the data takes some work as shown above.
The trouble is that nearly in all problems at the JEW level, ellipses are in
the standard form. As a result, many students will simply assume that (10)
is the equation of E2 . Such students will get an unfair advantage over the
scrupulous ones who derive it with some work (which is simple, but takes
some time nevertheless). Yet another sad casualty of the competitive MC
based tests.

SECTION II

Multiple Correct Answer(s) Type

This section contains five questions with four answers to each, out of
which one or more may be correct. 4 points for identifying all correct
answers, 0 otherwise.

Q.11 A ship is fitted with three engines E1 , E2 and E3 . The engines function
1 1 1
independently of each other with respective probabilities , and .
4 2 2
Let X denote the event that the ship is operational and let X1 , X2 and
X3 denote, respectively, the events that the engines E1 , E2 and E3 are
functioning. Which of the following is (are) true?

20
 3
(A) P [X1c |X] =
16
7
(B) P [exactly two engines of the ship are functioning |X] =
8
5
(C) P [X|X2 ] =
16
7
(D) P [X|X1 ] =
16

Answer and Comments: (B, D). A question of calculating the con-

ditional probabilities. The ship is operational if and only if at least
two of E1 , E2 , E3 are operational. This resolves into four mutually dis-
joint events : either all are operating or exactly one of the three is not
functioning and the other two are. As a result,
P (X) = P (X1 X2 X3 ) + P (X1C Z2 X3 ) + P (X1 X2c X3 ) + P (X1 X2 X3c ) (1)
We are given the values of P (Xi ) for i = 1, 2, 3. Further they are given
to be mutually independent. So, (1) gives
111 111 131 113
P (X) = + + +
244 244 244 244
1 1 3 3
= + + +
32 32 32 32
1
= (2)
4
In all the statements except (B), the desired probability is already
expressed in terms of symbols. In (B), it is expressed in words. It is
the conditional probability that exactly two of the events X1 , X2 and
X3 occur, given that X occurs. Call the former event ay Y . Then Y is
the disjunction of the three mutually disjoint events X1c X2 X3 , X1 X2c X3
and X1 X2 X3c , already considered in computing (1). So, taking the last
three terms in (1), we get
7
P (Y ) = (3)
32
Moreover, Y is a sub-event of X. So Y ∩ X is the same as Y . So
P (Y ) 7/32 7
P (Y |X) = = = (4)
P (X) 1/4 8

21
Hence statement (B) is true. The calculations in the remaining three
statements are more direct. In (A),

P (X1c |X) = P (X1c ∩ X)/P (X) = P (X1c X2 X3 )/P (X)

111
1
= 2 41 4 = (5)
4
8

Hence (A) is incorrect. In (C), X ∩ X2 equals (X1 X2 X3 ) ∪ (X1c X2 X3 ) ∪

(X1 X2 X3c ). Hence

P (X ∩ X2 )
P (X|X2) =
P (X2 )
P (X1X2 X3 ) + P (X1c X2 X3 )) + P (X1 X2 X3c )
=
P (X2 )
111 111 113
+ +
= 2 4 4 2 41 4 2 4 4
4
5/32 5
= = (6)
1/4 8

So the statement (C) is also incorrect. Finally, for (D), the calculations
are similar to that in (C) except that we interchange X1 and X2 . So,
analogously to (6), we have

P (X ∩ X1 )
P (X|X1) =
P (X1 )
P (X1X2 X3 ) + P (X1 X2c X3 )) + P (X1 X2 X3c )
=
P (X2 )
111 131 113
+ +
= 2 4 4 2 41 4 2 4 4
2
7/32 7
= = (7)
1/2 12

Therefore the statement (D) is correct.

This is a simple but highly repetitious problem. The statement (B)
by itself (without the answer) could have been asked as an MCQ with
only one correct answer. (B) tests the ability to correctly translate
a verbally described event into a mathematical formula as well as the

22
 knowledge of the formula for conditional probability. Nothing more and
a lot less is tested in the other three statements. They only waste a
candidate’s precious time and increase the chances of numerical errors.
2
Q.12 If S is the area of the region enclosed by y = e−x , y = 0, x = 0 and
x = 1, then

1 1
(A) S ≥ (B) S ≥ 1 −
e ! e !
1 1 1 1 1
(C) S ≤ 1+ √ (D) S ≤ √ + √ 1 − √
4 e 2 e 2

Answer and Comments: (A, B, D). Clearly,

Z 1 2
S= e−x dx (1)
0
If we could evaluate S exactly, then we can answer all alternatives by
2
direct comparisons. The trouble here is that the integrand e−x does
not have an andtiderivative in an explicit form and so the fundamental
theorem of calculus cannot be applied. Therefore, the present problem
is not about the evaluation of a definite integral. The best we can do
is to find some upper and lower bounds on S. Parts (A) and (B) of
the problem give possible lower bounds for S while (C) and (D) give
possible upper bounds and our job is to find which of these are correct.
Z b
For any definite integral I = f (x)dx the most immediate up-
a
per and lower bounds (obtained from the very definition of a definite
integral) are the upper and lower Riemann sums of f (x) for various
partitions, say,
P = (x0 , x1 , . . . , xi , . . . xn ) (2)
(where a = x0 < x1 < x2 < . . . < xn = b). In the present problem,
the integrand is strictly decreasing on [0, 1]. So, the upper and lower
Riemann sums for a partition P like (2) are
n
X
U(P ; f ) = f (xi−1 )(xi − xi−1 ) (3)
i=1
Xn
L(P ; f ) = f (xi )(xi − xi−1 ) (4)
i=1

23
 2
We now take f (x) = e−x and [a, b] as [0, 1]. Let us first see which
of the given four numbers occur either as lower or as upper Riemann
sums for I. Trivially, if we take the partition P to be the one whose
only nodes are 0 and 1, then

U(P ; f ) = f (0) × 1 = 1 (5)

1
and L(P : f ) = f (1) × 1 = (6)
e
Thus we see that the the expression in (A) is a lower bound for S.
Hence (A) is true.
To see which of the other three expressions arise as lower or upper
bounds for I, we need to recognise some points in [0, 1] at which f
assumes some values which equal some of the numbers occurring in
1
these expressions. We already know (and used) that 1 and equal,
e
1
respectively f (0) and f (1). A less obvious functional value is √ . Its
e
1 1
natural logarithm is −1/2 which equals −( √12 )2 . Hence f ( √ ) = √ .
2 e
1
Let us now consider the partition P = (0, √ , 1) with three nodes. For
2
this partition we have
1 1 1 1 1 1
U(P ; f ) = f (0)( √ − 0) + f ( √ )(1 − √ ) = √ + √ (1 − √ ) (7)
2 2 2 2 e 2
1 1 1 1 1 1
L(P ; f ) = f ( √ )( √ − 0) + f (1)(1 − √ ) = √ + (1 − √ ) (8)
2 2 2 2e e 2
The R.H.S. of (7) is precisely the expression in (D). So we see that
(D) is correct. (As it turns out, calculating (8) was essentially a waste.
Unfortunately there is no easy way to fortell this.)
The expressions in (B) and (C) cannot be recognised as upper
or lower Riemann sums in an obvious way. So we try some other
approaches. One of them is based on the simple fact that integrals
preserve inequalities of functions. Specifically, if we have two functions,
say f1 (x) and f2 (x) such that

f1 (x) ≤ f2 (x) (9)

24
for all x ∈ [a, b]. Then
Z b Z b
f1 (x) dx ≤ f2 (x) dx (10)
a a

Actually, the inequalities based on upper and lower Riemann sums can
be viewed as special cases of this where one of the functions is the given
integrand and the other function is a step function whose values are the
maxima/minima of the integrand on the subintervals of the partition.
But we have already exhausted these possibilities. So we try some
other functions for comparison. In the present problem, we take f1 (x)
as f (x). We now look for a function f2 (x) for which the inequality (9)
(or its opposite) will be true throughout [0, 1]. Since the exponential
function is monotonically increasing (i.e. preserves inequalities), we
can take f2 (x) of the form eg(x) where the exponent g(x) is comparable
to −x2 throughout [0, 1]. One such choice is to take g(x) as −x. (Of
course there are many others. But we are guided by the requirement
Z1
that eg(x) dx should coincide with the expression in (B) or (C).) With
0
this choice, we get

−x2 ≥ −x (11)

for all x ∈ [0, 1] and hence

Z 1
S≥ e−x dx (12)
0

This integral, unlike the original integral, can be easily evaluated in a

1 1
closed form. It equals −e−x = 1 − which is precisely the expression
0 e
in (B). So we see that (B) is also true.
Finally, we check if (C) is correct. The method we are applying so
far requires us to find a suitable function h(x) which is comparable to
2
e−x throughout the interval [0, 1] and whose integral over [0, 1] equals
the given expression. This approach did work in (B) by taking h(x)
as e−x (and also indirectly in (A) and (D) where we took h(x) to be
a suitable step function). Unfortunately, in (C), there is no obvious
h(x) which will fit the prescription. So here a more delicate analysis is
needed.

25
 We can take some clue from
1 y
the sum 1 + √ . We already
e
1
recognised 1 as f (0) and √
e C
1
as f ( √ ). Therefore we get
2 !
1 1
that the expression 1+ √ S P
2 e 1

is, geometrically, the middle S2

x
O A B
height of the trapezium OAP C
shown in the accompanying
diagram , where O, A, P and
C are respectively, the points
1 1 1
(0, 0), ( √ , 0), ( √ , √ ) and
2 2 e
(0, 1).
Therefore, the area, say ∆, of this trapezium is,
! ! !
1 1 1 1 1
∆ = 1+ √ √ −0 = √ 1+ √ (13)
2 e 2 2 2 e
Let us now compare this area with the area, say S1 , under the portion
2 1
of the curve y = e−x between x = 0 and x = √ . That is,
2
√
Z 1/ 2 2
S1 = e−x dx (14)
0
It is not easy to evaluate S1 in a closed form, the difficulty being ex-
actly the same as that in evaluating S, viz. that the integrand has no
antiderivative that can be expressed in a closed form. But our interest
is not so much in evaluating S1 as in comparing it with ∆. Here we
2
crucially need the concept of concavity of a function. Since f (x) = e−x
a direct calculation gives
2
f ′ (x) = −2xe−x (15)
′′ 2 −x2
and f (x) = (4x − 2)e (16)
for all x. The exponential factor is always positive while the other
1 1
factor (4x2 − 2) is negative for 0 ≤ x < √ and positive for x > √ .
2 2

26
 1
So we get that f is concave downward on [0, √ ]. We also get that it is
2
1
concave upwards on [ √ , ∞) and hence that the point P on the graph
2
−x2
y =e is a point of inflection, where the tangent crosses the curve
as shown in the figure. But that is not relevant here. What matters
1
is only the first part, viz., that f (x) is concave downward on [0, √ ].
2
Geometrically, this means that the curve always lies above the chord.
Since ∆ is the area below this chord, while S1 is the area below the
curve we have

S1 > ∆ (17)
2
Further as the integrand e−x is positive throughout, we also have

S > S1 (18)
√
Further since 2 2 < 4 we also have
!
1 1
∆> 1+ √ (19)
4 e
Combining the last three inequalities, we finally have
!
1 1
S> 1+ √ (20)
4 e
which shows that the statement (C) is false.
This is an excellent problem on the estimation of definite integrals.
It takes certain! perceptivity on the part of the candidate to recognise
1 1 √
√ as f √ and thereby to realise that the point 1/ 2 has a special
e 2
role in the problem. A further realisation that it is the point where f (x)
changes its concavity gives the necessary hint for (C). Unfortunately,
the work demanded for various alternatives is of a different kind. The
work for (C) is unusual since it is based on the concept of concavity
downwards. The time needed far exceeds the time allowed. Note that
the inequality (17) we have proved using concavity is much stronger
than what is needed to disprove (C). It is possible that the paper-
setters had some simpler argument in mind to disprove (C).

