EDUCATIVE COMMENTARY ON

JEE 2022 ADVANCED MATHEMATICS PAPERS

(Revised on 6/10/2022)

Contents
Paper 1 2

Paper 2 36

Concluding Remarks 72

This is a commentary on the problems from the Mathematics section of

the JEE 2022 Advanced papers. The 2019 commentary was a departure
from the practice for the then preceding 16 years (from 2003 to 2018) where
the commentaries covered all questions. Instead, a few questions were omit-
ted from the commentary. There were two reasons for this change. First,
some problems are straightforward and similar problems had been asked in
the past. They lacked any significant educative content. Secondly, detailed
solutions to such problems are easily available on many websites and there
is not much point in duplicating them. For example, routine problems in
coordinate geometry and trigonometry were omitted.
Since 2020 this selective practice has been modified. The solutions to all
the problems are included. But when there is not much to comment, only
the answer and the important steps in the solution are given. Hopefully, the
readers can fill the gaps.
Readers who notice any errors, want to comment or give alternate solu-
tions are invited to send an email to the author at [email protected] or
send an SMS or a WhatsApp message to the author at 9819961036.
I thank Siddhesh Naik for clearing my misunderstanding about Q.17 in
Paper 1 and Vinayak Antarkar for pointing out a slip in the solution to Q.10
in Paper 2.
Unless otherwise stated, all the references made are to the author’s book
Educative JEE (Mathematics) published by Universities Press, Hyderabad.
 PAPER 1

Contents

Section - 1 (Numerical Answer Type) 2

Section - 2 (One or More correct Options Type) 14

Section - 3 (Matching Entries in Columns Type) 26

SECTION - 1
This section contains EIGHT questions each of which is to be answered
with a number with a numerical value (after truncating or rounding off) with
two places of decimals.
There are 3 marks for a correct answer and 0 in all other cases.

Q.1 Considering only the principal values of the inverse trigonometric func-
tions, the value of
s √ √
3 −1 2 1 −1 2 2π −1 2
cos 2
+ sin 2
+ tan
2 2+π 4 2+π π
is ......... .

Answer and Comments: (02.36). The complicated expressions of

the three angles actually are an implied hint. If the problem involved
simple expressions like tan−1 1 or sin−1 12 , it would be tempting to take
their separate numerical values (viz. π4 and π6 respectively) and add.
This is not a workable proposition in the present problem. The only
way out is to relate the three angles to each other in such a way that
adding them becomes a manageable task.
Call the given sum as S. None of the three terms contains a ‘famil-
√
iar’ angle. Even in the third term which is simplest, we have tan−1 π2
whose numerical value can hardly be found without trigonometric ta-
bles (or some other means such as calculators). So the right way is
not to calculate the three terms individually, but to relate them to
each other using suitable trigonometric identities so that the sum S

2
is some succinct expression. (Analogous situations sometimes arise in
evaluation of a sum of two integrals. Neither integral is manageable,
but because of some relationship in their integrands, their jaggy parts
cancel each other so to speak and the sum of the two integrals is easy to
evaluate. Sometimes only one integral is given and the other we supply
from our pocket.)
In the present problem, a simple clue is provided by the fact that if
tan θ = ab , where a, b are positive, then sin θ = √a2a+b2 and cos θ =
√
√ b . Taking θ as the last term of S, i.e. i.e. as tan−1 2
, we get
a2 +b2 π

√ s
2 2
sin θ = √ = (1)
2 + π2 2 + π2
π
and cos θ = √ (2)
2 + π2

Because of (1), we can express the first term of S in terms of θ. The

catch is that it involves the inverse cosine rather than the inverse sine.
But this is easy to fix using that for x ∈ [0, 1],
π
sin−1 x + cos−1 x = (3)
2
Therefore,
s s
−1 2 π 2 π
cos = − sin−1 = −θ (4)
2 + π2 2 2 + π2 2
Hence we can now simplify S partially as
√
3 π 1 −1 2 2π
S = ( − θ) + sin +θ
2 2 4 2 + π2
√
3π θ 1 −1 2 2π
= − + sin (5)
4 2 4 2 + π2
To simplify S further, we need to tackle the last term, which is actually
the hardest term of S. Ironically, here the hardship itself provides a
clue. We would like to simplify S to a level where it does not involve
θ. As long as it involves θ, calculating its numerical value is not an
easy task as already commented. So we pray that the last term in (5)

3
 reduces to 2θ plus some easy number.
√
Ideally, it should simplify to 2θ .
Equivalently, we hope that sin−1 22+π2π2 comes out to be 2θ.
To test√ if this optimism is justified, we calculate sin 2θ from θ which is
tan−1 π2 . Using (1) and (2) we have,

sin 2θ = 2 sin√θ cos θ

2 π
= 2√ √
2 + π 2 + π2
2
√
2 2π
= (6)
2 + π2
So, S indeed simplifies to 3π
4
. Taking π approximately as 3.1415, this
comes to 2.356125. After rounding the answer in the desired format is
02.36.

A simple but very well designed problem. The trigonometric iden-

tities needed are simple and standard. but the skill lies in deciding
which ones of them to apply. And the hardship of the expressions itself
provides a clue. If this question were asked in the integer type answer
format, it would have asked to give the value of something like 4S
π
. But
that would have given an unwarranted hint that S is a rational multiple
of π and diluted the problem.

Q.2 Let α be a positive real number. Let f : IR −→ IR and g : (α, √ ∞) −→ √ IR

2 log e ( x − α)
be the functions defined by f (x) = sin( πx
12
) and g(x) = √
x
√
α
.
loge (e − e )
Then the value of lim+ f (g(x)) is ...... .
x→α

Answer and Comments: (00.50). A classic example of how an

unrelated appendage is added to a main problem. By the chain rule
(for limits), if we let y = g(x) and β = lim+ g(x), then the desired
x→α
limit is lim f (y) which is simply f (β) as the function f is continuous
y→β
everywhere and so as soon as we know β, the answer is sin( πβ
12
).
The real task is to evaluate β, which is a limit in an indeterminate√form.
The calculations can be simplified with a substitution. Let c = α. If

4
 √
we call x as u, then u → c+ as x → α+ .
2 loge (u − c)
β = lim+ (1)
u→c loge (eu − ec )
To simplify further, we put h = u − c. Then eu − ec = ec+h − ec =
ec eh − ec = ec (eh − 1). Hence
loge h
β = 2 lim+
h→0 loge ec (eh − 1)
loge h
= 2 lim+ (2)
h→0 c + loge (eh − 1)

As h → 0+ , loge h → −∞. Further since eh − 1 also tends to 0+ ,

+
its
√ log tends to −∞ tooh as h → 0 . As c is some fixed number (viz.
+
α), we have c + loge (e − 1) → −∞ as h → 0 . So, in (2), both the
numerator and the denominator tend to −∞. To evaluate the limit, it
is convenient to recast it as
1
β = 2 lim+ c h
h→0
loge h
+ loglog
e (e −1)
eh
2
= (3)
c loge (eh − 1)
lim + lim+
h→0+ loge h h→0 loge h
√
The first limit in the denominator is 0 because c = α is a fixed
number while loge h → −∞ as h → 0+ . The second limit in the
denominator is in an indeterminate form. If we were not taking logs,
0 eh − 1
then it would have been in the 0 indeterminate form lim+ and
h→0 h
can be evaluated without l’ôpital’s rule simply as the derivative of the
expoenential function eh at h = 0. It would come to 1. But because of
−∞
the logarithms, the indeterminate form is and we need l’ôpital’s
−∞
rule. Applying it,
loge (eh − 1) heh
lim+ = lim+
h→0 loge h h→0 eh − 1
h
= e0 lim+ h
h→0 e − 1
= 1×1=1 (4)

5
 So, at long last, we get
2
β= =2 (5)
0+1
Hence, finally, the desired limit equals f (β) = sin( 2π
12
) = sin( π6 ) = 1
2
=
00.50.

Normally, one would expect that the limit depends on α and

some candidates may be puzzled that it is independent of α. But the
discerning ones will realise that the very fact that the question asks for
a single, numerical answer indicates that β is independent of α.
Some candidates may sloppily ignore the logarithms in the second
eh − 1
limit in the denominator in (3) and conclude that it equals lim .
x→0 h
Luckily, they will go scot free since both the limits happen to be equal.
The constant 2 in the definition of g(x) has no role except to make
π
the answer come out as sin( π6 ) instead of sin( 12 ). The numerical value
of the latter can be found from that of sin( π6 ) (which is 21 ) by solving a
quadratic. But that would make the problem more laborious.

Q.3 In a study about a pandemic, data of 900 persons was collected. It was
found that
190 persons had symptom of fever,
220 persons had symptom of cough,
220 persons had symptom of breathing problem,
330 persons had symptom of fever or cough or both,
350 persons had symptom of cough or breathing problem or both,
340 persons had symptom of fever or breathing problem or both,
30 persons had all three symptoms (fever, cough and breathing prob-
lem).
If a person is chosen randomly from these 900 persons, then the prob-
ability that the person has atmost one symptom is ........ .

Answer and Comments: (00.80). Mathematically, this problem

is not very different from Q.13 in Papaer 1 of JEE Advanced 2021.
In both the problems, there is a set S and three subsets A, B, C of
it. They, along with their complements (w.r.t. S) decompose S into

6
8(=23) mutually disjoint subsets of the form X ∩ Y ∩ Z where X is
either A or its complement A′ , Y is either B or B ′ and Z is either C
or C ′ . These eight subsets are the prototypes and any of the subsets
of S which can be expressed in terms of A, B, C can be expressed
uniquely as the union of some of these eight subsets. This can be done
systematically by logic but often by inspection when the subsets and
their intersections are shown in a Venn diagram. (The areas of the
regions are not necessarily proportional to the subsets they represent.)
We show below the Venn diagram for the present problem. The
ambient set S is the sample space, consisting of all 900 persons sur-
veyed, F, C, B are, respectively, subsets of persons with fever, cough
and breathing problems. Small letters denote the numbers of persons
in the respective subsets.

F
C

a z b
S

y 30

x
c u

The data gives a system of equations in these seven unknowns

a, b, c, x, y, z and u. For example, from |F | = 120, we get

a + y + z = 120 − 30 = 90 (1)

It is an easy (albeit tedious) exercise to write down and solve all these
equations. For the final answer, we need a + b + c + u which comes out
to be 240 + 480 = 720. As the sample size is 900, the probability P
equals 720
900
= 00.80.

7
 A very straightforward problem, especially after a very similar ques-
tion last year (which was slightly less trivial since the data only gave
some inequalities and was not sufficient to determine all 8 variables).
2 + 3z + 4z 2
Q.4 Let z be a complex number with a non-zero imaginary part. If
2 − 3z + z 2
is a real number, then the value of |z|2 is ..... .