27
 x2 y2
Q.13 Tangents are drawn to the hyperbola − = 1, parallel to the
9 4
straight line 2x − y = 1. The points of contacts of the tangents on the
hyperbola are
! !
9 1 9 1
(A) √ ,√ (B) − √ , − √
2√ 2 √2 2√ 2 √ 2
(C) (3 3, −2 2) (D) (−3 3, 2 2)

Answer and Comments: (A, B). This is a straightforward problem

about finding the point of contact of a tangent to a conic given its
slope. But apparently the paper-setters thought it too straightforward
and so they have given it a twist by not specifying the slope directly,
but, instead saying that the tangent is parallel to a given line. This line
has no other role in the whole problem. It would be naive to suppose
that a candidate who gets stuck by not being able to find the slope of
the tangent from this piece of data would have been able to solve the
problem correctly had the slope be given directly as 2. So no testing
purpose is served by such corn husks. In a conventional examination,
perhaps some partial credit could be allotted for merely finding the
slope of the tangent. But in an MCQ, that is not possible either.
Now coming to the problem itself, for the formula lovers, there is a
formula for the equation of a tangent to a conic in terms of its slope.
Knowing the equation, one can find its point of contact by solving
simultaneously with the equation of the conic. But a better method is
available if we keep in mind that our interest is more in the point of
contact than with the tangent itself. Suppose (x1 , y1) is a point on the
hyperbola. Then the equation to the tangent at (x1 , y1 ) is
xx1 yy1
− =1 (1)
9 4
If this is to be parallel to the line 2x − y = 1, then its slope must be 2.
This gives an equation in x1 , y1 , viz.
x1 /9
=2 (2)
y1 /4
which upon simplification becomes
9
x1 = y1 (3)
2
28
Putting this into the equation of the hyperbola gives a quadratic in y1 ,
viz.,
9y12 y12
− =1 (4)
4 4
1 1
i.e. y12 = which gives y1 = ± √ . Putting these values in (3), the
2 2
two points of contact are those on (A) and (B).
This solution is so simple that it is relatively pointless to see if
it can be improved. But there is a way to improve. Our interest is
in the slope rather than in the entire equation of the tangent. If we
take parametric equations, there is a handy formula for the slope of the
tangent at a point using the chain rule. So we let

x = 3 sec θ, y = 2 tan θ (5)

for θ ∈ [0, 2π]. Differentiating,

dx dy
= 3 sec θ tan θ, = 2 sec2 θ (6)
dθ dθ
So, by the chain rule,
dy dy/dθ
=
dx dx/dθ
2 sec θ
=
3 tan θ
2
= (7)
3 sin θ
Equating this with 2 we get
1
sin θ = (8)
3
√
2 2 1
which then gives cos θ = ± and further tan θ = ± √ and sec θ =
3 2 2
3
± √ . Putting these into (5) gives the desired points of contact. (Ac-
2 2
tually, it suffices to find only one of the points of contact, since points
of contact of parallel tangents are always symmetric w.r.t. the centre.)

29
Q.14 If y(x) satisfies the differential equation y ′ − y tan x = 2x sec x and
y(0) = 0, then

π π2 π π2
   
(A) y = √ (B) y ′( =
4 8 2 4 18
π π2 π 4π 2π 2
   
′
(C) y = (D) y ( = + √
3 9 3 3 3 3

Answer and Comments: (A, D). A straightforward problem on solv-

ing a differential equation. The given equation is linear with integrating
factor
R
tan xdx
e −
= e− ln sec x = cos x (1)

The formula lovers can now also use the ready-made formula for the
solution. But it is often a good idea to multiply the equation through-
out by the integrating factor and cast the L.H.S. as the derivative of
some function. Indeed this is the very purpose of integrating factors.
So, multiplying both the sides by cos x, the given D.E. becomes

y ′ cos x − y sin x = 2x (2)

i.e.
d
(y cos x) = 2x (3)
dx
whose general solution is

y cos x = x2 + c (4)

where c is an arbitrary constant. The initial condition y(0) = 0 deter-

mines c as 0. Hence we have

y = x2 sec x (5)

and hence

y ′(x) = 2x sec x + x2 sec x tan x (6)

30
 π π
To complete the solution we have to substitute x = and x = in (5)
4 3
π π √
and (6). Since tan = 1 while sec = 2, a direct calculation gives
4 4
π π2 √ π2
 
y = 2= √ (7)
4 14 8 2
π π√ π2 √
 
y′ = 2+ 2 (8)
4 2 16
π
 
which shows that (A) is true while (B) is false. Similarly, from tan =
3
√ 
π

tan 60◦ = 3 and sec = 2 we see, after substitution, that (C) is
3
false while (D) is true.
A very routine problem about differential equations with a purely
clerical trigonometrical appendage.

Q.15 Let θ, φ ∈ [0, 2π] be such that

!
2 θ θ
2 cos θ(1 − sin φ) = sin θ tan + cot cos φ − 1,
2 2
√
3
tan(2π − θ) > 0 and −1 < sin θ < − . Then φ cannot satisfy
2

π π 4π
(A) 0 < φ < (B) <φ<
2 2 3
4π 3π 3π
(C) <φ< (D) < φ < 2π
3 2 2

Answer and Comments: (A, C, D). Superficially, this is a problem

about trigonometric equations. But in reality it is not so. There are
two variables here θ and φ. But only one equation in them. So, the
data is insufficient to determine θ and φ. All we can do is to use this
equation to express the variable φ in terms of θ. We are also given
(indirectly) the interval(s) over which θ varies and our job is to find
the set, say S, over which φ varies. Once this set S is identified, the
problem asks you to determine which of the given four intervals is (are)
disjoint from S. (The language used in framing the last part of the

31
question is likely to be confusing to some candidates. Generally, the
paper-setters are restrained from asking questions where the assertion
is in the negative.)
Once again, the paper-setters have given a twist to the data. A
straight-forward simplification gives that for any angle θ,
sin θ cos θ
tan θ + cot θ = +
cos θ sin θ
1
=
sin θ cos θ
2
= (1)
sin 2θ
Using this identity (with θ replaced by θ/2), the given equation becomes

2 cos θ(1 − sin φ) = 2 sin θ cos φ − 1 (2)

which upon further simplification gives

2 sin(θ + φ) = 2 cos θ + 1 (3)

We are also given certain restrictions on the possible values θ can take.
First θ ∈ [0, 2π]. Secondly, tan(2π − θ) > 0 which is a twisted way of
π 3π
 
saying that tan θ < 0. So, θ is further restricted to ,π ∪ , 2π .
√ 2 2
3
Finally we are also given that −1 < sin θ < − which means θ ∈
2
4π 5π 3π
( , ) − { } and further restricts θ to
3 3 2
3π 5π
 
θ∈ , (4)
2 3
From this and (3) we have to determine what possible values φ can
take. One way to do this would be to solve (3) for φ to get
1
φ = nπ + (−1)n sin−1 (cos θ + ) − θ (5)
2

where n is an integer. We shall then have to consider various values of

n for which the R.H.S. has at least some values between 0 and 2π and

32
in each case determine which possible values φ can take. Moreover, the
analysis would be different for even n than for odd n. This would be
complicated. So we try a simpler approach. We recast (3) as

1 + cos θ
sin(θ + φ) = (6)
2
1
Because of (4) the R.H.S. lies between and 1. Further since both θ, φ
2
lie in [0, 2π], θ + φ ∈ [0, 4π]. Therefore
π 5π π 5π
   
θ+φ∈ , ∪ 2π + , 2π + (7)
6 6 6 6
Thus, we have two cases:
π 5π
< φ+θ < (8)
6 6
13π 17π
or < φ+θ < (9)
6 6
We add the inequality
5π 3π
− < −θ < − (10)
3 2
to each of these two inequalities to get
3π 2π
− < φ < − (11)
2 3
π 4π
or < φ < (12)
2 3

The first possibility is ruled out because φ ≥ 0. The second possibility
π 2π
means that φ must lie in the interval , . This is precisely the
2 3
interval in (B). As the intervals in the remaining three possibilities are
all disjoint from this interval we see that φ can never lie in any of them.
We were somewhat lucky that we got the answer relatively easily.
From the data that θ and φ both lie in [0, 2π], we concluded that
θ + φ lies in [0, 4π]. Implicitly this was obtained by adding the two
inequalities 0 ≤ θ ≤ 2π and 0 ≤ φ ≤ 2π. This is a true but a

33
 crude estimate. When θ and φ are independent of each other, this is
the best estimate we can give. But when two quantities are related
to each other, sharper estimates are often possible for their sums. As
an extreme example, both sin2 x and cos2 x vary over [0, 1]. But their
sum does not vary all over [0, 2]. In fact, it is a constant. In the
present problem, φ is related to θ through the equation (5). So the
variation of θ + φ may be much smaller than 0 to 4π. There was also
some extravagance in concluding (11) from (8) and (10) (and similarly,
in deriving (12) from (9) and (10)). Accordingly, the set, say S, of
π 4π
possible values of φ may be only a subset of the interval , . The
2 3
intervals in (A), (C) and (D) are all disjoint from this interval and
hence also from S. But if they were bigger then it is possible that they
are not disjoint from this interval but still disjoint from the subset S.
In that case our approach would fail. To identify the set S exactly,
we need to follow the other approach of a detailed analysis of (5) for
various values of the integer n. In that case the problem would have
been far more challenging and complicated. As it stands, the problem
is simple for the naive candidate. A scrupulous candidate who goes for
analysing (5) is penalised in terms of the time he spends.