Answer and Comments: (00.50). A brute force method to tackle a

problem in complex algebra is to write z as x + iy and resolve the given
expression in the form u + iv where u, v are real valued functions of x
and y. But if the expression involves a ratio, expressing it in the form
u + iv becomes complicated. So we try something else.
A real number remains real if any other real number is added or sub-
tracted from it and the converse is also true. Call the given number as
w. We notice that w is a ratio of two quadratic polynomials in z with
the same leading coefficient. So, if we subtract 1 from it, the quadratic
term in the numerator will be gotten rid of. So,
2 + 3z + 4z 2 6z
w−1 = 2
−1 = (1)
2 − 3z + z 2 − 3z + 4z 2
6
is real. We can recast it as 2 −3+4z , divide by 6 (which is a real number)
z
and use that the reciprocal of a real number is also real. That gives
that 4z − 3 + z2 is real. Finally, adding 3 gives that 4z + z2 is real and
division by 2 gives 2z + z1 is real.
So far, we have reduced the given number w to another, simpler number
which is real. Now for the first time we write z as x + iy. Then
1
z
= xx−iy
2 +y 2 . Therefore,

1 x − iy
2z + = 2(x + iy) + 2 (2)
z x + y2
Since this is a real number, its imaginary part vanishes. That is,
y
2y − 2 =0 (3)
x + y2
We are given that the imaginary part of z, viz. y is non-zero. Therefore,
1
we get x2 +y 2
2 = 2 or, x + y
2
= 12 . But |z|2 is precisely x2 + y 2. So,
|z|2 = 21 = 0.50.

8
 Instead of obtaining (3) by resolving into real and imaginary parts,
we could have applied the simple observation that a complex number
z is real if and only if it equals its complex conjugate z. So, once we
get that 2z + 1z is real, we can write

1 1
2z + = 2z + (4)
z z
which simplifies to
1 1
2(z − z) = −
z z
z−z
=
zz
z−z
= (5)
|z|2

We are given that z is not real. Hence z 6= z. Canceling z − z we get

|z|2 = 21 , the same answer as before.
In fact, we could have applied this characterisation of reality right
at the start and written w = w. But this would be long winded.
A basically simpe problem, which is an excellent test of the ability
to simplify a problem through a series of elementary operations.

√ z denote the complex conjugate of a complex number z and i =

Q.5 Let
−1. In the set of complex numbers, the number of distinct roots of
the equation z − z = i(z + z 2 ) is ..... .

Answer and Comments: (04.00). Yet another problem where it

would be foolish to resolve into real and complex parts. Instead, we
observe that the equation can be rewritten as

(1 − i)z = (1 + i)z 2 (1)

This can be simplified if we notice that the ratio of the coefficients can
be simplified. Indeed,

1+i (1 + i)2 2i
= = =i (2)
1−i (1 − i)(1 + i) 2

9
 As a result, the given equation is reduced to

z = iz 2 (3)

Equating the absolute values of both the sides,

|z| = |i||z|2 = |z|2 (4)

So either |z| = 0 or |z| = 1. The first possibility gives z = 0 as a

solution. If the second possibility holds, then multiplying both the
sides of (3) by z gives

1 = iz 3 (5)

Multiplying both the sides by −i, we have

z 3 = −i (6)

In other words, z is a cube root of the non-zero complex number −i. For
any positive integer n, every non-zero complex number has n distinct
n-th roots. They can be written explicitly by taking the number in
its polar form (i.e. as reiθ where r > 0 and θ is real), But in the
present problem, we are interested only in the number of roots and not
in identifying them. So, the given equation has four distinct roots, viz.
0 and the three cube roots of −i.

Although not asked, in the present problem it is easy to explicitly

identify the roots. Note that i is a cube root of −i. Hence the three
solutions of (6) are i, iω and iω 2 , where ω is a complex cube root of
unity. Along with 0 you get the four solutions of the given equation.
A very good problem based on elementary simplifications and the
fact that every non-zero complex number has n distinct n-th roots.

Q.6 Let l1 , l2 , . . . , l100 be consecutive terms of an arithmetic progression with

common difference d1 , and let w1 , w2 , . . . , w100 be consecutive terms
of another arithmetic progression with common difference d2 , where
d1 d2 = 10. For each i = 1, 2, . . . , 100, let Ri be a rectangle with length
li , width wi and area Ai . If A51 − A50 = 1000, then the value of
A100 − A90 is ...... .

10
Answer and Comments: (18900.00). If the lenghts as well as the
widths of a sequence of rectangles are in G.P.’s, so are their areas as
follows immediately from the definition. But this is not so when the
lengths form an A.P. and so do the widths.
The easiest start is to express each li in terms of l1 and d1 and
similarly each wi in terms w1 and d2 . This will allow us to express Ai in
terms of l1 , w1 , d1 and d2 . Since li = l1 +(i−1)d1 and wi = w1 +(i−1)d2 ,
we have

Ai = li wi = l1 w1 + (i − 1)(l1 d2 + w1 d1 ) + (i − 1)2 d1 d2 (1)

Writing a similar equation for Aj where j < i and subtracting it from

(1) gives

Ai − Aj = (i − j)(l1 d2 + w1 d1 ) + (i − j)(i + j − 2)d1d2 (2)

Taking i = 51, j = 50 and using that d1 d2 = 10, we get

1000 = (l1 d2 + w1 d1 ) + 990 (3)

which gives

l1 d2 + w1 d2 = 10 (4)

Similarly, we can write down an expression for A100 − A90 . Its value is
not given. But from (4) we can determine it as 18900.

An unusual but a little laborious problem on progressions. The

calculations, although straightforward, are prone to numerical errors.
A discerning candidate may think that the data is insufficient. There
are four variables, l1 , d1 , w1 and d2 . So we need four equations to deter-
mine them. A51 − A50 = 1000 is one such equation. We are also given
that d1 d2 = 10. With only two equations in four unknowns, we cannot
determine them uniquely. A closer look, however, shows why the infor-
mation given is sufficient. We are only concerned with the differences
of the A’s. From (2), we see that they depend only on d1 d2 (which is
given as 10) and l1 d2 + w1 d1 (which can be found from A51 − A50 ). So,
effectively, there are only two unknowns, d1 d2 and l1 d2 + w1 d1 .

11
Q.7 The number of 4-digit integers in the closed interval [2022, 4482] formed
by using the digits 0, 2, 3, 4, 6, 7 is ..... .

Answer and Comments: (569.00). Let S be the set of all integers in

[2022, 4482] which satisy the given restrictions. For each such number,
the starting digit must be either 2, 3 or 4. Each of the remaining
three digits can have any of the six values 0, 2, 3, 4, 6, 7. To find |S|,
we construct a superset T of it and first find T . Let T be the set of
all numbers which satisfy the same restrictions as those in S (as far
as their digits are concerned, viz. they have to come from 0, 2, 3, 4, 6
and 7 with the first digit from 2, 3, 4) except that they need not be in
[2022, 4482]. Then S ⊂ T . Moreover,

|T | = 3 × 6 × 6 × 6 = 648 (1)

To find |S|, we have to subtract |T − S| from this. An element can be

in T − S when it is either less than 2022 or greater than 4482. The two
possibilities cannot occur together. So we separately count numbers of
each type and add.
If n ∈ T and n ≤ 2021, then the first three digits of n must be either
200 or 202 since 1 is not allowed as a possible digit. In the first case,
the last digit can be any of the six numbers 0, 2, 3, 4, 6, 7. In the
second case, the only possible number is 2020. Hence there are, in all,
7 numbers in T which are less than 2022.
Now suppose that n = n1 n2 n3 n4 ∈ T and n > 4482. Then n1 must be
4. Also n2 ≥ 4. The possibility n2 = 4 is ruled out because in that
case, n3 will have to be at least 8, which is not the case for any number
in T . So, n2 is either 6 or 7. In either case, n3 and n4 can assume
any of the six values 0, 2, 3, 4, 6, 7 independently. Hence the number of
integers in T that exceed 4482 is 2 × 6 × 6 = 72.
Summing up, we have shown that

|T − S| = 7 + 72 = 79 (2)

By complementary counting,

|S| = |T | − |T − S| = 648 − 79 = 569 (3)

12
 So, this is the number of integers in [2022, 4482] which can be formed
using only the digits 0, 2, 3, 4, 6 and 7.

A good problem on complementary counting since the cardinality of

the complement needs to be found carefully. Although we have given
the reasoning in detail, it does not take much time to conceive it and
once it is conceived, the counting can be completed even with a rough
sketch, requiring hardly any words. So this problem is excellent where
the very correctness of the answer is a sure proof that the candidate
has thought it out correctly. Such problems are ideally suited for ex-
aminations where the candidates have to write only the final answer in
the form of a number.
There are two restrictions on the numbers n we are looking for, viz.
the digital restriction, viz. that the first digit is 2, 3 or 4 and the other
three come from {0, 2, 3, 4, 6, 7} and secondly the range restriction, viz.
that they lie in the closed interval [2022, 4482]. We constructed the su-
perset T of S by dropping the second restriction. What if we construct
T by dropping the digital restriction only? Then T would be the closed
interval [2022, 4482]. This is very tempting because in this case, |T | is
very easy to find. It is simply 4482 − 2021 = 2461. But finding |T − S|
is not an easy task since the set T − S will have to be decomposed into
four subsets depending upon the occurrence of the forbidden digits 1,
5, 8 and 9. As these subsets are overlapping, the principle of inclusion
and exclusion will have to be applied and the calculations will be hor-
rendous. This shows that in complementary counting, sometimes art
lies in choosing the superset. (Sometimes, the choice of the superset
is very natural. But then the real art lies in finding the cardinality of
the complement, see Andre’s solution in Comment No.6 of Chapter 1
for the problem of finding the number of balanced arrangements of n
pairs of parenteses.)
Q.8 Let ABC be the triangle with AB = 1, AC = 3 and 6 BAC = π2 . If
a circle of radius r > 0 touches the sides AB, AC and also touches
internally the circumcircle of the triangle ABC, then the value of r is
..... .

Answer and Comments: (00.84). It is convenient to introduce

coordinates with axes lying along the sides. Then A is the origin and

13
 we take B = (1, 0) and C = (0, 3). As △BAC is right angled at A, its
circum-centre, say P , is the midpoint of BC, viz. ( 12 , 23 ) and radius is
1
√ 1
√
2
9 + 1 = 2
10. Let Q be the centre of the circle of radius r which
touches the sides AB, AC. Then Q = (r, r). If this circle touches the
circumcircle internally, then the difference between their radii equals
the distance between their centres. This gives an equation for r, viz.
s
1√ 1 3
10 − r = (r − )2 + (r − )2 (1)
2 2 2
Squaring and simplifying gives a quadratic in r, viz.
√
r 2 = 4r − r 10 (2)
√
As there is no√constant term and r 6= 0, we are left with r = 4 − 10 ≈
0.84, taking 10 as 3.16.

A simple geometry problem. A figure is hardly necessary since

the key idea is merely a simple property about the distance between
touching circles and the numerical dada is easy to find.

SECTION -2
This section contains SIX (06) questions.
Each question has FOUR options (A), (B), (C) and (D). ONE OR MORE
THAN ONEof these four option(s) is (are) correct answer(s).
For each question, choose the option(s) corresponding to (all) the correct an-
swer(s). When no incorrect answer is chosen, 4 points are given for choosing
all correct answers, otherwise some partial credit is given.
e (loge x)1/2
Z
Q.9 Consider the equation dx = 1, a ∈ (−∞, 0) ∪
1 x(a − (loge x)3/2 )2
(1, ∞). Which of the following statements is/are TRUE ?

(A) No a satisfies the above equation

(B) An integer a satisfies the above equation
(C) An irrational number a satisfies the above equation
(D) More than one a satisfy the above equation

14
 Answer and Comments: (C,D). Yet another problem where two
unrelated themes are patched together. The problem is really about
evaluating an integral. But a parameter a is added to the integrand to
make the integral a function of a, say I(a). Then I(a) is set equal to
1 (an essentially arbitrary choice) and then solutions of the equation
I(a) = 1 are asked.
Coming to the integral I(a), the substitution u = loge x)3/2 reduces it
to
2Z 1 1
I(a) = du
3 0 (a − u)2
2 1
= ( )|10
3 a−t
2 1 1
= ( − )
3 a−1 a
2
= (1)
3a(a − 1)

So the equation I(a) = 1 is merely the quadratic equation

3a2 − 3a − 3 = 0 (2)
√
which can be solved explicitly to get two solutions, viz. 3±6 33 . The
roots are distinct and both are irrational. So (C) and (D) hold.

An artificial combination of two topics, evaluation of a definite inte-

gral and solving a quadratic. The problem would have been interesting
if the truth of the statements could be tested without evaluating the
integral explicitly, for example, by finding only upper and lower bounds
for it.

Q.10 Let a1 , a2 , a3 , ..... be an arithmetic progression with a1 = 7 and common

difference 8. Let T1 , T2 , T3 , ........ be such that T1 = 3 and Tn+1 −Tn = an
for n ≥ 1. Then, which of the following is /are TRUE ?
20
X
(A) T20 = 1604 (B) Tk = 10510
k=1
X30
(C) T30 = 3454 (D) Tk = 35610
k=1

15
Answer and Comments: (B,C). The paper-setters’ passion for
adding appendages to the main theme of the problem continues. In
essence the problem amounts to solving the recurrence relation
Tn+1 − Tn = an (1)
for n ≥ 1 with T1 = 3. an is given as the n-th term of an A.P. with
the first term 7 and common difference 8. Anybody can get from this
that an = 7 + 8(n − 1) = 8n − 1 for all n ≥ 1. At least a candidate
who has cleared JEE (Mains) can safely be assumed to know this. So
nothing is gained by introducing an in this manner. There is, in fact,
a danger that a slight numerical slip at this starting point will vitiate
the subsequent work.
Anyway the recurrence relation (1) now become
Tn+1 − Tn = 8n − 1 (2)
for every n ≥ 1. Our job is to express Tn directly as a function of n.
For this we apply (2) again and again for indices 1, 2, . . . , n − 1 and get
T2 − T1 = 8 × 1 − 1
T3 − T2 = 8 × 2 − 1
.. .
. = ..
Ti − Ti−1 = 8(i − 1) − 1
.. .
. = ..
Tn − Tn−1 = 8(n − 1) − 1
If we add these n − 1 equations, most of the terms cancel and we are
left with
Tn − T1 = 8 × (1 + 2 + . . . + n − 1) − (n − 1)
= 4n(n − 1) − (n − 1) = 4n2 − 5n + 1 (3)
Since T1 is given as 3, we finally have
Tn = 4n2 − 5n + 4 (4)

Testing whether (A) and (C) hold is now a matter of sunstitution.

T20 = 1600 − 1000 + 4 = 1504 which makes (A) false. T30 = 3600 −
150 + 4 = 3454 whence (C) is true.

16
 For (B) and (D), we need to add the terms of the sequence {Tn }.
n n
4k 2 − 5k + 4
X X
Tk =
k=1 k=1
4n(n + 1)(2n + 1) 5n(n + 1)
= − + 4n (5)
6 2
We need not simplify. Putting n = 20, the sum in (B) is 10510. Simi-
larly, putting n = 30, the sum in (D) is 35615. So, (B) is true but (D)
is false.

The only interesting part is to solve (2) by applying it repeatedly

and adding the resulting equations. This works because one term in the
L.H.S. of each equation gets cancelled with one term in the next equa-
n
X 1
tion. This is also what happens for telescopic series (e.g.
k=1
k(k + 1)
1 1 1 1
can be summed as 1 − n+1 by writing k(k+1) as k − k+1 .) For this rea-
son the method adopted in arriving at (4) is called telescoping the
equation. Of course, one can prove by induction on n, that (4) holds
for all n ≥ 1. But that is a verification and not quite the same as arriv-
ing at the answer. Yet another approach is to assume that Tn is some
quadratic polynomial in n, say Tn = An2 + Bn + C, and determine
A, B, C from the values of T1 , T2 , T3 which can be obtained by a direct
calculation as 3, 10, 25 respectively.
The subsequent work after deriving a formula for Tn is sheer drudgery.
Hardly anything new is tested (except possibly the knowledge of the
formula for the sum of the squares of the integers from 1 to n). On the
contrary the calculations involved are time consuming, retetitious and
prone to numerical slips. The draconian policy of awarding −2 points
even for a single slip makes the good work in deriving (4) nugatory and
such a candidate is worse off than someone who simply skips the ques-
tion. It would have been far more equitable to relegate this question to
Section I and ask only the numerical value of, say, T1 + T2 + . . . + T20 .
Q.11 Let P1 and P2 be two planes given by
P1 : 10x + 15y + 12z − 60 = 0,
P2 : −2x + 5y + 4z − 20 = 0.
Which of the following straight lines can be an edge of some tetrahedron
whose two faces lie on P1 and P2 ?

17
 x−1 y−1 z−1 x−6 y z
(A) = = (B) = =
0 0 5 −5 2 3
x y−4 z x y−4 z
(C) = = (D) = =
−2 5 4 1 −2 3

Answer and Comments: (A,B,D). A slight verbal correction is due

in the last part of the question. An edge is a segment, a line is infinite
in both directions. So it would be better to say that a line contains an
edge rather than is an edge.
Every tetrahedron has 4 faces and six edges. Given any pair, say F1
and F2 of faces, the edges behave in three different ways w.r.t. this
pair. For example consider the tetrahedron with vertices A, B, C, D
shown below. Let F1 be the ‘front’ face consisting of the triangle ACD
and F2 the ‘back’ face consisting of the triangle ABD. Let P1 and P2
be the planes containing F1 and F2 respectively and let L be the line
of their intersection shown in thick.

F2 B
L
F
1

A C

Then the edges can be classified as follows.

(i) The edge AD is common to the faces F1 and F2 and is contained

in the line L
(ii) The edges CA and CD lie completely in the plane P1 and do not
meet P2 except at the points A and D respectively
(iii) Similarly, the edges BA, BD lie completely in P2 and meet P1 only
at A and D respectively.
(iv) The edge BC is skew to the line L. It intersects P1 only at the
point C and P2 only at the point B,

18
The four possibilities are obviously disjoint. In the present problem we
are not given any particular tetrahedron. Instead, we are given two
planes P1 and P2 and for each of the four given lines, we are asked to
find which possibility holds, viz. whether the line is contained in both
P1 and P2 , only in one of them (but intesecting the other) or contained
in neither plane but still intersects both.
Call the lines in the four options (A), (B), (C), (D) as L1 , L2 , L3 , L4
respctively. Each is given with its direction numbers and one point
lying on it. We first determine the direction of L = P1 ∩P2 . L is parallel
to ~n1 ×~n2 where ~n1 , ~n2 are normals to P1 and P2 respectively. From the
equations of P1 , P2 , we may take ~n1 = (10, 15, 12) and ~n2 = (−2, 5, 4).
So the direction numbers of ~n1 × ~n2 and hence of L are (proportional
to) 0, −4, 5.
Thus none of the lines L1 , L2 , L3 , L4 is even parallel to L, forget about
being an edge, if possibility (i) is to apply. But there could be other
reasons to qualify to be an edge of some tetrahedon with faces conained
in P1 and P2 . In each case, we have to check whether the line Li , i =
1, 2, 3, 4 is contained in the plane Pj , j = 1, 2 or merely intersects it at
one point. This can sometimes be done by inspection. But there is
a also a systematic procedure for this. Suppose that the line is given
parametrically as

x = u1 t + v1 , y = u2 t + v2 , z = u3 t + v3 , t ∈ IR (1)

and the equation of the plane is

ax + by + cz + d = 0 (2)

Substituting from (1) into (2) will give a linear equaion in t of the form

At + B = 0 (3)

If A 6= 0 , there is a unique solution and the line cuts the plane in one
point. If both A, B vanish, it is completely contained in the plane. If
A = 0 and B 6= 0, then the line is parallel to the plane.
Take, for example, the line L1 in option (A). Its guiding vector is (0, 0, 5)
and (1, 1, 1) is a point on it. So its parametric equations are

x = 1, y = 1, z = 5t + 1 (4)

19
 Putting this into the equation of P1 ,

10 + 5 + 12(5t + 1) − 60 = 0 (5)

which has a unique solution. Hence L1 meets P1 at the point (1, 1, 15 4

).
By a similar argument, it intersects P2 too, but at a point different from
(1, 1, 15
4
). So L1 is a line which is skew to L and there is a tetrahedron
with faces contained in P1 , P2 with L1 as an edge, Therefore, (A) is
true.
By similar arguments, (B) and (D) are true. However, with (C), the
parametric equations of L3 are

x = −2t, y = 5t + 4, z = 4t (6)

If we substitute these into the equation of P1 we get

−20t + 75t + 60 + 48t − 60 = 0 (7)

which has a unique solution t = 0. So, L3 meets P1 at the point (0, 4, 0).
Its point of intersection with P2 is given by

4t + 25t + 20 + 16t − 20 = 0 (8)

which also gives t = 0 and (0, 4, 0) as the point of intersection. But

that means L3 meets both P1 and P2 at the same point. A single point
cannot be an edge. Hence (C) is false.

The conceptual part of the problem is quite novel. Although

tetrahedra are commonly studied, the classification of their edges w.r.t.
a particular pair of faces is generally not covered. A candidate has to
figure it out. Unfortunately, after doing this correctly, there is a highly
repetitious, long numerical work which is very prone to numerical slips.
So, as in the last question, unless a candidate is very careful, he is likely
to get a negative score despite tackling the conceptual part correctly.
Q.12 Let S be the reflection of a point Q with respect to the plane given by

~r = −(t + p)î + tĵ + (1 + p)k̂

where t, p are real parameters and î, ĵ, k̂ are the unit vectors along the
three positive coordinate axes. If the position vectors of Q and S are

20
10î+15ĵ +20k̂ and αî+β ĵ +γ k̂ respectively, then which of the following
is/are TRUE ?
(A) 3(α + β) = −101 (B) 3(β + γ) = −71
(C) 3(γ + α) = −86 (D) 3(α + β + γ) = −121

Answer and Comments: (A,B,C). The manner in which the plane,

say P , is given is (perhaps intentionally) clumsy. By specifying that
î, ĵ, k̂ are unit vecrors along the three coordinate axes, the paper-setters
have given a hint that the first step is to cast the plane in a coordinate
form, viz. as the solution set of a linear equation in x, y, z. This is
very easy to do by eliminating the parameters t and p from the three
equations

x = −(t + p), y = t, z = 1 + p (1)

Simply adding them gives the equation of P as

x+y+z =1 (2)

The points Q and S are reflections of each other in this plane. They
are given in terms of their position vectors. In more familiar terms this
means Q = (10, 15, 20) and S = (α, β, γ). As they are reflections of
each other, w.r.t. P , two conditions hold:
(i) the midpoint of the segment QS lies on P
(ii) the line QS is normal to P
From (i) we get
α + 10 β + 15 γ + 20
+ + =1 (3)
2 2 2
which simplifies to

α + β + γ = −43 (4)

and shows that (D) is false.