SECTION III
Integer Answer Type

This section contains five questions. Each has a single digit correct an-
swer. Correct answer gets 4 points, incorrect one none.

Q.16 Let p(x) be a real polynomial of least degree which has a local maximum
at x = 1 and a local minimum at x = 3. If p(1) = 6 and p(3) = 2, then
p′ (0) is ....... .

Answer and Comments: 9. The data implies that p′ (x) has 1 and 3
among its roots. So, its degree is at least two. Since p(x) has the least
possible degree among polynomials which satisfy this requirement, so
does p′ (x). Hence p′ (x) is of the form

p′ (x) = α(x − 1)(x − 2) (1)

34
for some (real) constant α. Expanding and integrating,

x3
p(x) = α( − 2x2 + 3x) + β (2)
3
where β is some other real constant. The values of α and β can be
found from the given conditions, viz, p(1) = 6 and p(3) = 2 which
imply
1
α( − 2 + 3) + β = 6 (3)
3
and α(9 − 18 + 9) + β = 2 (4)

respectively. The second equation immediately gives β = 2. Putting

this into (3) we get
4
α =6−2=4 (5)
3
which implies α = 3. This gives

p′ (x) = 3(x2 − 4x + 3) (6)

whence p′ (0) = 9.
A very straightforward problem about determining a polynomial,
being given its points of local maxima and local minima. The compu-
tations involved are reasonable and not prone to numerical errors. The
values of p(x) at any two distinct points would serve to determine α
and β uniquely. No particular advantage is gained by specifying the
values of p(x) at the points where it has a local maximum or a local
minimum. However, with these values, the same data could have been
given a little more subtly, by specifying that the graph of p(x) has a
local maximum at the point (1, 6) and a local minimum at (3, 2).
The problem could have been made a little more interesting by
specifying that p(x) is a cubic polynomial which has a point of inflection
at x = 2 and a local minimum at x = 3. In that case a candidate would
first have to recognise that the local maximum of p(x) must be at x = 1
(because, for a cubic, the point of inflection is the arithmetic mean of
the point of local maximum and the point of local minimum). The rest
of the solution would remain the same.

35
Q.17 Let S be the focus of the parabola y 2 = 8x and let P Q be the common
chord of the circle x2 + y 2 − 2x − 4y = 0 and the given parabola. Then
the area of the triangle P QS is ..... .

Answer and Comments: 4 (square units). A straightforward prob-

lem of finding the area of a triangle from the coordinates of its vertices.
In the present problem, none of the three vertices P, Q, and S is spec-
ified directly. So the real task is to determine these first. As the focus
of the parabola is at S, we get S = (2, 0). To determine P and Q, we
have to solve the equations
x2 + y 2 − 2x − 4y = 0 (1)
and y 2 = 8x (2)
simultaneously. Substituting (2) into (1) and then solving for y gives
4y = x2 + 6x (3)
Squaring and using (2) again, we get an equation of degree 4 in x, viz.
x4 + 12x3 + 36x2 − 128x = 0 (4)
x = 0 is a solution of this equation. It gives (0, 0) as one of common
points of the circle and the parabola which is obvious anyway. Let us
take it to be P . The other solutions satisfy the cubic As soon as we
determine the other end of the common chord, viz. Q, the problem is
almost solved. But it is not easy to find Q by mere inspection. The
x-coordinate of Q must satisfy the cubic equation
x3 + 12x2 + 36x − 128 = 0 (5)
By trial we see that x = 2 is a solution of this. (Geometrically, we
already know that there are no other real solutions. For an analytical
proof we factorise the cubic as (x − 2)(x2 + 14x + 64) and note that the
quadratic factor never vanishes for any real x since 72 − 64 < 0.)
Since x = 2, y 2 = 8x = 16. So y = ±4. The point (2, −4) does not lie
on the circle. So we must have Q = (2, 4). Now that we know all three
vertices of the triangle P QS its area is
0 0 1
1
= 2 0 1 =4 (6)
2
2 4 1

36
Determination of Q is the only non-trivial part in the problem. Our
solution was based on eliminating one of the variables (viz. y) from (1)
and (2) and trial. An alternate (but essentially equivalent) approach
is to take one of the two curves in a parametric form. The parametric
equations of the parabola t2 = 8x are

x = 2t2 , y = 4t (7)

where −∞ < t < ∞. Substitution into (1) gives

4t4 + 16t2 − 4t2 − 16t = 0 (8)

or, in a factorised form

t(t3 + 3t2 − 4) = 0 (9)

Once again, 0 is an obvious root (which corresponds to the root 0 of (4)

and to the point P = (0, 0) of the common chord). The other roots are
the roots of the cubic factor. As the sum of the coefficients is 0, t = 1
is a root. The cubic factors as (t − 1)(t2 + 4t + 4) and the quadratic
factor has no roots. So, t = 0, 1 are the only real roots of (8) and the
corresponding common points are (0, 0) and (2, 4). In this approach
too there was some guesswork. But it was not as much a shot in the
dark as in (5). Adding the coefficients of the terms in a polynomial is
usually the first thing one does to see if it has any obvious non-zero
roots.
Instead of the parabola, we can represent
√ the circle in a parametric
form. Its center is at (1, 2) and radius 5, we have
√ √
x = 1 + 5 cos θ, y = 2 + 5 sin θ (10)

Substituting these into the equation of the parabola, we get

√ √
(2 + 5 sin θ)2 = 8(1 + 5 cos θ) (11)

which, after simplification, becomes

√ √
5 sin2 θ + 4 5 sin θ − 8 5 cos θ − 4 = 0 (12)

If the cosine term were absent this would be a quadratic in sin θ. To

get rid of the cosine term we can take it on one side and square both

37
 the sides and replace cos2 θ by 1 − sin2 θ. But then the equation that
results would be a a fourth degree equation in sin θ. So again we would
face the same difficulty.
Yet another way a candidate can hit upon the point Q is if it
strikes him that the ends of the latus rectum of the parabola y 2 = 8x
are at (2, ±4). If he also observes luckily that (1, 2) is the centre of the
circle, then he would hardly fail to notice that (2, 4) is the end of the
diameter passing through (0, 0). In particular, it is a common point of
the parabola and the circle.
Summing up, no matter which approach is tried there is an element
of luck in determining the point (Q. The contribution of luck is least
in doing it through (9). Once Q is determined, hardly anything is left
in the problem.

Q.18 The value of

 v v 
u u s
 1 1 1 1
u u
 √ √ √ 4 − √ . . .
u 
6 + log3/2  t4 − t4 −
3 2 3 2 3 2 3 2 

is ......

Answer and Comments: 4. This is a somewhat unusual question

because the expression given involves an infinitistic process. But it
is different from the way infinite sums are defined. The sum of an
infinite series is defined as the limit of the sequence of its partial sums.
Here each partial sum is an ordinary finite sum. In the present problem,
however, a number is defined by an infinitistic process in such a manner
that a part of that number is defined by the same infinitistic process.
Logically, such a definition suffers from what is called a vicious cycle.
It is beyond the scope of the JEE level to consider the question as to
whether such a number indeed exists. (A similar situation arises for
what are called continued fractions which were popularly studied at
one time.)
We bypass these philosophical questions and proceed to find the
value of the number, assuming that the expression does define a number

38
uniquely. Denote the number
v v
u u s
1 1 1
u u
√ √4 − √ ...
u
t4 − t4 − (1)
3 2 3 2 3 2

by x. Then, we get the equation

s
1
x= 4− √ x (2)
3 2
Squaring both the sides we get a quadratic in x, viz.
1
x2 + √ x − 4 = 0 (3)
3 2
Solving,
q
− 3√1 2 ± 1
18
+ 16
x =
2
√ q 289
−1 ± 3 2 18
= √
6 2
−1 ± 17
= √ (4)
6 2
The negative sign is discarded since obviously x is non-negative. So,
8
x = √ . Completing the solution is now straightforward. The
3 2 !
1 8
given expression equals 6 + log3/2 √ × √ which simplifies to
3 2 3 2
 2
 
4 4 2
6 + log3/2 9 . Since = = (3/2)−2, the expression equals
9 3
6 − 2 = 4.
Problems like this are not asked frequently. For those who have seen
similar problems earlier, obtaining (2) and hence (3) is rather routine.
(A fairly well-known simple problem of this type is to evaluate
s r q √
2+ 2+ 2+ 2 + ...

39
 which comes out to be 2.) But in the present problem, even after getting
(3), the subsequent work requires considerable numerical calculation
involving radicals. The candidate has also to realise that 289 is a
perfect square. A numerical slip can be fatal here.

Q.19 Let f : IR −→ IR be defined as f (x) = |x| + |x2 − 1|. The total number
of points at which f attains either a local maximum or a local minimum
is ..... .

Answer and Comments: 5. The expression x changes sign at x = 0

while x2 −1 does so at the points ±1. Therefore because of the absolute
value signs in the definition of f (x), to study its behaviour, we have to
separately consider the intervals (−∞, −1), (−1, 0), (0, 1) and (1, ∞).
Further, f (x) is evidently an even function of x. So, it suffices to study
its behaviour only on the first two of these four intervals.
For x < −1, x is negative but x2 − 1 is positive. So,

f (x) = x2 − x − 1 (1)

This is a quadratic whose derivative vanishes only at x = 1/2 which

is outside the interval (−∞, −1). So, f (x) is strictly decreasing on
(−∞, −1). Next we consider f on (−1, 0). Here f (x) = −x + 1 − x2
which is a quadratic in x with derivative vanishing at x = −1/2. The
function is strictly increasing on (−1, −1/2) and strictly decreasing on
(−1/2, 0). As f is an even function, this also implies that f is strictly
increasing on (0, 1/2) and strictly decreasing on (1/2, 1). Similarly, it
is strictly increasing on (1, ∞).
Summing up, f changes its behaviour from increasing to decreasing
or vice versa at the points x = −1, −1/2, 0, 1/2 and x = 1. Therefore
it has a local maximum or a local minimum at these five points.

40
 A fairly simple problem. The rea-
soning given need not be in words.
A good graph will give the answer
equally easily. In fact the graphic
approach has the advantage that one
need not determine the points ±1/2
exactly. All that matters for the
question is that there are two local
maxima, somewhere in (−1, 0) and in
(0, 1).
The question is well suited to test the ability to reason correctly without
having to express the reasoning in words.
Q.20 If ~a, ~b, ~c are unit vectors satisfying |~a − ~b|2 + |~b −~c|2 + |~c −~a|2 = 9, then
|2~a + 5~b + 5~c| is ..... .

Answer and Comments: 3. We first express the lengths in terms of

the dot products. Thus,
|~a − ~b|2 = (~a − ~b) · (~a − ~b)
= |~a|2 + |~b|2 − 2~a · ~b
= 2 − 2~a · ~b (1)

with similar expressions for |~b − ~c|2 and |~c − ~a|2 . Adding these three
expressions the data means
3
~a · ~b + ~b · ~c + ~c · ~a = − (2)
2
To use this information we need some other expression where the term
~a · ~b + ~b · ~c + ~c ·~a occurs. One such expression is |~a + ~b + ~c|2 which, upon
expansion, becomes

|~a + ~b + ~c|2 = |~a|2 + |~b|2 + |~c|2 + 2(~a · ~b + ~b · ~c + ~c · ~a) (3)

The values of all the terms on the R.H.S. are known to us. Putting
them we get
|~a + ~b + ~c|2 = 3 − 3 = 0 (4)

41
This means that the vector |~a + ~b + ~c| has zero length. Hence it must
be the zero vector, i.e.

~a + ~b + ~c = ~0 (5)

Therefore, ~b + ~c = −~a. Substituting this into the given vector gives

2~a + 5~b + 5~c = 2~a − 5~a = −3~a (6)

Therefore

|2~a + 5~b + 5~c| = | − 3~a| = 3|~a| = 3 (7)

since ~a is a unit vector.

The method applied here works only in an extreme case. In general,
knowing the length of a vector does not determine it uniquely. This
happens to be the case only when the length is 0. Even after getting
(5), if, instead of 2~a + 5~b + 5~c we had some other linear combination,
say α~a + β~b + γ~c where the scalars α, β, γ are all distint, we would not
be able to find its length. In the present problem it works because two
of the scalars are equal.
These two unusual features make the problem a little tricky. An
analytically minded candidate is likely to give up when he realises that
the data is, in general, insufficient to give the answer uniquely. If he
has the perseverance to see that in the present problem, the method
goes through, then only he gets the answer.
Problems where an equation can be solved only in an extreme
case are not uncommon. Indeed nearly all problems of trigonometric
characterisations of equilateral triangles are of this kind. (See Chapter
14.) As a typical example, if ABC is a triangle, then an equation like

cos A + cos B + cos C = k (8)

does not determine the angles A, B, C uniquely. But the maximum

3
value k can take is . And when k equals this extreme value, there
2
is a unique solution, viz. an equilateral triangle. The first part of the
present problem can be looked at as a characterisation of an equilateral
triangle. If ABC is a triangle with circumcentre O and circumradius 1

42
 −→ −→ −→
(an essentially arbitrary choice) and we let ~a =OA, ~b =OB and ~c =OC,
then it is not hard to show that these vectors satisfy the data in the
problem if and only if the triangle ABC is equilateral. Questions in-
volving characterisations of equilateral triangles are not suitable for an
MCQ format because here the answer is trivial to guess but not so
trivial to prove. A candidate who has studied such characterisations is
likely to be prompted by the very symmetry of the data to think that
the triangle formed by the end points of the three vectors (assuming
they have a common origin) is equilateral. Unless he is a scrupulous
candidate, he will not bother to prove this. He will just take it for
granted. However, even after getting this, the last part does require
that two of the scalars in the given linear combination of ~a, ~b, ~c are
equal.

43
 PAPER 2

Contents

Section I (Single Correct Choice Type)

Section II (Paragraph Type)

Section III (Multiple Correct Answers)

Single Correct Choice Type

This section contains eight multiple choice questions. Each question has
4 choices out of which ONLY ONE is correct. 4 points for a correct answer
−1 for an incorrect one.

Q.21 Let a1 , a2 , a3 , . . . be in harmonic progression with a1 = 5 and a20 = 25.

The least positive integer n for which an < 0 is

(A) 22 (B) 23 (C) 24 (D) 25

Answer and Comments: (D). Problems on H.P.’s are not as common
as those on A.P.’s or G.P.’s. One of the reasons is that unlike in the
case of an A.P. or G.P., there is no handy formula for the sum of an
H.P. The harmonic progressions are defined in terms of arithmetic ones.
So the standard approach to tackle any problem about numbers in an
H.P. is to consider the A.P. of their reciprocals.
So, we let
1
bi = (1)
ai
for i = 1, 2, . . . , 25. Then the b’s are in an A.P. We are also given that
1 1
b1 = and b20 = (2)
5 25

Let the common difference of this progression be d. Then we get

b20 = b1 + 19d (3)

44
 Substituting the values from (1), this gives an equation for d, biz.
1 1 4
19d = − =− (4)
25 5 25
4
which gives d = − . Therefore,
475
4(n − 1) 1 96 − 4n
bn = − + = (5)
475 5 475
Clearly, an and bn have the same signs. So, an and hence bn is negative
if and only if

96 < 4n (6)

The least n for which this happens is n = 25.

A very straightforward problem. The problem is more about A.P.’s
than H.P.’s. But had it been asked about A.P.’s it would be too easy.
So the paper-setters have given it a slight twist.

Q.22 Let α(a) and β(a) be the roots of the equation

√ √ √
( 3 1 + a − 1)x2 + ( 1 + a − 1)x + ( 6 a + 1 − 1) = 0

where a > −1. Then lim+ α(a) and lim+ β(a) are
a→0 a→0

5 1 7 9
(A) − and 1 (B) − and −1 (C) − and 2 (D) −
2 2 2 2

Answer and Comments: (B). The problem involves a novel com-

bination of the roots of a family of quadratic equations, parametrised
by a parameter (viz. a here) and their limits as this parameter tends
to some value. Such combinations are often baffling to the average
candidates. So, the problem is a good test of a candidate’s ability to
correctly understand what the problem is all about.
An honest approach to solve the problem is to first express α(a)
and β(a) as functions of a and then take the limits of these functions as
a → 0+ . The various fractional powers make the expression look ugly.

45
So we make a suitable substitution
√ to make the expression look a little
more manageable. We put u = a + 1. Then the quadratic becomes
6

(u2 − 1)x2 + (u3 − 1)x + (u − 1) = 0 (1)

The coefficient u2 − 1 is non-zero as a > −1. So the roots are

q
(1 − u3 ) ± (u3 − 1)2 − 4(u2 − 1)(u − 1)
x= (2)
2(u2 − 1)
If we want, we could express these in terms of the original parameter a.
A better way out, instead, is to work with u instead of a while finding
the limits of these roots. As a → 0+ , u → 1+ . So, our task is to find
the limits of the two expressions in (2) as u → 1+ . This can be done
easily even without using the all time favourite l’Hôpital’s rule. We
simply cancel the factor (u − 1) and get
q
−(u2 + u + 1) ± (u2 + u + 1)2 − 4(u + 1)
x= (3)
2(u + 1)
√
−3 ± 9−8 1
Then the limits are , i.e. − and −1. (The data is
4 2
insufficient to decide which one is the limit of α(a) and which one β(a).
Nor does the problem ask it.)
This gives the limits of the roots of the quadratic (1) as the
parameter u tends to 1+. But the operation of a limit commutes with
many other operations in mathematics. That is why we have theorems
which can be verbally expressed by statements like ‘A limit of a sum
is the sum of the limits’ or ‘The limit of a quotient is the quotient of
the limits provided the limit of the denominator is non-zero’. We also
have theorems which assert that the limit of the sine is the sine of the
limit, or more precisely, if lim f (x) = L, then lim sin(f (x)) = sin L.
x→c x→c
Indeed this is equivalent to saying that the sine function is continuous
(at L). More generally the result holds if we replace the sine function
by any continuous function. But the function has to be single valued.
(Otherwise it is not a function at all!)
If we allow an extension of such results even when we have multi-
valued functions, then we can give a shorter solution of the problem.

46
 Here the roots of the quadratic (1) form a bivalued function of the
parameter u. We identified them separately and found their limits as
u → 0+ . We can as well interchange the two processes. That is, we first
take the limit of the quadratic (1) as u → 0+ and then find the roots
of this limiting quadratic. If we do this indiscriminately, the limiting
quadratic comes out to be 0x2 + 0x + 0 = 0. To avoid this degeneracy,
we replace (1) by an equivalent quadratic valid for u 6= 1, viz.

(u + 1)x2 + (u2 + u + 1)x + 1 = 0 (4)

We now let u → 1. Then the limiting quadratic is

2x2 + 3x + 1 = 0 (5)
√
−2 ± 9−8 1
whose roots are , i.e. −
and −1.
4 2
This alternate approach is certainly a short cut. But a rigorous
proof of its validity is beyond the JEE level. So, once again a candidate
who just assumes its validity gets rewarded.

Q.23 Four fair dice D1 , D2 , D3 and D4 each having six faces numbered 1,2,3,4,5
and 6 are rolled simultaneously. The probability that D4 shows a num-
ber appearing on one of D1 , D2 and D3 is
91 108 125 127
(A) (B) (C) (D)
216 216 216 216

Answer and Comments: (A). Superficially, this is a problem on

probability. But since the desired probability is merely the ratio of
the number of favourable cases to the total number of cases, in reality
it is a counting problem. As the figures shown by the four dice are
independent of each other the total number of cases is 64 .
To find the number of favourable cases, let xi denote the figure shown
by Di for i = 1, 2, 3, 4. We want the number of ordered quadruples
(x1 , x2 , x3 , x4 ) with 1 ≤ xi ≤ 6 (i = 1, 2, 3, 4) for which x4 equals at
least one of x1 , x2 , x3 . We resort to complementary counting, i.e. we
count the number of (ordered) quadruples (x1 , x2 , x3 , x4 ) in which x4
is different from all x1 , x2 , x3 and then subtract this number from 64 ,
the total number of quadruples. Here x4 can take six possible distinct

47
 values. For any fixed value of x4 , the remaining three x1 , x2 , x3 are free
to take any of the five remaining values independently of each other.
This they can do in 53 ways. So the number of unfavourable cases
is 6 × 53 . Hence the number of favourable cases is 64 − 6 × 53 , i.e.
6 × (63 − 53 ) = 6 × (216 − 125) = 6 × 91.
6 × 91 91
Therefore the desired probability is 4
= .
6 216
Instead of using complementary courting, the number of favourable
cases could also have been obtained directly. But this resolves into
several cases depending upon how many and which of x1 , x2 , x3 equals
x4 . The complementary counting gives the answer is one shot. So,
this is a simple problem which rewards those who get the idea of using
complementary counting.
The language used in the question can be confusing. Some candi-
dates are likely to interpret ‘one’ as exactly one. What is intended here
is ‘at least one’. Fortunately, in this problem the wrong interpretation
75
would give the answer as which is not one of the alternatives. So
216
that will alert the candidate, but only after wasting some precious time.
The best thing is to word the question so as to leave no ambiguity.
7 5
Q.24 Let P QR be a triangle with area ∆ with a = 2, b = and c = , where
2 2
a, b, c are the lengths of the sides of the triangle opposite to angles at
2 sin P − sin 2P
P, Q and R respectively. Then equals
2 sin P + sin 2P
2 2
3 45 3 3
 