To make use of (ii), the vector (1, 1, 1) is normal to P . Hence the line
QS is parallel to (1, 1, 1) which gives

α − 10 = β − 15 = γ − 20 = u (5)

21
 for some u ∈ IR. Adding, we get

3u = α + β + γ − 45 = −88 (6)

because of (4). Hence u = − 88

3
. We can avoid fractions by taking

3α = 3u + 30 = −58 (7)
3β = 3u + 45 = −43 (8)
and 3γ = 3u + 60 = −28 (9)

By direct substitutions, we see that (A),(B),(C) are all true.

A simple problem once a candidate realises that it is a problem in

coordinate geometry garbed as a problem in vectors. The computations
are reasonable and there are some tacit hints. For example, the very
fact that in all options a factor of 3 occurs suggests that this is done
to avoid fractions.

Q.13 Consider the parabola y 2 = 4x. Let S be the focus of the parabola. A
pair of tangents drawn to the parabola from the point P = (−2, 1) meet
the parabola at P1 and P2 . Let Q1 and Q2 be points on the lines SP1
and SP2 respectively such that P Q1 is perpendicular to SP1 and P Q2
is perpendicular to SP2 . Then, which of the following is/are TRUE ?
√
3 10
(A) SQ1 = 2 (B) Q1 Q2 = 5
(C) P Q1 = 3 (D) SQ2 = 1

Answer and Comments: (B,C,D). A diagram would be helpful

to understand the problem. But one of the strongest justifications of
coordinates is that they convert a problem in geometry to a problem in
algebra through some formulas. Diagrams may help in the derivation
of these formulas. But once they are obtained, they can be applied
without diagrams.
A typical point (x, y) on the parabola y 2 = 4x is (t2 , 2t) for some t ∈ IR.
dy/dt 2 1
The tangent at this point has slope = . It passes through
dx/dt 2t t
P = (−2, 1) if and only if
2t − 1 1
2
= (1)
t +2 t

22
 which simplifies to a quadratic in t,

t2 − t − 2 = 0 (2)

The roots are 2 and −1. Without loss of generality, we let P1 correspond
to 2 and P2 to −1. Then

P1 = (4, 4), P2 = (1, −2) (3)

Q1 , Q2 are the feet of the perpendiculars from P to the lines SP1 and
SP2 respectively. Since S = (1, 0) we get the equations of SP1 and SP2
and then determine Q1 , Q2 . They come out as
2 4
Q1 = ( , − ), Q2 = (1, 1) (4)
5 5
Now that we know the coordinates of all the points P, S, Q1 and Q2 , all
the four options can be tested simply using the distance formula. (B),
(C) and (D) come out to be true and (A) false.

A chain of computational tidbits assembled together without any

purpose. For the final answer the distance formula has to be applied
four times and even one slip renders all the work useless.

Q.14 Let |M| denote the determinant of a square matrix M. Let g : [0, 2ı ] −→
IR be the function defined by
π
q r
g(θ) = f (θ) − 1 + f( − θ) − 1
2
where
cos(θ + π4 ) tan(θ − π4 )
   
 1 sin θ 1   sin π 
1   
π

f (θ) =  − sin θ 1 sin θ  +
 
sin(θ − 4 ) − cos π2 loge ( π4 ) 
2 

−1 − sin θ 1 cot(θ + π4 ) loge ( π4 ) tan π
  
  

Let p(x) be a quadratic polynomial whose roots are the maximum

√ and
the minimum values of the function g(θ), and p(2) = 2 − 2. Then
which of the following is/are TRUE ?
√ √ √ √
(A) p( 3+4 2 ) < 0 (B) p( 1+34 2
)>0 (C) p( 5 2−1
4
) >0 (D) p( 5−4 2 ) < 0

23
Answer and Comments: (A, C). It was commented above that
the last problem was a hotchpotch of computational tidbits assembled
together. But at least those tidbits were coherent in that they all were
from coordinate geometry. Nothing can match the present problem
in terms of the diversity of the topics touched. It has matrices and
determinants, trigonometric identities, not to mention logarithms too.
In case that is not enough there are maxima and minima, a quadratic
polynomial and in case inequalities feel hurt, they are there in all four
options along with surds. It has something for everybody. One wonders
why integrals are omitted!
The right approach in such problems begins by focusing on the relevant
details. Matrices are really not involved. The notation for a determi-
nant is very standard and people study them even before studying
matrices. So we concentrate on the two determinants that appear in
the definition of f (θ). By a direct expansion, the first determinant
comes to be 2 + 2 sin2 θ.
As for the second determinant, a direct evaluation can hardly be
called straightforward. So the key to success is to suspect that there
is a deliberate purpose behind giving the entries in such a complicated
form. Sure enough, the diagonal entries all vanish. Among the re-
maining entries, the entries loge ( π4 ) and loge ( π4 ) cannot be evaluated
easily. But even without evaluating, they are negatives of each other.
Moreover they are located symmetrically w.r.t. the diagonal. This
leads to checking whether the same holds for the other two pairs of
entries symmetrically located w.r.t. the diagonal. Consider, for exam-
ple, cos(θ + π4 ) and sin(θ − π4 ). Upon expansions they are cos θ−sin
√
2
θ
and
sin θ−cos θ
√
2
. So, they are indeed negatives of each other.
The only remaining pair is tan(θ − π4 ) and cot(θ + π4 ). The first one
tan θ − 1 1 1
equals . Writing the second one as π = 1+tan θ we
tan θ + 1 tan(θ + 4 ) 1−tan θ
see that they are negatives of each other too.
The search is over. The second determinant is that of a skew sym-
metric matrix of an odd order. So it vanishes and we get a succint
expression for f (θ) viz.
1
f (θ) = (2 + 2 sin2 θ + 0) = 1 + sin2 θ (1)
2
24
We now immediately get g(θ).
π
q r
g(θ) = f (θ) − 1 + f ( − θ) − 1
2
q √
= sin2 θ + cos2 θ
= sin θ + cos θ (2)

where positive signs are justified as θ ∈ [0, π2 ]. Over this interval the
√
maximum and the minimum √ of g(θ) are 2 and 1 respectively as can
be seen by writing g(θ) as 2 sin(θ + π4 ) which is maximum when θ = π4
and minimum when θ = 0 or θ = π2 .
That gives
√
p(x) = c(x − 2)(x − 1) (3)
√
for some constant c. Since p(2) = 2 − 2, we get
√ √
c(2 − 2) = 2 − 2 (4)

which determines c as 1. Thus√p(x) is a quadratic which is positive √

for x ∈ (−∞, 1) and for x ∈ ( 2, ∞) and negative for 1 < x < 2.
For each of the four options we have to decide√where √ the given number
3+ 2
falls. Thus, in (A) we have to check if 1 < 4 < 2, or equivalently,
√ √ √
whether 4 < 3 + 2 < 4 2. This is true because 1 < 2 < 2. Hence
(A) is true. Similar calculations show that (C) is true, but (B) and (D)
are false.

An extremely long problem. The heart of the solution is in recognis-

ing that the second determinant in f (θ) vanishes by skew-symmetry.
The rest of the problem is a drudgery. And once again, even after do-
ing it correctly, choosing the correct options involves a lot of repetitious
work with inequalities and even a single slip is costly. (We commented
above that the question did not involve integrals. The paper-setters
could have accommodated them by replacing two terms of the second
π
Z π/2
determinant by a mutually canceling pair, say cos(θ + )dθ and
0 4
π
Z π/2 Z 2 Z 1
2 2
sin(θ − )dθ or, ex dx and ex dx.)
0 4 1 2

25
 SECTION -3

This section contains FOUR (04) Matching List Sets.

Each set has TWO lists: List-I and List-II
List-I has four entries (I), (II), (III) and (IV) and List-II has five entries.
FOUR options are given in each Multiple Choice Question based on List-
I and List-II and ONLY ONE of these four options satisfies the condition
asked in the Multiple Choice Question.
There are 3 points for choosing the correct option, 0 for no answer and
−2 in other cases.

Q.15 Consider the folloing lists:

List-I List-II

(I) {x ∈ [− 2π
3
, 2π3
] : cos x + sin x = 1} (P) has two elements
√
(II) {x ∈ [− 5π ,
18 18
5π
] : 3 tan 3x = 1} (Q) has three elements
√
(III) {x ∈ [− 62π , 6π ] : 2 cos(2x) = 3} (R) has four elements
5 5

(IV) {x ∈ [− 7π , 7π ] : sin x − cos x = 1} (S) has five elements

4 4

(T) has six elements

The correct option is :

(A) (I) → (P); (II) → (S); (III) → (P); (IV) → (S)

(B) (I) → (P); (II) → (P); (III) → (T); (IV) → (R)
(C) (I) → (Q); (II) → (P); (III) → (T); (IV) → (S)
(D) (I)→(Q); (II) → (S); (III) → (P); (IV) → (R)

Answer and Comments: (B). Each part in the first list asks for the
number of solutions of a trigonometric equation in a given interval.
In (I) rewrite the equation as sin(x + π4 ) = √12 . The general solution
is x + π4 = nπ + (−1)n π4 . For n = 0 we get x = 0. For n = 1 we get
x + π4 = π − π4 which gives x = π2 . Other values of n give solutions
outside the interval [− 2π
3
, 2π
3
]. So there are two solutions. Hence (I) →
(P).

26
 In (II) we rewrite the equation as tan 3x = √13 = tan π6 . Hence the
general solution is 3x = nπ + π6 . But 3x ∈ [− 5π , π ]. The only solutins
6 6
in this range occur for n = 0 and n = −1. Hence here too (P) is the
right match.
It is not necessary to decide the matching statements for the other
items in List I, because (B) is the only option in which both (I) and
(II) are matched with (P). So (B) holds.

Another repetitious problem. But some alertness obviates the need

to find the matches for III and IV.

Q.16 Two players, P1 and P2 play a game against each other. In every round
of the game, each player rolls a fair die once, where the six faces of the
die have six distinct numbers. Let x and y denote the readings on the
die rolled by P1 and P2 , respectively. If x > y, then P1 scores 5 points
and P2 scores 0 point. If x = y, then each player scores 2 points. If
x < y, then P1 scores 0 point and P2 scores 5 points. Let Xi and Yi be
the total scores of P1 and P2 , respectively, after playing the i-th round.
List-I List-II
3
(I) Probability of (X2 ≥ Y2 ) is (P) 8
11
(II) Probability of (X2 > Y2 ) is (Q) 16
5
(III) Probability of (X3 ≥ Y3 ) is (R) 16
355
(IV) Probability of (X3 > Y3 ) is (S) 864
77
(T) 432
The correct option is :
(A) (I) → (Q); (II) → (R); (III) → (T); (IV) → (S)
(B) (I) → (Q); (II) → (R); (III) → (T); (IV) → (T)
(C) (I) → (P); (II) → (R); (III) → (Q); (IV) → (S)
(D) (I) → (P); → (II) (R); (III) → (Q); (IV) → (T)

Answer and Comments: (A). We calculate the probabilities one

by one till we have calculated sufficiently many to identify the correct
option.