(A) = (B) = (C) = (D) =
4∆ 4∆ 4∆ 4∆

Answer and Comments: (C). Before tackling the problem, it is well

to comment the entertaining part of the notations used in its state-
ment. In the good old days, the triangle would invariably be taken as
ABC instead of P QR and then nobody needed to be told what a, b, c
stood for. The notation ∆ for the area of the triangle is also standard.
When the JEE became an MCQ test, this time honoured practice was
changed, the rationale apparently was that since the alternatives to
the question are also marked as (A), (B), (C), (D), the use of the same
symbols to denote anything else could confuse the candidates. It is

48
 doubtful if anybody would really get confused for this reason!
Now coming to the problem itself, let us first simplify the expression
to be evaluated.
2 sin P − sin 2P 1 − cos P 2 sin2 P2 2 P
= = 2 P = tan (1)
2 sin P + sin 2P 1 + cos P 2 cos 2 2

There are numerous formulas for the various trigonometrical func-

tions of the angles of a triangle in terms of its sides. In the present
problem, there is an implicit hint that we should choose a formula
which involves the area ∆. (The sides are given and they determine
∆ numerically. So, there must be some purpose in giving all the al-
ternatives in terms of ∆ rather than as numbers.) One such formula
is
v
P u (s − b)(s − c)
u
tan = t (2)
2 s(s − a)
1
where s is the semi-perimeter (a+b+c). Combining this with another
2
well-known formula for the area, viz.
q
∆= s(s − a)(s − b)(s − c) (3)
we get
P (s − b)(s − c)
tan
= (4)
2 ∆
7 5 1 7 5
 
Squaring and putting b = , c = and s = 2+ + = 4, we get
2 2 2 2 2
2
P (1/2)2(3/2)2 3

2
tan = 2
= (5)
2 ∆ 4∆
A straightforward problem once you get an idea which formula to
use. The hint for this is given in a subtle and an imaginative manner.
Q.25 If P is a 3 ×3 matrix such that P T = 2P + I, where P T is the transpose
of P and
 Iis the
 3×3
 identity matrix, then there exists a column matrix
x 0
X =  y  6=  0  such that
   

z 0

49
  
0
(A) P X =  0  (B) P X = X (C) P X = 2X (D) P X = −X
 

Answer and Comments: (D). As P (and hence also P T ) is a 3 × 3

matrix, the single matrix equation

P T = 2P + I (1)

is equivalent to a system of 9 equations obtained by matching the cor-

responding entries of the matrices on both the sides. But to try to
determine P this way would be cumbersome. Instead we use the fact
that every square matrix can be expressed uniquely as a sum of a sym-
metric and a skew symmetric matrix. So, suppose

P =A+B (2)

where A is symmetric and B is skew symmetric. Then

P T = AT + B T = A − B (3)

So, (1) becomes

A − B = 2A + 2B + I (4)

We rewrite this as

−A − I = 3B (5)

The L.H.S. is a symmetric while the R.H.S. is a skew symmetric matrix.

Hence both the sides equal the zero matrix of order 3. This gives

B = 0 and A = −I (6)

which combined with (2) gives

P = −I (7)

As a result, for every 3 × 1 matrix X,

P X = −X (8)

50
 Therefore (D) is true. In fact what we have proved is much stronger
than (D). (D) only requires some non-zero 3 × 1 matrix X for which
P X = −X. (A slicker derivation of (7), poined out by Sujeet Singh, is
to take the transpose of both the sides of (1) and combine with (1) to
get 3P = −3I.)
Those familiar with eigenvalues will recognise (D) to mean that P
has −1 as an eigenvalue. There are methods for finding eigenvalues of
a square matrix using determinants. But in this problem this is of little
help.
√
Q.26 If ~a, ~b are vectors such that |~a + ~b| = 29 and

~a × (2î + 3ĵ + 4k̂) = (2î + 3ĵ + 4k̂) × ~b

then a possible value of (~a + ~b) · (−7î + 2ĵ + 3k̂) is

(A) 0 (B) 3 (C) 4 (D) 8

Answer and Comments: (C). For brevity, denote the vector 2î +
3ĵ + 4k̂ by ~v . Then the given condition is

~a × ~v = ~v × ~b (1)

The R.H.S. equals −~b × ~v. So (1) gives

(~a + ~b) × ~v = ~0 (2)

which is equivalent to saying that ~a + ~b is a scalar multiple of ~v , say

~a + ~b = λ~v (3)

Taking lengths of both the sides

|~a + ~b| = |λ||~v| (4)

√
The L.H.S.
√ is given as √29. By a direct calculation |~v| also comes out
to be 4 + 9 + 16 i.e. 29. So (4) gives |λ| = 1 and hence λ = ±1.
Therefore

~a + ~b = ±(2î + 3ĵ + 4k̂) (5)

51
 and hence

(~a + ~b) · (−7î + 2ĵ + 3k̂) = ±(2î + 3ĵ + 4k̂) · (−7î + 2ĵ + 3k̂)
= ±(−14 + 6 + 12) = ±4 (6)

So there are two possible values of the given expression and one of them
is 4.
The problem is a simple but good test of the knowledge of the basic
properties of the cross product, specifically, its anti-commutativity and
the condition for its vanishing. It is a pure coincidence that the value
of |~a + ~b| is given to be the same as |~v|. Apparently, the paper-setters
wanted to keep the computations simple and thereby reward a candi-
date who gets the conceptual part correctly. By using the words ’a
possible value’ instead of merely saying ‘a value’ the paper-setters have
made it abundantly clear that there may be more than one correct an-
swers but only one of them is asked. This leaves no room for confusion
in the mind of the candidate.

Q.27 The equation of a plane passing through the line of intersection of the
2
planes x + 2y + 3z = 2 and x − y + z = 3 and at a distance √ from
3
the point (3, 1, −1) is
√ √
(A) 5x − 11y + z√= 17 (B) 2x√+ y = 3 2√
−1
(C) x + y + z = 3 (D) x − 2y = 1 − 2

Answer and Comments: (A). The family of planes passing through

the line of intersection of two given planes is a 1-parameter family
and the equation of any member of it can be written by combining
the equations of these two planes. Thus, in the present problem, the
desired plane will have an equation of the form

(1 − λ)(x + 2y + 3z − 2) + λ(x − y + z − 3) = 0 (1)

for some real number λ. (See Chapter 9, Comment No. 13 for a

discussion about the analogous result in the plane for the equation of a

52
line passing through the point of intersection of two given lines. Often
the coefficient (1 − λ) is dropped. But the danger in doing so is that
we miss the second plane and if it happens to be a right answer, we
miss it too. So we work with (1).) Rewriting it,

x + (2 − 3λ)y + (3 − 2λ)z − 2 − λ = 0 (2)

2
The distance of this plane from the point (3, 1, −1) is given to be √ .
3
This gives us an equation for λ, viz.

3 + 2 − 3λ + 2λ − 3 − 2 − λ 2
q =√ (3)
1 + (2 − 3λ)2 + (3 − 2λ)2 3

Squaring and simplifying

λ2 1
2
= (4)
13λ − 24λ + 14 3
This becomes a quadratic in λ, viz.,

10λ2 − 24λ + 14 = 0 (5)

whose roots are

√
12 ± 144 − 140
λ= (6)
10
7
Hence λ = 1 or . The first possibility does give the second plane, viz.
5
x − y + z = 3 as a possible answer. The second value, viz., λ = 75 , after
putting into (2) gives the plane
11 1 17
x− y+ z− =0 (7)
5 5 5
which is the plane in (A). Technically, the problem is not wrong because
it asks for the equation of a plane with certain properties and the plane
in (A) meets the prescription. But in reality there are two such planes.
It would have been better to ask this question as an MCQ where more
than one alternatives may be correct and also give the equation of the
second plane as one of the alternatives.

53
 What is really disturbing about this problem is that the majority
of the candidates would begin the solution by taking the parametrised
family of planes as

(x + 2y + 3z − 2) + λ(x − y + z − 3) = 0 (8)

where λ is a real parameter. With this choice the resulting equation in

λ comes out to be

3λ2 = 3λ2 + 4λ + 14 (9)

7
which is a linear equation, with λ = − as the only solution. This
2
solution gives the plane in (A) correctly. But an analytically minded
candidate would realise that unless d = 0, there are two planes in the
given family which are at a distance d from a given point. He will then
wonder which is the other plane. It is only when he realises that the
second of the given two planes is also a right answer, that he would
identify that the villain was to use (8) instead of (1) which is the right
thing.
It is hard to tell whether this possibility eluded the paper-setters
too. Probably it did, as otherwise, they would have asked the question
differently. That is, they would have made it abundantly clear that one
of the answers given is only one of the possible answers to the problem,
as they did in framing the last question. As the problem stands, those
who miss the subtle difference between (1) and (8) get unduly rewarded.
Ignorance is bliss!
π/2 π+x
Z  
2
Q.28 The value of the integral x + ln cos x dx is
−π/2 π−x
π2 π2 π2
(A) 0 (B) −4 (C) +4 (D)
2 2 2

Answer and Comments: (B). When the integrand of a function

splits as the sum of two functions, the integral also splits as the sum of
two integrals. In the present case we can write the integral, say I as

I = I1 + I2 (1)

54
where
Z π/2
I1 = x2 cos x dx (2)
−π/2
π+x
Z π/2  
and I2 = ln cos x dx (3)
−π/2 π−x
Occasionally, in some tricky problems on integration, it is not a good
idea to split the integrand. Instead, it is easier to evaluate the integral
as a whole by some trick. See for example, the JEE 1983 problem in
Comment No. 24 of Chapter 17. But the present integral is not of
this tricky type. Here I1 can be evaluated by a standard application of
integration by parts. So, the evaluation of I depends more on that of I2 .
Here no matter which of the two factors of the integrand is taken as the
first function, integration by parts will only lead to more complicated
integrands. No substitution suggests itself either. So, it is futile to
try to evaluate I2 by finding an antiderivative of the integrand. Nor
is it necessary. The integral asked is a definite integral and as pointed
out in Chapter 18, there are ways to evaluate a definite integral taking
advantage of some special features of the problem such as the symmetry
of the interval around 0 or some special property of the integrand.
In the present problem, the interval of integration is symmetric
π+x π−x
about the point 0. Further for any x in this interval, and
π−x π+x
are reciprocals of each other and so their logarithms are negatives of
each other. Therefore the first factor of the integrand in (3) is an
odd function. The second factor, cos x, is an even function. So the
integrand of I2 is an odd function over the interval [−π/2, π/2] which
is symmetric about 0. Thus we get
I2 = 0 (4)
without any computation. As for I1 , the integrand is an even function
of x. So again by the symmetry of the interval of integration, we have
Z π/2
I2 = 2 x2 cos x dx (5)
0

The integral on the R.H.S. can be evaluated by two standard applica-

tions of integration by parts. First we find the indefinite integral.
Z Z
x2 cos x dx = x2 sin x − 2x sin x dx

55
 Z
= x2 sin x + 2x cos x − 2 cos x dx
= x2 sin x + 2x cos x − 2 sin x (6)
So, finally,
π/2 π/2 π/2
I = I2 = 2x2 sin x − 4x cos x − 4 sin x
0 0 0
π2 π2
= −0−4 = −4 (7)
2 2
π+x
The only non-trivial part in the solution is to recognise that ln
π−x
is an odd function of x. Thereafter the problem is a routine one.