27
By symmetry, for every i, P (Xi > Yi ) is the same as P (Xi < Yi ).
Further 1 = P (X1 = Yi ) + P (xi > Yi ) + P (Xi < Yi ). So any one
of these three probabilities determines the othet two. Note also that
P (Xi ) ≥ Yi ) = P (Xi > Yi ) + P (Xi = Yi ). So if we can calculate
P (X2 = Y2 ) and P (X3 = Y3 ) we can answer all the four parts.
Clearly, E(X1 = Y1 ) occurs in 6 out of possible 36 cases. So P (X1 =
6
Y1 ) = 36 = 16 . So P (X1 > Y1 ) = P (X1 < Y1 ) = 21 (1 − 16 ) = 12
5
.
6
To begin with, we notice that P (X1 = Y1 ) = 36 = 16 . Hence P (X1 >
1 1 5
Y1 ) = 2 (1 − 6 ) = 12 . Note that X1 > Y1 means X1 = 5 and Y1 = 0,
while X1 = Y1 can occur only when X1 = Y1 = 2, because X1 , Y1 are
simply the scores of P1 and P2 on the first round.
For each i > 1, to calculate P (Xi = Yi ) we not only need the outcomes
at the i-th round, but also those at each of the earlier i − 1 rounds,
The score of P2 at each round can be determined from that P1 . It is
0, 2, 5 depending upon whether the score of P1 is 5, 2, 0 respectively. So
it suffices to keep track of the scores of either player. We choose P1 .
We first remark that in any round, the three possible scores of P1 are
5 5
5, 0 and 2, with probabilities 12 , 12 and 16 respectively.
The possible scores of P1 on the first two rounds are given by the nine
ordered pairs of the form (u, v) where each u, v can be 5, 2 or 0. Out
of these 9 possibilities only six are favourable for X2 ≥ Y2 . They are
listed here along with their probabilities.

Scores of P1 in Probability
first two rounds

5 5 25
(5, 5) 12
× 12 = 144
5
(5, 2) 12
× 61 = 72
5
1 5 5
(2, 5) 6
× 12 = 72
5 5 25
(5, 0) 12
× 12 = 144
5 5 25
(0, 5) 12
× 12 = 144
1 1 1
(2, 2) 6
× 6 = 36

Adding all, we get

75 + 40 + 4 99 11
P (X2 ≥ Y2 ) = = = (1)
144 144 16
28
Hence (I) → (Q). This rules out options (C) and (D) and restricts the
correct option to (A) or (B).
To find P (X2 > Y2 ), we drop the last three possibilities above (which
correspond to X2 = Y2 ) and get
99 54 45 5
P (X2 > Y2 ) = − = = (2)
144 144 144 16
which shows (II) → (R). As this is the case both in (A) and (B), we
need to calculate further.
In both (III) and (IV) we need to take the possible scores of P1 in the
first three rounds. This time there are 27 (= 33 ) possibilities of the
form (u, v, w) where each u, v, w is 5, 2 or 0. Since E(X3 > Y3 ) is a
subevent of E(X3 ≥ Y3 ), there are fewer possibilities and so we first
find the match for option (IV) rather than (III).
Instead of listing the favourable possibilities for E(X3 > Y3 ) in a table,
we classify them into four types. The probabilities of all possibilities
of the same type are equal.

Type Probability

5 3 125
(5, 5, 5) ( 12 ) = 1728
25
(5, 5, 2), (5, 2, 5) and (2, 5, 5) 144
× 61 each
5 3
(5, 5, 0), (0, 5, 5) and (5, 0, 5) ( 12 ) each
5 1 2
(5, 2, 2), (2, 5, 2) and (2, 2, 5) 12
× ( 6
) each

Note that we have not listed (2, 2, 2) as a possibility since it gives

X3 = Y3 . Adding,
125 + 125 × 3 + 50 × 3 + 20 × 3
P (X3 > Y3 ) =
1728
710 355
= = (3)
1728 864
which show (IV) → (S). So far we have shown that (I), (II) and (IV)
match with (Q), (R) and (T) respectively. As the only option in which
this happens is (A), even without finding the match for (III), (A) is the
right option.

29
 Conceptually, a simple but good problem on probability. Unfortu-
nately, the calculations involved are too many and liable for a numerical
slip. This question should have been a full length question. Then asking
even a single probability, say, P (X3 > Y3 ) would enable an examiner
to tell merely looking at the answer whether the candidate has done
the correct thinking, because 355
864
is not an answer anybody can get by
fluke. It is only when a candidate gives a different numerical answer
that the examiner will have to check if the candidate’s mistake was
conceptual or merely a computational slip.

Q.17 Let p, q, r be nonzero real numbers that are, respectively, the 10-th,
100-th and 1000-th terms of a harmonic progression. Consider the
system of linear equations
x+y+z = 1
10x + 100y + 1000z = 0
qrx + pry + pqz = 0
List-I List-II

(I) If qr = 10, then the system of lin- (P) x = 0, y = 109

, z = − 19
ear equations has as a solution

(II) If pr 6= 100, then the system of lin- (Q) x = 109

, y = − 19 , z = 0
ear equations has as a solution

(III) If pq 6= 10, then the system of lin- (R) infinitely many solu-
ear equations has tions

(IV) If pq = 10, then the system of lin- (S) no solution

ear equations has
(T) at least one solution
The correct option is :
(A) (I) → (T); (II) → (R); (III) → (S); (IV) → (T)
(B) (I) → (Q); (II) → (S); (III) → (S); (IV) → (R)
(C) (I) → (Q); (II) → (R); (III) → (P); (IV) → (R)
(D) (I) → (T); → (II) (S); (III) → (P); (IV) → (T)

Answer and Comments: (B).

If we had only the first two equations then eliminating x from them we

30
get 9y + 11z = −1. Hence
−11z − 1
y= (1)
9
Substituting this into x + y + z = 1, we can get x in terms of z. As z
can be arbitrary, there are infinitely many solutions. x = 0 gives the
solution in (P) while z = 0 gives the solution in (Q).
The third equation complicates the problem, especially because the
coefficients of x, y, z are given in terms of an unknown harmonic pro-
gression. Harmonic progressions do not figure as frequently in problems
as the arithmetic or the geometric progressions. The best way to han-
dle an H.P. is to keep in mind that its terms are the reciprocals of
those of an A.P. So, in the present problem, p1 , 1q and 1r are the 10-th,
100-th and the 1000-th term of some A.P., say, with the first term a
and common difference d. Then
1
= a + 9d (2)
p
1
= a + 99d (3)
q
1
and = a + 999d (4)
r

These formulas will be of immediate use if we divide the third given

equation by pqr. Doing so the third equation becomes

(a + 9d)x + (a + 99d)y + (a + 999d)z = 0 (5)

We now have a system of three linear equations in three unknowns. But

since the values of a and d are not given, the existence as well as the
nature of the solutions of this system could conceivably depend upon
the particular values of a and d. A comforting observation is that the
last equation depends only upon the ratio ad rather than on the actual
values of a and d.
In each statement in List-I, some condition is given about the ratio of
two of the numbers p, q and r. Let us see what these conditions imply
about the relationship between a and d.

31
In (I), we are given qr = 10 which, from (3) and (4), gives a+999d
a+99d
= 10.
This simplifies to 9a = 9d and hence a = d. But then (5) becomes the
same as the second equation of the system and hence redundant. We
have already seen that in this case there are infinitely many solutions,
of which one is as in (P) and the other as in (Q). Hence (P), (Q), (R)
and (T) are all true in this case.
By very similar calculaions, in (IV), pq = 10 implies a = d and this
leads to infinitely many solutions with one as in (P) and another as in
(Q). So here too (P), (Q), (R) and (T) are all true.
But in (II) and (III), the same calculations show that a 6= d. From this
we claim that the system has no solution. For, now the third equation
of the system is (5). Since ax + ay + az = a we get

9dx + 99dy + 999dz = −a (6)

Adding dx + dy + dz = d to both the sides gives

10dx + 100dy + 1000dz = a − d (7)

But, on the other hand because of the second given equation the L.H.S.
is 0. So, a = d a contradiction.
Thus in (II) and (III) the only match is with (S). The only correct
option which matches (II) and (III) to (S) and (I) and (IV) to any of
the other four entries in List-II. So the correct choice is (B).

There is a systematic method due to Gauss based on what is called

row reduction which completely analyses a system of linear equa-
tions, that is, it tells if the system is consistent or not and in case it
is, gives the totality of all possible solutions and no others. In other
words, it speaks the truth, the full truth and nothing but the truth.
It is amusing that a great mathematician like Gauss also invented this
method which involves little more than elementary arithmetic. But
many candidates at the JEE level may be unfamiliar with it. And in
the present problem its superiority is not very pronounced. So we have
given a direct argument.
Problems about systems of linear equations where the nature of the
solution set depends on some parameters appearing in the coefficeints

32
 of one or more equations in the system are fairly common in JEE.
The present problem is of this type. Its unusual feature is that the
conditions about the parameters (such as pq = 10) have to be translated
in terms of two other parameters, viz. a and d, which do not appear in
the statement of the problem. It is for the candidate to choose them.
That makes the problem novel and interesting.
x2 y2
Q.18 Consider the ellipse + = 1. Let H(α, 0), 0 < α < 2, be a
4 3
point. A straight line drawn through H parallel to the y-axis crosses
the ellipse and its auxiliary circle at points E and F respectively, in
the first quadrant. The tangent to the ellipse at the point E intersects
the positive x-axis at a point G. Suppose the straight line joining F
and the origin makes an angle φ with the positive x-axis.
List-I List-II
√
( 3−1)4
(I) If φ = π4 , then the area of the triangle (P) 8
F GH is
(Q) 1
(II) If φ = π3 , then the area of the triangle 3
F GH is (R) 4
1
(III) If φ = π6 , then the area of the triangle (S) √
2 3
F GH is √
3 3
(T) 2
π
(IV) If φ = 12 , then the area of the triangle
F GH is
The correct option is :
(A) (I) → (R); (II) → (S); (III) → (Q); (IV) → (P)
(B) (I) → (R); (II) → (T); (III) → (S); (IV) → (P)
(C) (I) → (Q); (II) → (T); (III) → (S); (IV) → (P)
(D) (I) → (Q); → (II) (S); (III) → (Q); (IV) → (P)

Answer and Comments: (C). In the past, some matching type

questions invited criticism for the diversity of the topics covered by the
entries in the first column. Thus, one would involve a trigonometric
equation, some other would be an integral, yet another would ask for
the focus of some conic and so on. The present question is an example
of the opposite extreme. Here not only all four items involve an ellipse,

33
there is no variation in what is asked about the ellipse, like its eccen-
tricity or its foci. All items ask for the area of the same trinagle F GH,
the only difference being the angle φ.
Obviously, we first have to express the area as a function of φ and then
evaluate it at the values given in the first column.
The auxiliary circle of the ellipse has equation x2 + y 2 = 4. A point
F on it can be taken in the parametric form as (2 cos φ, 2 sin φ), where
φ is the angle OF makes with the √ positive x-axis. The corresponding
point E on the ellipse is (2 cos φ, 3 sin φ).

(0,2) F

φ
x
O H ( α,0) (2,0) G

Since E, F and H lie on the same vertical line, we have α = 2 cos φ.

(The statement of the problem begins with α. But later an angle φ is
introduced. And the area is to be calculated in terms of φ, rather than
α.)
As F GH is a right angled triangle, with F H = 2 sin φ, to find its area
we only need to identify the point G.√ The equation of the tangent to
the ellipse at the point E = (2 cos φ, 3 sin φ) is

x cos φ y sin φ
+ √ =1 (1)
2 3

34
Its point of intersection with the x-axis is G = ( cos2 φ , 0) = (2 sec φ, 0).
Since H = (α, 0) = (2 cos φ, 0), GH = 2 sec φ − 2 cos φ. Hence

area of△F GH = (sec φ − cos φ)2 sin φ

= 2(1 − cos2 φ) tan φ
= 2 sin2 φ tan φ (2)

Now it is merely a matter of giving φ the values π4 , π3 , π6 and 12

π
. Only
the last one is complicated because the values of sin 15 and tan 15◦ are
◦

not as √common as those of the other three. They can be calculated by

using 23 = cos 30◦ = 1 − 2 sin2 15◦ . But the paper-setters have been
merciful by giving (P) as a match for (IV) in all options. So, we only
need to evaluate 2 sin2 φ tan√φ when φ = π4 , φ3 and π6 . In the first two
cases the answer is 1 and 3 2 3 by easy calculations. They correspond
to (Q) and (T) respectively. The only option in which this happens is
(C). So, without any further calculation, (C) is the right option.

Basically, a simple problem. The calculations are repetitious and

some of them complicated. But a little (non-mathematical) alertness
reduces the work by half.

35
 PAPER 2

Contents

Section - 1 (Single Digit Answer type) 36

Section - 2 (One or More correct Options Type) 47

Section - 3 (Only One Option correct Type) 61

SECTION - 1
This section contains EIGHT questions each of which is to be answered
with an integer from 0 to 9, both inclusive with a numerical value (after
truncating or rounding off) with two places of decimals.
There are 3 marks for a correct answer 0 if no answer and −1 in all other
cases.
π π
Q.1 Let α and β be real numbers such that − < β < 0 < . If sin(α +
4 4
1 2
β) = and cos(α − β) = , the greatest integer less than or equal to
3 3 !2
sin α cos β cos α sin β
+ + + is .... .
cos β sin α sin β cos α

Answer and Comments: 1. There is little point in finding α, β from

the data and then taking their sines and cosines. A better option is
to recast the given expression, say E, so that it will involve only the
trigonometric functions of α + β and α − β.
It is hardly practical to expand E directly by squaring the sum of four
terms as they are. The trick is to group them suitably, add each group
and then take the square. For example, combining the first and the
third term gives
sin α cos α cos α cos β + sin α sin β
+ =
cos β sin β sin β cos β
cos(α − β) 2 cos(α − β)
= = (1)
sin β cos β sin 2β

36
 From the given value of cos(α − β) as 23 , this simplifies to 3 sin4 2β . Sim-
ilarly, the sum of the second and the fourth term simplifies to 3 sin4 2β .
Therefore,
16 1 1 2
E = ( + ) (2)
9 sin 2β sin 2α
With more trigonometric identities and using the given values again,
the expression inside the parentheses simplifies to 1. So, E = 16
9
and
hence its integral part is 1.

A problem where the ability to simplify a given expression is the

sole key. Squaring has little role. Possibly it is a red signal to alert a
candidate that it is better to first simplify the sum.

Q.2 If y(x) is a solution of the differential equation xdy − (y 2 − 4y)dx = 0

for x > 0, y(1) = 2, and√ the slope of the curve y = y(x) is never 0,
then the value of 10y( 2) is .... .
Answer and Comments: 8. Separating the variables and integrat-
ing, the general solution of the d.e. comes out as
y−4
ln | | = 4 ln x + ln c (1)
y
for some constant c. The initial condition y = 2 when x = 1 determines
c as 1. Therefore the particular solution is

|y − 4|
= x4 (2)
|y|

whence
y−4
= ±x4 (3)
y
The negative value has to be chosen because when x is in a neighbour-
hood of 1, by continuity y is in a neighbourhood of 2 and so the L.H.S.
is negative. Therefore the solution is

y − 4 = −yx4 (4)

37
 √ √ 40
At x = 2, this becomes 5y = 4 and so 10y( 2) = 5
= 8, which is
actually an integer.

The differential equation part is easy. Some thought is needed in

choosing the negative sign in (3).

Q.3 The greatest integer less than or equal to

Z 2 Z log2 9
log2 (x3 + 1)dx + (2x − 1)1/3 dx
1 1

is ..... .
Answer and Comments: 1. Neither integral is easy to evaluate.
But as commented in Q.1 of Paper 1, sometimes the sum of two com-
plicated expressions is easy to evaluate.
Z a A frequentZ a occurrence of this
situation is in integrals of the type f (x)dx + f (a − x)dx if the
0 0
sum function f (x) + f (a − x) is very easy to integrate.
To guess the relationship between the two integrals, we note that
in the first integrand, log w.r.t. base 2 is taken while in the second,
powers of 2 are involved. This suggests that they are inverses of each
other. This indeed turns out to be the case. If we put y = log2 (x3 + 1),
then 2y = x3 + 1 and hence x = (2y − 1)1/3 . Thus the two integrands
are inverses to each other. If we call the first integrand as f (x) then
the second is simply f −1 x. Denote it by g(x). Further, the limits of
the first integral viz. 1 and 2 are taken by f to those of the second.
Therefore the sum is of the form
Z b Z f (b)
f (x)dx + g(x)dx (1)
a f (a)

where f (x) is strictly increasing and differentiable on [a, b] and its in-
verse function g(x) is also differentiable on [f (a), f (b)].
Call the two integrals as I1 and I2 respectively. There is a tricky way
to evaluate this sum. In the second integral, put x = f (u). Then
g(x) = g(f (u) = u since f and g are inverses of each other. Also
dx = f ′ (u)du. Hence
Z b
I2 = uf ′(u)du (2)
a

38
 If we integrate this by parts, then
Z b
I2 = uf (u)|ba − f (u)du (3)
a

The frst term is simply bf (b)−af (a). If we change the dummy variable
of integration from u to x the integal on the R.H.S. is simply I1 . So we
get

I1 + I2 = bf (b) − af (a) (4)

Applying this with a = 1, b = 2 and f (x) = log2 (x3 + 1), the given sum
equals 2 log3 9 − 1 log2 2 = log2 81 − 1. Since 26 = 64 < 81 < 128 = 27 ,
we get that the integral part of I is 6 − 1 = 5.

A tricky problem. Easy for those who think of (4). A somewhat

similar trick was needed in Q.7 of Part B of ISI Entrance Test 2016.
Q.4 The product of all positive real values of x satisfying the equation
3 −68 log
x(16(log5 x) 5 x)
= 5−16

is .... .

Answer and Comments: 1. As powers of 5 as well as logarithms

with base 5 are involved, there is an imlicit hint to take logarithms of
both the sides w.r.t the base 5. Doing so,

(16(log5 x)3 − 68 log5 x) log5 x = −16 (1)

Putting u = log5 x, this becomes (16u3 − 68u)u = −16 i.e.

16u4 − 68u2 + 16 = 0 (2)

Let u1 , u2, u3 , u4 be the four roots of this equation. Then those of the
original equation (in x) will be 5u1 , 5u2 , 5u3 , 5u4 . Their product will be
5u1 +u2 +u3 +u4 . This reduces the problem to finding the sum of the roots
of (2). As there is no cubic term in it, the sum is 0. Therefore the
product of the roots of the given equation is 50 = 1.

Those who cannot think of this short cut, can rewrite (2) as 4u4 −
17u2 + 4 = 0 and further as quadratic in v = u2 , viz. 4v 2 − 17v + 4 = 0

39
 √
whose roots are 17± 289−64
8
= 17±15
8
, i.e. 4 and 14 . Therefore u = ±2, ± 21
±2 ±1/2
and, correspondingly, x = 5 , 5 . The product of these four values
is 1.
3
 
ex − (1 − x3 )1/3 + (1 − x2 )1/2 − 1 sin x
Q.5 If β = , then the value of 6β
x sin2 x
is .... .

Answer and Comments: 5. The very form of the limit suggests

that it is in an indeterminate form. As the answer is to be an integer
between 0 and 9, there is an implied hint that β is a fraction with
denominator dividing 6.
Near 0, sin x is comparable to x. (A formal statement is that sinx x tends
to 1 as x → 0.) So we can safely replace sin x by x to rewrite

3
 
ex − (1 − x3 )1/3 + (1 − x2 )1/2 − 1 x
β= (1)
x3
An easy method to evaluate β is to expand the numerator in powers of
x and collect the coefficients of x3 . For this we need the power series
3
expansions of ex , (1 − x3 )1/3 and (1 − x2 )1/2 . The first one is obtained
from the standard expansion of ex and repalcing x by x3 .

x3 x6 x9
3
e =1+x + + + ... (2)
2! 3!
which is valid for all x ∈ IR. The other expansions we need, viz.
(1 − x3 )1/3 and (1 − x2 )1/2 are both examples of binomial expansions
with fractional exponents. The coefficients are calculatedexactly as
n
if the exponent n were a positive integer n, except that r has no
combinatorial significance and has to be interpreted algebraically as
n(n−1)(n−2)...(n−r+1)
r!
. The expansion of (1 − x3 )1/3 thus becomes
1 1
1 ( − 1) 6
(1 − x3 )1/3 = 1 − x3 + 3 3 x + ... (3)
3 2!
(As we are interested only in the coefficient of x3 , we need not expand
beyond the second term.)

40
In the expansion of the last term in the numerator of (1), we need to
take only the coefficient of x2 in the expansion of (1 − x2 )1/2 . This is
very analogous to (3) and gives (because of the factor x at the end),
1
((1 − x2 )1/2 − 1)x = − x3 + . . . (4)
2
Collecting the coefficients of x3 in (2), (3) and (4), we get that β =
1 + 13 − 21 = 56 . Hence 6β = 5.

Power series provide a powerful tool which makes the calculation

of certain tricky limits almost mechanical as in the present problem,
thereby obviating the need for other methods like the l’Hôpitals’s rule.
(It is another matter that sometimes l’Hôpitals’s rule itself can be ob-
viated by recasting the limit as a derivative.)
The catch with power series expansions is that they are not uncondi-
tionally valid. For example, although (2) is valid for all x ∈ IR, (3)
and (4) hold only for |x| < 1. As an example of how an indiscriminate
application of power serries leads to a ridiculous result, the binomial
expansion of (1 − x)−1 comes out to be
(1 − x)−1 = 1 + x + x2 + . . . + xn + . . . (5)
1
If we write the L.H.S. as 1−x , this is the formula for the sum of an
infinite geometric series. But if we put x = 2, we get that the infinite
series 1 + 2 + 4 + . . . + 2n + . . . adds to the negative number −1. This
happens because 2 is outside the interval (−1, 1) of validity. (When the
exponent
  n is a positive integer, this difficulty does not arise because
n
r
vanishes for all r > n and so the series is really a finite sum.)
An ‘ethical’ objection to the use of the power series is that the
justifications needed to establish their validity are above the JEE level.
This objection would apply if a candidate has to show the reasoning
and not in a situation where all that matters is how rapidly you get
the answer and not how you get it.
Still, for the sake of the consciensious readers here is a a proof
without power series. Call the given limit as L. After replacing sin x
by x, we split it as L1 − L2 + L3 , where
3
ex − 1
L1 = lim (6)
x→0 x3
41
 (1 − x3 )1/3 − 1
L2 = lim (7)
x→0 x3
(1 − x2 )1/2 − 1
and L3 = lim (8)
x→0 x2
The substitution u = x3 converts L1 and L2 , respectively, to the deriva-
tives (at 0) of the functions eu and (1 − u)1/3 . These are, 1 and − 13 .
Similarly, the substitution v = x2 converts L3 to the (right handed)
derivative of (1 − v)1/2 at 0. Hence L3 = − 12 . Thus we have found the
limit not only without power series but without l’Hôpital’s rule too.
(A question may arise how we know these derivatives. They follow
from those of their inverse functions which are ln u, 1 − u3 and 1 − v 2
respectively.)
 