SECTION II
Paragraph Type

This section contains three paragraphs each with two multiple choice
questions. Each question has four possible answers out of which only one
is correct. 3 marks for a correct answer, −1 for an incorrect one.

Paragraph for Question No.s 29 to 30

Let f (x) = (1 − x)2 sin2!x + x2 for all x ∈ IR and let

Z x
2(t − 1)
g(x) = − ln t f (t)dt for all x ∈ (1, ∞).
1 t+1
Q.29 Which of the following is true?

(A) g is increasing on (1, ∞)

(B) g is decreasing on (1, ∞)
(C) g is increasing on (1, 2) and decreasing on (2, ∞)
(D) g is decreasing on (1, 2) and increasing on (2, ∞)

Answer and Comments: (B). By the second form of the fundamental

theorem of calculus, we have
!
′ 2(x − 1)
g (x) = − ln x f (x) (1)
x+1

56
 for all x ∈ (1, ∞). The factor f (x) is positive for all x ∈ (1, ∞). So
the sign of g ′ (x) depends only on that of the first factor. Denote it
by h(x). The problem now reduces to deciding the sign of h(x) for
x ∈ (0, ∞). This is not easy if attempted directly because we would
have to compare a rational function with a logarithmic function. But it
is easier to compare their derivatives as they are both rational functions.
Rewriting h(x) as
4
h(x) = 2 − − ln x (2)
x+1
we have
4 1 −(x − 1)2
h′ (x) = − = (3)
(x + 1)2 x x(x + 1)2
which is negative for all x ∈ (1, ∞). Hence h is strictly decreasing on
(1, ∞). By a direct calculation, h(1) = 0. Therefore h(x) < h(1) = 0
for all x ∈ (1, ∞). Hence g ′(x) = h(x)f (x) is also negative for all
x ∈ (1, ∞) making g strictly decreasing on (1, ∞).
The problem is straightforward except that the determination of
the sign of h(x) has to be done by taking its derivative.
Q.30 Consider the statements:

P : There exists some x ∈ IR such that f (x) + 2x = 2(1 + x2 )

Q : There exists some x ∈ IR such that 2f (x) + 1 = 2x(1 + x)

Then

(A) both P and Q are true (B) P is true and Q is false

(C) P is false and Q is true (D) both P and Q are false

Answer and Comments: (C). Let us first consider statement P. The

given equality means
(1 − x)2 sin2 x + x2 + 2x = 2 + 2x2 (4)
which reduces to
(1 − x)2 sin2 x = (x − 1)2 + 1 (5)

57
 and further, to

(1 − x)2 cos2 x = −1 (6)

As the L.H.S. is always non-negative, there can be no solution to (4).

So the statement P is false.
As for the statement Q, the given equality means

2(x − 1)2 sin2 x + 2x2 + 1 = 2x + 2x2 (7)

i.e.

p(x) = 0 (8)

where

p(x) = 2(x − 1)2 sin2 x − 2x + 1 (9)

Suppose x > 0. When sin2 x = 0, p(x) equals −2x+ 1 which is negative

for x > 1/2. But when sin2 x = 1, p(x) = 2(x − 1)2 − 2x + 1 which
is positive for x sufficiently large. Thus we see that p(x) assumes both
positive and negative values on IR. Clearly p(x) is continuous every-
where. Therefore by the Intermediate Value Property of continuous
functions, p(x) vanishes at least once in IR. Hence the statement Q is
true. (Actually, p(x) has many zeros, for example, the reasoning above
gives that it has a zero in every interval of the form [nπ, nπ + π/2] for
every positive integer n. But, for the truth of the statement Q, all we
need is one zero.)
The part of the problem involving the statement Q is more interest-
ing because it takes merely a glance to see that p(x) can assume both
positive and negative values.

Paragraph for Question No.s 31 to 32

Let an denote the number of all n-digit positive integers formed by the
digits 0 and 1 or both such that no consecutive digits in them are 0. Let
bn be the number of such n-digit integers ending with digit 1 and cn be the
number of such n-digit integers ending with digit 0.

58
Q.31 Which of the following is correct?

(A) a17 = a16 + a15 (B) c17 6= c16 + c15

(C) b17 =
6 b16 + c16 (D) a17 = c17 + b16

Answer and Comments: (A). Evidently, there is nothing great

about the subscript 17. The statements could as well have been made
about an , bn and cn for any n ≥ 3. Let us, in fact, do that. It is often
easier to think of something for a general n, than for a particular value
of n.
When we talk of the number of digits of a positive integer, there is
some ambiguity as to whether the leading 0’s are to be counted or not.
We proceed on the assumption that the leading zeros do not count. As
the only digits allowed are 0 and 1, an n-digit number must begin with
a 1 as its first or topmost digit. (Caution: Here by first, we mean as
we go from left to right. Sometimes, by the first digit one means just
the opposite, i.e. the digit in the units place.)
With this clarification, we now introduce certain sets. Thus for every
positive integer n, let An be the set of all n-digit number in which no
two consecutive digits are 0, Bn be the set of those members of An in
which the last digit is 1 and Cn be the set of those members of An which
end with 0. Obviously, |An | = an , |Bn | = n and |Cn | = cn . Further, it
is clear that An is the disjoint union of Bn and Cn . So, we have

an = bn + cn (1)
for every n = 1, 2, 3, . . .. Now suppose n ≥ 2 and take a typical member
~x = (1x2 . . . xn−1 1) of Bn . If we remove the last digit 1, then the (n−1)-
digit number (1x2 . . . xn−1 ) is in An−1 . Conversely, from any number in
An−1 we get a number in Bn by appending a 1 at the end. This gives
us a bijection between Bn and An−1 . Therefore
bn = an−1 (2)
However, if ~x = (1x2 x3 . . . xn−1 0) ∈ Cn , then xn−1 has to be 1 (as
otherwise the last two digits of ~x will both be 0). Hence the (n − 1)-
digit number (1x2 x3 . . . xn−2 1) lies in Bn−1 . Conversely, from every

59
 number in Bn−1 we get a number in Cn by appending a 0 at the end.
This gives us a bijection between Cn and Bn−1 and hence

cn = bn−1 (3)

For n ≥ 3 we can combine (1), (2) and (3) to get

an = bn + cn = an−1 + bn−1 = an−1 + an−2 (4)

We now get (A) as a special case, with n = 17. As only one of the
alternatives is given to be correct, we need not bother about the others.
Still, for the sake of completeness we point out that in view of (2) and
(3) the sequences {bn } and {cn } also satisfy the same recurrence relation
as {an }. That renders (B) false. Replacing c16 by b15 we see that (C)
is also false. Finally, replacing a17 by b17 + c17 we see that for (D) to
be true, we must have b17 = b16 . But that is not true.
The recurrence relation (4) satisfied by the sequence {an } is the
well-known Fibonacci relation. As commented in Exercise (5.17), it
arises naturally in many combinatorial problems. The present problem
is also strikingly similar to the JEE 2000 problem in Comment No. 14
1
of Chapter 22. Indeed, if the coin in that problem is fair (i.e. p = ),
2
an+1
then the probability pn is that problem is precisely n where an is as
2
in this problem.
Q.32 The value of b6 is

(A) 7 (B) 8 (C) 9 (D) 11

Answer and Comments:. (B). We already proved that bn = an−1

for all n > 1. So it suffices to calculate a5 . For this we repeatedly
use the recurrence relation (4). But we have to feed the initial values
a1 and a2 first. Clearly a1 = 1 and a2 = 2 because A2 = {11, 10}.
So successive applications of (4) give a3 = a1 + a2 = 1 + 2 = 3 and
a4 = a2 + a3 = 2 + 3 = 5. Finally, a5 = a3 + a4 = 3 + 5 = 8. This is
also the value of b6 .
Although this solution is simple enough, those who recognise (4)
as the Fibonacci relation and also recognise that a1 and a2 are the
Fibonacci numbers 1 and 2 will realise that all subsequent terms of

60
 the sequence {an } are Fibonacci numbers. The first few Fibonacci
numbers, viz. 1, 2, 3, 5, 8, 13, 21, ... are quite well-known. As (B)
is the only answer with a Fibonacci number, it must be the right one
without doing any calculation.
Finally, if we count leading zeros too, then an n-digit number with
only 0 and 1 as possible digits is nothing but a binary sequence of length
n, with no restriction on the first term. With this interpretation, the
values of an , bn and cn will come out to be the values of an+1 , bn+1 and
cn+1 with the present interpretation, because if we take any member
of An+1 and remove the leading 1, we get a binary sequence of length
n with no consecutive zeros and conversely, if we start with such a
sequence and put a 1 in front we get a member of An+1 . With this new
interpretation, there would be no change in the answer to Q.31. But
in Q.32, the value of b6 will not be 8 but rather the next Fibonacci
number after 8, that is 5+8, i.e. 13. Since this is not in any given
alternatives, that will alert the candidate that his interpretation of the
problem was wrong.

Paragraph for Question No.s 33 and 34

√
A tangent P T is drawn to the circle x2 + y 2 = 4 at the point P ( 3, 1). A
straight line L, perpendicular to P T is a tangent to the circle (x−3)3 +y 2 = 1.

Q.33 A common tangent of the two circles is

√ √
(A) x = 4 (B) y = 2 (C) x + 3y = 4 (D) x + 2 2y = 6

Answer and Comments: (D). Call the first circle ac C1 and the
second as C2 . Their centres, say O and C, lie at (0, 0) and (3, 0)
respectively, while their radii, say r1 and r2 are 2 and 1 respectively.
Since OC = 3 = r1 + r2 , the two circles touch each other externally at
the point (2, 0). So, the line

61
 T

L
D

C
O A B

x = 2 is their internal common tangent. Since both the centres lie on

the x-axis, the two common external tangents must meet at some point,
say B, on the x-axis. This point divides the segment OC externally
in the ratio r1 : r2 , i.e. in the ratio 2 : 1. This determines the point
B as (6, 0). Among the given alternatives, the line in (D) is the only
one which passes through the point (6, 0). Therefore (D) is the right
answer. Nevertheless, to arrive at it honestly, we must find the slopes
of the common tangents.
For this, suppose one of the common tangents touches the circle C1 at
D. Then from the right angled triangle ODB, we get
OD 2
tan 6 OBD = =√
DB OB − OD 2
2

2 2 1
= √ = √ = √ (5)
36 − 4 4 2 2 2
1
Therefore the slopes of the two common external tangents are ± √ .
2 2
As they pass through the point (6, 0) their equations are
√
x ± 2 2y = 6 (6)

Q.34 A possible equation of L is

√ √
(A) x − 3y = 1 (B) x + 3y = 1
√ √
(C) x − y = −1 (D) x + y = 5

62
 Answer and Comments: (A). Let √ us first find the equation of the
tangent to the first circle C1 at P ( 3, 1). This is obtained from the
popular formula T = 0, which gives
√
3x + y = 4 (7)

1
The line L is perpendicular to this and hence has slope √ . Also L
3
touches C2 . We can now find the equation of L from the formula for
the equation of a tangent to a conic in terms of its slope. But in the
present case, as the conic is a circle, there is an easier geometric way
1
out. Since L has slope √ , its equation is of the form
3
√
x − 3y + c = 0 (8)

To determine c, we equate the perpendicular distance of the center

(3, 0) from (8) with the radius, which is 1. This gives

3+c
√ =1 (9)
1+3

i.e. |c + 3| = 2 which has two solutions,

√ c = −5 √
and c = −1. So the
possible equations of L are x − 3y = 1 and x − 3y = 5.

SECTION III
Multiple Correct Answer(s) Type

This section contains six questions with four answers to each, out of which
one or more may be correct. 4 points for identifying all correct answers, 0
otherwise.
1 4 4
Q.35 If the adjoint of a 3 × 3 matrix P is 2 1 7 , then the possible
1 1 3
value(s) of the determinant of P is (are)

(A) −2 (B) −1 (C) 1 (D) 2

63
Answer and Comments: (A, D). It is futile to try to find the matrix
P from its adjoint. Doing so would require us to solve a system of 9
equations in 9 unknowns. And to make things worse, these equations
are not linear! So, we look for some other method. We observe that our
job is not so much to find the matrix P but merely to find its determi-
nant, say D. Now, there is a relationship between the determinant of
P and its adjoint. For an n × n matrix P it is given by

QP = DIn (1)

where Q is the adjoint matrix of P . (Sometimes, Q itself is called the

adjoint of P . But in the present problem the adjoint is to be treated
as a number and so it equals the determinant of the adjoint matrix Q.)
Taking determinants of both the sides,

det(Q)det(P ) = D n (2)

The second factor on the L.H.S. is simply D (which we want to find in

the present problem). The first factor is called the adjoint of P and in
1 4 4
the present problem given to us as the determinant 2 1 7 whose
1 1 3
value comes out to be 4 by a direct expansion. Taking determinants of
both the sides in (1) (and keeping in mind that n = 3), we have

4D = D 3 (3)

This has three roots, D = 0, 1 and −1. Two of these are included in the
answers. The value 0 is not given as a possible answer and so we need
not bother about it. But if it were given as one of the alternatives, we
would have to dismiss it. (Note that it is not given in the statement of
the problem that the matrix P is non-singular.) This is easy but not
trivial. We go back to equation (1). If D = 0 then the R.H.S. would
be the zero matrix 0 and we would get

QP = 0 (4)

But Q is non-singular (as its determinant is non-zero) and so Q−1 exists.

Multiplying by it on the left gives

P = In P = Q−1 QP = Q−1 0 = 0 (5)

64
 But this is a contradiction because if P is the zero matrix so would be
Q.
Thus we have eliminated the possibility D = 0. Most candidates will
either not see the need for doing so or will simply take it for granted.
(In fact, some people go to the extent of saying that the adjoint always
equals D n−1 . Although this is true, the proof needed for D = 0 is dif-
ferent from that when D 6= 0.) So, once again the scrupulous candidate
gets a beating. That mars this otherwise excellent question.
Q.36 If the straight lines
x−1 y+1 z x+1 y+1 z
= = and = =
2 k 2 5 2 2
are coplanar then the plane(s) containing these two lines is (are)

(A) y + 2z = −1 (B) y + z = −1 (C) y − z = −1 (D) y − 2z = −1

Answer and Comments: (B, C). The wording of the problem is a

little confusing. If two (distinct) lines are given to be coplanar, then
there is only one plane containing them. So the possible plural in the
answer can be confusing. The catch here is that the value of k is not
specified. It is an unknown parameter which can take any non-zero real
value. (Even k = 0 is allowed with the understanding that in that case
one of the equations for the first line is y = −1 and that for the second
line is z = 0.) As the value of k changes, the given lines will change and
so will their coplanarity. Thus first we have to determine those values
of k for which the given lines are coplanar and then for each such value
find the equation of the (unique) plane containing them.
Let us do both these tasks simultaneously. Call the two lines as L1
and L2 . (Strictly speaking, we should call them L1 (k) and L2 (k) since
they depend on k. But let us not be so fussy.) Also, in case they are
coplanar, let the (unique) plane containing them be given by
ax + by + cz = d (1)
(where again, a, b, c, d depend on k). The parametric equations of L1
and L2 are, respectively,
x = 1 + 2t, y = −1 + kt, z = 2t (2)
and x = −1 + 5s, y = −1 + 2s, z = ks (3)

65
The plane (1) will contain the line L1 if and only if the equation

a(1 + 2t) + b(−1 + kt) + c(2t) = d (4)

is true for all t. This means that it must vanish identically, which gives

2a + kb + 2c = 0 (5)
and a − b = d (6)

Similarly, if (1) contains the line L2 then we get

s(5a + 2b + kc) − a − b = d (7)

for all s which means

5a + 2b + kc = 0 (8)
and − a − b = d (9)

(6) and (9) together imply a = 0 and d = −b. Putting these values
into (5) and (8) and eliminating b gives

4c = k 2 c (10)

Note that c 6= 0. For if c = 0, then since already a = 0, from (8) we

would get b = 0 and then (1) will not represent a plane. So, from (10)
we get k 2 = 4 and hence k = ±2. Thus there are only two values of
k for which the given lines are coplanar. To find the equation of the
plane containing them, for k = 2, (5) gives c = −b. Since d = −b, and
b 6= 0, equation (1) becomes

y − z = −1 (11)

while for k = −2, (5) gives c = b and hence the equation (1) becomes

y + z = −1 (12)

The problem can be done a little more elegantly if we use vectors. Let
A and B, be repsetively, the points (1, −1, 0) and (−1, −1, 0). These
are given to lie on L1 and L2 respectively. Further let ~u = 2î + k ĵ + 2k̂
and ~v = 5î + 2ĵ + k k̂. (It is a little unfortunate that k appears twice.

66
 But one of the appearances is a scalar and the other a vector. So no
confusion is likely.) These vectors are parallel to the lines L1 and L2
respectively.
Now suppose there is some plane, say P , which contains both L1 and
L2 . Then the vectors ~u, ~v are parallel to P . Further, since P contains
−→
both A and B, the vector AB is also parallel to P . Therefore the scalar
triple product of these three vectors must vanish. That is,

−2 0 0
2 k 2 =0 (13)
5 2 k

which is simply k 2 = 4. Hence k = ±2. So for these two values of k

the lines L1 and L2 are coplanar. To find the equations of the plane
containing them, we note that the vector ~u × ~v is normal to this plane.
For k = 2, ~u × ~v = (2î + 2ĵ + 2k̂) × (5î + 2ĵ + 2k̂) = 6ĵ − 6k̂. Since
A = (1, −1, 0) is a point in this plane, its equation is

((x − 1)î + (y + 1)ĵ + z k̂) · (6ĵ − 6k̂) = 0 (14)

which comes out as y − z = −1. A similar computation for k = −2

gives ~u × ~v = (2î − 2ĵ + 2k̂) × (5î + 2ĵ − 2k̂) = 14ĵ + 14k̂. So, the
equation of the plane containing L1 and L2 is

((x − 1)î + (y + 1)ĵ + z k̂) · (14ĵ + 14k̂) = 0 (15)

which comes out as y + z = −1.

This is an interesting problem, except for its somewhat confusing
statement. Moreover, no matter which method is followed, the compu-
tations involved are lengthy and prone to errors. The question would
be suitable as a full length question in the classical JEE.
Z x 2
Q.37 If f (x) = et (t − 2)(t − 3)dt for all x ∈ (0, ∞), then
0

(A) f has a local maximum at x = 2

(B) f is decreasing on (2, 3)
(C) there exists some c ∈ (0, ∞) such that f ′′ (c) = 0
(D) f has a local minimum at x = 3

67
 Answer and Comments: (A, B, C, D). The setting of the problem is
similar to that of Q. 29. In both these problems, a function is defined
by a definite integral and its properties are asked. But these properties
require the derivative of this function, which, by the second form of the
fundamental theorem of calculus, equals the integrand itself. So the
question is really about the integrand. The twist given by considering
a function defined by an integral only serves to test if the candidate
can correctly apply the second FTC. But similar questions have been
asked so frequently in the past that hardly any originality of thought
is tested. Applying it is more of a ritual. It is bad enough to have to
do it once. But to have to do it twice in the same paper beats reason.
Now, coming to the question itself, by the second form of FTC, we
have
2
f ′ (x) = ex (x − 2)(x − 3) (1)

for all x ∈ (0, ∞). All alternatives except possibly (C) involve the
increasing/decreasing behaviour of f and hence the sign of f ′ on various
subintervals of (0, ∞). The exponential factor is always positive. So,
the sign of f ′ (x) is the same as that of (x − 2)(x − 3) which is positive
for x < 2 and x > 3 and negative for 2 < x < 3. So f is increasing on
(−∞, 2), decreasing on (2, 3) and again increasing on (3, ∞). Therefore
(A), (B), (D) are all true. As for (C), since f ′′ is the derivative of f ′ ,
the existence of the point c will follow from Rolle’s theorem, if we could
find two distinct points, say a, b ∈ (0, ∞) such that f ′ (a) = f ′ (b). This
is very easy. As f has a local maximum at x = 2 and a local minimum
at x = 3, f ′ vanishes at both these points. So there is some c ∈ (2, 3)
at which f ′′ vanishes. Since (2, 3) ⊂ (0, ∞), (C) is also true.
1 1
Q.38 Let X and Y be two events such that P (X|Y ) = , P (Y |X) = and
2 3
1
P (X ∩ Y ) = . Which of the following is (are) correct?
6

2
(A) P (X ∪ Y ) = (B) X and Y are independent
3
1
(C) X and Y are not independent (D) P (X c ∩ Y ) =
3

68
 Answer and Comments: (A, B). A straightforward problem on con-
ditional probability, requiring little more than Baye’s theorem. The
first two pieces of data mean, respectively,
P (X ∩ Y ) 1
= (1)
P (Y ) 2
P (X ∩ Y ) 1
and = (2)
P (X) 3
As we are also given
1
P (X ∩ Y ) = (3)
6
from (1) and (2) we get
1 1
P (Y ) = and P (X) = (4)
3 2
We now have all the data to find the probabilities occurring in the given
answers.
1 1 1 2
P (X ∪ Y ) = P (X) + P (Y ) − P (X ∩ Y ) = + − = (5)
2 3 6 3
which shows that (A) is true. To test (B), we need to check if P (X ∩
Y ) = P (X)P (Y ). From the values we now, this is the case. So, (B) is
also true and hence (C), which is the negation of (B) is false. Finally,
1 1 1
P (X c ∩ Y ) = P (Y ) − P (X ∩ Y ) = − = (6)
3 6 6
which shows that (D) is false too.