β 0 1
Q.6 Let β be a real number. Consider the matrix A =  2 1 −2  . If


3 1 −2
A7 − (β − 1)A6 − βA5 is a singular matrix, then the value of 9β is .... .

Answer and Comments: 3. By giving a high power of a 3×3 matrix,

the paper-setters have given a loud hint that its full expansion is not
the right way to get the answer. (Calculating even the square of a 3 × 3
matrix is often a tedious job, prone to errors. Calculating each of its 9
entries involves 3 multiplications and two additions if done directly.)
As A5 is an obvious factor of the given matrix, The right start is to
factorise the matrix as
A7 − (β − 1)A − βA5 = A5 (A2 − (β − 1)A − β) (1)
where the last term is really βI, I = I3 being the identity matrix of
order 3. This is given to be singular. As the matrix A is non-singular,
so is A5 . Hence the other factor A2 −(β−1)A−βI is singular. Therefore
its determinant vanishes. This is better than knwoing the singularity of
the matrix A7 − (β − 1)A − βA5 . But still calculating A2 and then the
determinant is a tedious task. So, we look for a further factorisation.
Luckily, we get this by inspection.
A2 − (β − 1)A − βI = A2 + A − βA − βI
= A(A + I) − β(A + I)
= (A + I)(A − βI) (2)

42
 So, once again, at least one of the two factors is singular and has a
vanishing determinant. It is easier to calculate the determinant of the
first factor.
 
β + 1 0 1 


|A + I| = 2
 2 −2  = 0 + 1(2 − 3) = −1 6= 0 (3)
3 1 −1 
 


We are now only left with the second factor A − β whose determinant
must vanish.
 
0
 0 1 

|A − β| = 2 1 − β −2 = 2 − 3 + 3β = 3β − 1 (4)
 
 
3 1 −2 − β
 
 

The vanishing of the determinant means β = 31 . Hence 9β = 3.

A simple problem once the idea of factorisation strikes. First fac-

torisation is easy. The second one needs a little inspection.
x2 y2
Q.7 Consider the hyperbola − = 1 with foci at S and S1 , where S
100 64
lies on the positive x-axis. Let P be a point on the hyperbola, in the
first quadrant. Let 6 SP S1 be α with α < π2 . The straight line passing
through the point S and having the same slope as that of the tangent
at P to the hyperbola, intersects the straight line S1 P at P1 . Let δ be
the distance of P from the straight line SP1 , and β = S1 P . Then the
greatest integer less than or equal to βδ
9
sin α2 is ..... .

Answer and Comments: 7.

The drafting of the problem is a masterpiece of obfuscation. It in-
troduces many symbols. So a diagram is necessary to understand it.
Further it is unnecessarily verbose. What is the great idea in describing
a line as a line having the same slope as some other line when there is
a familiar word ‘parallel’ ? Also, the point P1 has little role. It would
be far better to introduce the line through S parallel to the tangent
at P as, say L, and δ be the distance of P from L. It is silly to say
‘This is my colleague and this is his son Ram’ and then refer to the
colleague as Ram’s father. It is far better to introduce your colleague

43
by some name and then refer to him by that name without involving
Ram at all. In a test in which a candidate gets barely 3 minutes for
each question, such a waste is costly.
In the figure below we have shown all the data given by the paper-
setters. In addition, we have called the tangent at P as T , the line
through S parallel to (sorry, having the same slope as) T as L. We
have also shown (by dotted lines) the perpendiculars SR and S1 Q from
S and S1 , repectively, to T .

P P1
β α /2 δ
α/2
R
δ
S1 S x
O

There is a subtle hint in the data given. Everything in the problem,

such as α, β and δ depends on the point P which can be anywhere on
the hyperbola in the first quadrant. But the answer expected, viz. the
value βδ
9
sin α2 is some fixed number, i.e. independent of α, β and δ.
This suggests that we must appeal to some results which are true for
the tangent at any point of the hyperbola regardless of where the point
is.
Another hint is that the expression involves α2 which is half the an-
gle 6 SP S1 . So, somewhere the angle bisector of this angle should be
involved. But how is it related?
The first property of the tangent to a hyperbola at P gives a very
direct relationship. It says that the angle bisector of 6 SP S1 is itself the

44
tangent. In other words, the tangent at P bisects the angle subtended
at it by the segment joining the two foci. (There is a similar result
for an ellipse except that there it is the normal at P which bisects the
angle. It has a physical interpretation that all light rays originating
at one focus pass through the other focus after being reflected by the
ellipse regardless of which point of the ellipse they fall upon. The same
holds for sound waves and whispering galleries are based on it. So,
for an ellipse, as well as for a hyperbola, these properties are called
focusing properties. The focusing property for a parabola is most
well known and says that all incoming rays parallel to the axis pass
through the focus after being reflected by the parabola.)
Now that we know that the triangle S1 QP is right angled with 6 S1 P Q =
α
2
we get
α
β sin = S1 Q (1)
2

Thus we have partly unravelled the mysterious expression βδ9

sin α2 . We
δ 1
still need to work with 9 . The constant 9 has little role. But what
significance does δ have? It is given as the perpendicular diatance of
the point P from the line L. But since L and T are parallel, it also
equals the perpendicular distance of S from T . In other words,

δ = SR (2)

Therefore βδ sin α2 is simply the product of the perpendicular distances

of S1 and S from the tangent at P . There is a result which says that
this product is independent of P and always equals b2 where b is the
semi-minor axis of the hyperbola. In the present problem, b = 8. Hence
δβ sin α 64
9
2
= 9 whose integral part is 7.

Evidently, the problem is tailor made for those who know the two
properties of the tangent to a hyperbola. The first one is quite well-
known. The second one is not. But it is easy to prove. Taking P =
x2 y 2
(a sec θ, b tan θ) as a point on the hyperbola 2 − 2 = 1, the equation
a b
of the tangent at P comes as

b sec θx − a tan θy = ab (3)

45
 The foci lie at (±ae, 0). Their distances from the tangent are
abe sec θ − ab abe sec θ + ab
√ and √ repsectively. A direct
2
a2 tan θ + b2 sec2 θ a2 tan2 θ + b2 sec2 θ
algebraic simplification, using a2 e2 = a2 + b2 gives that their product
equals b2 .
The theme of the problem is good. But giving the data in a clumsy
form is repulsive. Also it is questionable whether a less known result
can be expected of the candidates. In the good old days, in a full length
question, Part (A) would ask a proof of this property and as ususal,
Part (B) would be the present problem, to be solved with or without
using (A).
5
Q.8 Consider the (functions f, g : IR −→ IR defined by f (x) = x2 + 12
2(1 − 4|x| ), |x| ≤ 43 ,
and g(x) = 3 . If α is the area of the region
n
0, |x| ≥ 43 o
(x, y)IR × IR : |x| ≤ 34 , 0 ≤ y ≤ min{f (x), g(x)} , then the value of 9α
is .... .

Answer and Comments: 6. Both the functions are even functions

of x. Also the region is symmetric about the y-axis. Hence it suffices
to find the area of its eight half, say,
3
 
R1 = (x, y)IR × IR : 0 ≤ , 0 ≤ y ≤ min{f (x), g(x)} (1)
4
The first task is to determine which of f (x) and g(x) is smaller for
which values of x. For this we find where their graphs intersect. Since
g(x) = 2 − 38 x for x ≥ 0, we look for the non-negative roots of the
equation
5 8
x2 + =2− x (2)
12 3
which reduces to the quadratic equation

12x2 + 32x − 19 = 0 (3)

The roots can be found by the quadratic formula or by noting that the
L.H.S. factors as (6x + 19)(2x − 1). The only positive root is 21 . As
the first factor is positive for all x ≥ 0, we get that for 0 ≤ x ≤ 12

46
 f (x) ≤ g(x) while for x ≥ 12 , g(x) ≤ f (x). Therefore the area of the
region R1 is
1/2 5 3/4 8
Z Z
2
(x + )dx + (2 − x)dx (4)
0 12 1/2 3
1
Straightforward evaluations give these integrals as 41 and 12 respectively.
1 2
Hence the area of R1 is 3 . So α, the area of R, is 3 . Therefore 9α = 6.

An absolutely routine computational problem.

SECTION - 2
This section contains SIX questions each of which has four options of
which one or more are correct. There are four marks if all correct and no
incorrect option are chosen, 0 marks if no option is selected and −2 marks
if an incorrect option is chosen regradless of whether any correct option is
chosen. When some but not all correct options and no incorrect options are
chosen, there is some partial marking.

Q.9 Let P QRS be a quadrilateral in a plane, where QR = 1, 6 P QR =

6 QRS = 70◦ , 6 P QS = 15◦ and 6 P RS = 40◦ . If 6 RP S = θ ◦ , P Q = α

and P S = β, then the interval(s) that contains the value of 4αβ sin θ◦
is/are √ √ √ √
(0, 2) (B) (1, 2) (C) ( 2, 3) (D) (2 2, 3 2)

Answer and Comments: (A,B) .

Accurate diagrams are not crucial in
R
such problems. In the accompaning 1
(indicative) diagram, the data im- 40
S 55
plies that 6 SQR = 70 − 15 = 55 de- 30
grees and also that 6 RSQ = 180 − 1
70 − 55 = 55 degrees. Hence the tri- β
angle QRS is isosceles. So SR = 1.
θ 80 55
70
P α Q

47
 To evaluate the expression, αβ sin θ, we can apply the sine rule. Since
β and θ are both involved in the triangle P RS, we apply it to this
triangle and get
sin θ◦ sin 40◦
= (1)
1 β
which gives

β sin θ◦ = sin 40◦ (2)

Similarly, applying the sine rule to the triangle P RQ gies

1
α= (3)
2 sin 80◦
Multiplying the two and using sin 80◦ = 2 sin 40◦ cos 40◦ we get
1
4αβ sin θ◦ = = sec 40◦ (4)
cos 40◦
The problem is now reduced to deciding which of the four intervals
contain sec 40◦ . The exact value of sec 40◦ is not a familiar expression.
But since the secant function is strictly increasing on the interval [0, π2 ]
we get lower and upper bounds on it√in terms of the secants of the
angles 30◦ and 45◦ , which are √23 and 2. Hence

2 √
√ < sec 40◦ < 2 (5)
3
Hence (A), (B) are correct options.

A good combination of two simple but unrelated trigonometric

problems.
∞
π
sin2k ( ). Let g : [0, 1] −→ IR be the function defined by
X
Q.10 Let α
k=1 6

g(x) = 2αx + 2α(1−x)

. Then, which of the following statements is/are TRUE?

48
(A) The minimum value of g(x) is 27/6
(B) The maximum value of g(x) is 1 + 21/3
(C) The function g(x) attains its maximum at more than one point
(D) The function g(x) attains its minimum at more than one point

Answer and Comments: (A,B,C). The problem splits naturally

into two (unrelated) parts, evaluating an infinite series and then study-
ing the maxima/minima of a function.
The first part is easy. The given series is a geometric series with the
first term as well as the common ratio sin2 π6 = 41 . Hence
1/4 1/4 1
α= = = (1)
1 − (1/4) 3/4 3
As a result,
1
g(x) = 2x/3 + 2 3 (1−x) (2)
g(x) is differentiable everywhere. So its maxima/minima on [0, 1] can
be found from its derivative.