Q.39 For every integer n, let an and bn be real numbers. Let f : IR −→ IR

be given by
(
an + sin πx, for x ∈ [2n, 2n + 1]
f (x) = , for all integers n.
bn + cos πx, for x ∈ (2n − 1, 2n)
If f is continuous, then which of the following hold(s) for all n ?

(A) an−1 − bn−1 = 0 (B) an − bn = 1

(C) an − bn+1 = 1 (D) an−1 − bn = −1

69
Answer and Comments: (B, D). This is one of the problems which
are hard to understand but once you understand them are very easy
to solve (like somebody asking you to give the name of your father’s
mother-in-law’s daughter!). The sine and the cosine functions are con-
tinuous everywhere. So, continuity of f at non-integral values means
nothing new. But for every integer n, f has a different definition on
the two sides of n and so continuity of f at n means that the left and
the right handed limits are equal to f (n).
Now fix some integer n. Then on the immediate right of 2n,
f (x) = an + sin πx while on its immediate left, f (x) = bn + cos πx.
Therefore,

lim f (x) = = lim+ an + sin πx = an + 0 = an (1)

x→2n+ x→2n
while lim− f (x) = lim bn + cos πx = bn + 1 (2)
x→2n x→2n−

So, continuity of f at x = 2n gives

an = bn + 1 (3)

which means (B) is true and also that (A) is false. . On the other hand,
for x = 2n + 1, f (x) = an + sin πx on the immediate left of 2n + 1 while
on its immediate right it equals bn+1 + cos πx. (Note that it is not bn
but bn+1 here. This is the only catch in the problem.) So,

lim f (x) = lim bn+1 + cos πx = bn+1 − 1 (4)

x→(2n+1)+ x→(2n+1)+

while lim f (x) = lim an + sin πx = an (5)

x→(2n+1)− x→(2n+1)−

So, continuity of f at x = 2n + 1 gives

an = bn+1 − 1 (6)

Replacing n by n − 1 this means an−1 − bn = 1. So (D) is true and (C)

is false.
This is a good problem to test a candidate’s ability to understand
a problem. Once he understands it correctly, the computational work
involved is minimal. Such problems are ideal for multiple choice tests.

70
Q.40 Let f : (−1, 1) −→ IR be such that
2
f (cos 4θ) =
2 − sec2 θ
π π π 1
     
for θ ∈ 0, ∪ , . Then the value(s) of f is (are)
4 4 2 3
s s s s
3 3 2 2
(A) 1 − (B) 1 + (C) 1 − (D) 1 +
2 2 3 3

Answer and Comments: (A, B, C, D). The problem is wrongly

designed. Suppose x ∈ (−1, 1). Then f (x) is not given explicitly. To
π

find it out, we must first express x as cos 4θ for some θ ∈ 0, ∪
4
π π
 
, . Now, as θ varies from 0 to π/4, cos 4θ varies from 1 to −1
4 2
in a one-to-one manner. But as θ varies over (π/4, π/2), then cos 4θ
again varies from −1 to 1 in a one-to-one manner.  So, foreach 
fixed
π π π
x ∈ (−1, 1) there are two distinct values of θ in 0, ∪ , for
4 4 2
which cos 4θ = x. One of these values, say α, lies in (0, π/4) and the
other, say β, lies in (π/4, π/2). These two values are related to each
other by
π
α+β = (1)
2
2
From the formula in the problem, f (x) would equal both
2 − sec2 α
2
and . This means sec2 α = sec2 β and hence
2 − sec2 β

cos α = ± cos β (2)

But (2) does not follow from (1). All that we get from (1) is that
cos α = sin β and sin β is different from cos β or − cos β for all β ∈
(π/4, π/2).
So the data of the problem can never hold. In such a case, any
statement you make about the function f is true by a quirk of logic,
because there is no counterexample to show its falsity, just as anything

71
you say about a six legged man is true. (Truth of this kind is called
vacuous truth. See Exercise (17.22)(a) for another example of a vac-
uous question. The 2003 problem about a 3 × 3 matrix with positive
entries which is its own inverse is also vacuous, because no such matrix
exists.)
Actually, the idea of a function is extraneous to the main theme of
the problem which is purely trigonometrical. The problem could have
2
asked for the possible value(s) of the expression given that
2 − sec2 θ
π π π 1
   
θ ∈ 0, ∪ , and cos 4θ = . Assuming that this is what the
4 4 2 3
paper-setters had in mind, a solution can be given as follows.
We are given something about cos 4θ and the expression to be
evaluated is in terms of cos θ. It is possible to express cos 4θ in terms
of cos θ. But the expression is a fourth degree polynomial in cos θ and
finding the value(s) of cos θ from the value of cos 4θ is not easy. So
we follow a ‘meet half way’ strategy so to speak. That is, instead of
coming down all the way from 4θ to θ we come down from 4θ to 2θ and
1
also go up from θ to 2θ. More precisely, from the data cos 4θ = , we
3
determine the possible values of cos 2θ. We also convert the expression
to be evaluated (which is given in terms of θ) to an expression in cos 2θ.
Then we simply substitute the value(s) of cos 2θ.
The first task is easy. Since cos 4θ = 2 cos2 2θ − 1, we get
1 4
2 cos2 θ = 1 + = (3)
3 3
which gives
s
2
cos 2θ = ± (4)
3
Next, let us simplify the expression to be evaluated.
2 2 cos2 θ
=
2 − sec2 θ 2 cos2 θ − 1
1 + cos 2θ
=
cos 2θ
1
= 1+ (5)
cos 2θ
72
(4) and (5) together imply
s
2 3
=1± (6)
2 − sec2 θ 2
So (A) and (B) are correct answers.
The trigonometric part of the question is interesting because
of the ‘meet half way’ strategy the candidate has to think of. By
unnecessarily twisting it and recasting it in terms of functional values,
the paper-setters have killed a good problem.

73
 CONCLUDING REMARKS

As in the past years, the JEE Mathematics papers this year are a collage of
all kinds of problems. But this year, the proportion of good problems is fairly
impressive. We mention here Q. 1 (about the asymptotic estimation of a ra-
tional function), Q.5 (about the ratio of the determinants of two related ma-
trices),
Q. 8 (about the differentiability of x2 cos πx ), Q.12 (about the estimation
Z 1 2
of the definite integral e−x dx), Q.18 (about a number defined by an in-
0
finitistic process), Q.19 (about the number of local extrema of |x| + |x2 − 1|),
Q.20 (about solving a vector equation in an extreme case), Q. 22 (about the
limits of the roots of a parametrised quadratic equation), Q.23 (which can be
done elegantly using complementary probability), Q.24 (about solving a tri-
angle with an imaginative hint for the answer), Q.25 (about a matrix whose
transpose can be expressed as a function of that matrix), Q.26 (based on sim-
ple properties of the cross product of vectors), Q.31 and 32 (about getting
a recurrence relation for the number of n-digit integers), Q.35 (where the
adjoint of a matrix but not the matrix is given), and Q.39 (involving limits
of sin πx and cos πx). Q.36 is a good question about coplanarity of two lines
but is marred by the confusing wording. Q. 40 would also have been a good
question about evaluating an expression in cos θ given the value of cos 4θ.
But the unnecessary twist given to it has made the question mathematically
wrong.
There are several questions where the work demanded is more than can be
reasonably expected given the time available. These include Q.7 (about find-
Z 1 2
ing a certain locus), Q.12 (about the estimation of the integral e−x dx),
0
Q. 14 (where a linear differential equation has to be solved and much more to
be done) and Q.36 (about coplanar lines). The work involved in Q.10 is also
substantial since it is not given that the second ellipse has its axes along the
coordinate axes. Those who simply assume this to be the case get unfairly
rewarded. In fact, there are several other questions where scruples do not
pay. These include Q.13 (where a relationship between two angles θ and φ is
given), Q.22 (asking for the limits of the roots of a parametrised quadratic
equation) and Q.35 (about the adjoint of a matrix). In such questions those
who are blissfully unaware that their claims need some justification are re-
warded. Q.27 (where we have to identify a member of a one-parameter family

74
of planes) is technically correct as it stands. But it also gives an unfair ad-
vantage to those who take the popular but incomplete parametrisation of
the family of planes because what they miss does not affect the answer. On
the contrary, those who start with the correct parametrisation will be baffled
that there are two correct answers to the problem and will have to resolve the
difficulty by resorting not so much to mathematics as to English grammar!
An element of luck is involved in Q.17 where the solution requires that you
guess a root of a certain cubic equation.
There are some questions inherently unsuitable to be asked as MCQ’s
because sneak methods are possible. These include Q.3 (about finding an
antiderivative of a given function) and Q.5 (about the ratio of two related
determinants). In Q. 32 (about finding a certain number) those who realise
that the number asked is a Fibonacci number get the answer without doing
any work.
Q.4 (about combinatorics) is a straight repetition of a JEE 1981 problem.
Q.28 (about evaluation of definite integrals of even and odd functions) has
little novelty in it since very similar questions have been asked many times
in the past. The paper-setters are helpless in this regard because of the
constraints on the credit to be awarded to a question. The charm of novel,
full length questions simply cannot exist in the present setting.
The paper-setters have given only a lip service to complex numbers by
asking just one, uninspiring question (Q.6). Areas like number theory, bino-
mial theorem (forget about binomial identities) and infinite series are totally
omitted. This could be due partly to the reduction in the number of ques-
tions to 20 in each paper. But another reason is the duplication of ideas
in several questions. There was no point in asking two questions (Q.11 and
Q.38) on conditional probability. One of them could have been replaced by
a good problem on probability involving some real life situation. The two
problems based on the second Fundamental Theorem of Calculus are also
almost replicas of each other.
Overall the paper is reasonable. Given the constraints they are subject
to, the paper-setters have managed to come up with a fair number of good
problems.

75