1 1 1
g ′(x) = (ln 2)2x/3 − (ln 2)2 3 (1−x) (3)
3 3
This vanishes if and only if x = 1 − x, i.e. at x = 12 . But since ln 2
is positive and x < 1 − x for x < 12 , the derivative changes sign from
negative to positive at 21 . Therefore the minimum occurs at x = 21 and
the maximum on the interval [0, 1] occurs at the end points 0 and 1.
So (C) is true and (D) false. A direct computation gives g(0) = g(1) =
1 + 21/3 and g( 12 = 27/6 . So both (A), (B) are true.

The purpose of asking this question in an advanced test is not clear.

Both the parts, viz. summing an infinite geometric series and finding
the maximum and the minimum of a function are very routine and a
candidate who has cleared JEE Mains can be safely assumed to know
them. The only testing point in the solution is to realise that g ′ (x) is
negative for x < 12 and positive for x > 21 . If this point is missed, the
maxima and the minima will be interchanged.

49
Q.11 Let z denote the complex conjugate of a complex number z. If z is a
non-zero complex number for which both real and imaginary parts of
1
(z)2 +
z2
are integers, then which of the following is/are possible value(s) of |z|2
?
 √ 1/4  √ 1/4
43+3 205 7+ 3
(A) 2
(B) 4
 √ 1/4  √ 1/4
9+ 65 7+ 13
(C) 4
(D) 6

Answeer and Comments: (A). In Q.4 of Paper 1, the problem

dealt with some complex number being real. The direct approach is
to equate its imaginary part with 0. But often the characterisation
that a complex number is real if and only if it equals its own complex
conjugate was useful. Unfortunately, there is no easy characterisation
of the integrality of the real and imaginary parts. That is more a
number theory question and ofen difficult.
The present problem does not ask to either prove or use any such
characterisation. Instead, it asks which of the given real numbers can
arise as |z|2 for some z which satisfies the condition that the real and
imaginary parts of z 2 + 1/z 2 are both integers. By giving all options
as the fourth roots of some real numbers, the paper setters have given
an implicit hint that we have to deal with not just squares but with
fourth powers.
To see how fourth powers are involved, note for example, that if we
take z in its polar form as
z = r(cos θ + i sin θ) (1)
then the expression whose real and imaginary parts we are interested
in can be rewritten as
1 |zz|2 + 1
z2 + =
z2 z2
r4 + 1
= 2
r (cos 2θ + i sin 2θ)
(r 4 + 1)(cos 2θ − i sin 2θ)
= (2)
r2
50
If the real and imaginary parts are integers, then there are some integers
α and β such that
r4 + 1
cos 2θ = α (3)
r2
r4 + 1
and sin 2θ = β (4)
r2
Eliminating θ, we get
!2
r4 + 1
= α2 + β 2 (5)
r2
or equivalently,
1
r4 + = α2 + β 2 − 2 (6)
r4
The R.H.S. is an integer. So, if z is a number for which z 2 + z12 has
integral real and imaginary parts, then its absolute value r = |z| must
satisfy that r 4 + r14 is an integer.
We now try the options one-by-one and decide for which of them this
condition holds. In all cases, the numbers given are quadratic surds
and while taking their reciprocals, we rationalise them. Thus in (A),
1 2
4
= √
r 43 + 3 205
q
2(43 − 3 205)
=
(43)2 − 9 × 205
√
2(43 − 205)
=
1849 − 1845
√
43 − 205
= (7)
2
Adding,
1 43 + 43
r4 + 4
= = 43 (8)
r 2
which is an integer. Similar calculations which we skip, give that in
(B), (C) and (D), the values of r 4 + r14 are 27 , 29 and 76 respectively. As

51
none of these is an integer, (B), (C), (D) are false. As at least one
option is true, it must be (A).

In the solution above we derived the integrality of r 4 + r14 as a

necessary condition for the existence of a complex number z for which
|z| = r and the real and imaginary parts of z 2 + z12 are integers. We did
not prove its sufficiency. To give a complete solution to the problem
we must show that in (A) such a complex number indeed exists. In
other words, we must find integers α and β such that (6) and hence
(5) holds. Once we find such integers, we let θ = 12 tan−1 αβ . With this
value of θ, the complex number z = r(cos θ + i sin θ) has the desired
property. From (8), the L.H.S. of (5) equals 45. So all we need to do is
to express 45 as the sum of two perfect squares. This is easy. Simply
 √ 1/4
write 45 as 32 + 62 . So with r = 43+32 205 and θ = 12 tan−1 2, we
get that z = r iθ is a number for which the real and imaginary parts of
z 2 + z12 are both integers.
It should be noted that not every positive integer can be expressed
as a sum of two perfect squares. (There is a criterion to tell which
ones can, but we skip it.) For example, 7 cannot. Hence if r is a real
number for which r 4 + r14 = 7 − 2 = 5, then even though r 4 + r14 is
an integer, there is no complex number z with |z| = r for which the
real and imaginary parts of z 2 + z12 are both integers. In other words,
the integrality of r 4 + r14 is a necessary but not a sufficient condition.
To make it so, we must add that not only is r 4 + r14 is an integer, but
r 4 + r14 + 2 can be expressed as a sum of two perfect squares. Thus, in
the case of (A), we have now given a legitimate certificate of its truth
rather than the back door argument that among the given options, all
others are false.
An excellent problem on complex numbers, requiring very unusual
thinking. The repetitious work in testing the integrality of r 4 + r14 in
all four options mars the challenge involved in reducing the problem to
testing the integrality of r 4 + r14 . Proving the necessity and sufficiency
of the condition would have been a perfect full length question for the
JEE in the past when the paper-setters had the discretion to allocate
marks according to the degree of difficulty of a problem. Sadly, in the
present JEE it ranks on par with the last question in which two simple
and worn out problems are clubbed together.

52
Q.12 Let G be a circle of radius R > 0. Let G1 , G2 , . . . Gn be n circles of
equal radius r > 0. Suppose each of n circles G1 , G2 , . . . , Gn touches
the circle G externally. Also, for i = 1, 2, ...n − 1, the circle Gi touches
Gi+1 externally, and Gn touches G1 externally. Then, which of the
following statements is/are TRUE?
√
(A) If n = 4, then ( 2 − 1)r < R
(B) If n = 5, then r < R
√
(C) If n = 8, then ( 2 − 1)r < R
√ √
(D) If n = 12, then ( 2( 3 + 1))r > R

Answe and Comments: (C,D). Let C and C1 , C2 , . . . , Cn be the

centres of G and G1 , G2 , . . . , Gn respectively. By summetry, the angle
between any two lines CCi and CCi+1 is the same, viz. 2π n
for all i.

G
2
G

P
1

r G
π/n 1

C R r C1

Gn

As all options involve the ratio Rr , we first obtain an expression for this
ratio in terms of n. Let P1 be the point of contact of G1 and G2 . Then
6 C1 CP1 = π . From the right angled triangle CC1 P1 we get sin π = r
n n R+r
which translates into
R π
= cosec − 1 (1)
r n

53
We now give n the values 4, 5, 8 and 12 and decide the truth of the
options one-by-one. It is good to keep in mind that as n increases, πn
decreases and cosec nπ increases. Therefore the ratio Rr also increases
strictly. This is intuitively obvious because for a fixed R, the outer
circles will get smaller and smaller when their number increases.
√ √
For n = 4, cosec( π4 ) = 2. Therefore Rr = ( 2 − 1). Hence (A) is
false as it requires inequality.
For n = 5, cosec( π5 ) = sin136◦ . Hence, Rr = sin136◦ − 1. The exact value
√ √
of sin 36 is 10−2
◦
4
5
. Hence Rr = √ 4 √ − 1. This is a complicated
10−2 5
number. But our interest is only in checking whether it exceeds 1 or
equivalently, whether √ 4 √ > 2. This reduces further to whether
10−2 5
q √ √ √
2 > 10 − 2 5, i.e. whether 4 > 10 − 2 5, √ or, whether 4 + 2 5 > 10.
Finally, this will be the case if and only if 5 > 3. But that is false
since 5 < 9. Hence (B) is false.
But there is a much easier way to show that for n = 5, R > r. In fact,
we do not need the value of sin 36◦ . As observed above, the ratio Rr
decreases as n increases. For n = 6 the ratio is 1, i.e. R = r. This can
be seen from (1) since sin π6 = 21 . But even without (1), it is possible
to see that if R = r, then n = 6, from a very common pattern in which
7 discs of equal radii are arranged with one disc at the centre and the
remaining six aligned along the periphery of the central disc, as in the
case of the queen at the centre of a carrom board, surrounded by six
pieces of the same size as the queen, touching one another and the
queen and forming a ring around the queen. So Rr = 1 for n = 6 and
hence Rr > 1 for all n < 6.
In (C), we need the value of sin π8 . This can be calculated from
the value of cos π4 = 12 using the identity sin2 θ = 1−cos
2
2θ
. But as our
r 1
interest is merely in checking whether R < √2−1 for n = 8, we can
once again use that as n increases, the ratio Rr decreases strictly. In
disproving (A) we actually showed that for n = 4 the ratio Rr equals
√ 1 . So, for any n > 4 it is less than √ 1 . In particular this holds
2−1 2−1
for n = 8. So, (C) is true.
This approach will not work for (D) though because we now want
to test whether for n = 12, the ratio Rr is greater than √2(√13+1) and

54
 so for this approach to succeed, for comparison we need to calculate
r
R
for some value of n greater than 12. As n increases, nπ gets smaller
and finding the trigonometric ratios of smaller submultiples of √an angle
π
gets complicated. So here we calculate sin 12 from cos π6 = r 2
3
using
√
1− 3/2
the identity sin2 θ = 1−cos 2θ
2
π
. Taking θ as 12 π
, we get sin 12 = 2
√
3−1
which simplifies to √ . Hence putting n = 12 in (1),
2 2

R 1
= π −1
r sin 12
√
2 2
= √ −1
3−1
√
2 2
< √ (2)
3−1
√ √
After rationalisation, the R.H.S. equals 2( 3 + 1). Thus we see that
(D) is true.

Basically, a simple but good problem. In many other questions with

four options, the work was repetitious. But in the present problem,
(A) is straightforward and interesting short cuts are available for (B)
π
and (C). For (D), however, we need to calculate the value of sin 12 .
Interestingly, this value also appears in Item (IV) of Q.18 of Paper 1.
But there because in all options it matched with (P), there was no need
to calculate it. But in the present problem, there is no alternative. In
Q.9 of this paper, the problem reduced to sec 40◦ . But there all options
were intervals and so the problem reduced to calculating bounds on
sec 40◦ which has no exact formula. The present problem also gives
the options in the form of inequalities. But the angles involved are
π/4, π/5, π/8 and π/12. Except for the first, their sines and cosines
are not very frequently needed, but they do have exact expressions and
that makes it tempting to calculate them. But only the last one is
mandatory.

Q.13 Let î, ĵ and k̂ be the unit vectors along the three positive coordinate
axes. Let

55
 ~a = 3î + ĵ − k̂
~b = î + b2 ĵ + b3 k̂, b2 , b3 ∈ IR
~c = c1 î + c2 ĵ + c3 k̂, c1 , c2 ∈ IR

be three vectors such that b1 b2 > 0, ~a · ~b = 0 and

    
0 −c3 c2 1 3 − c1
 c3 0 −c1   b2  =  1 − c2 
   

−c2 c1 0 b3 −1 − c3
Then, which of the following is/are TRUE?
√ √
(A) ~a · ~c = 0 (B) ~b · ~c = 0 (C) |~b| > 10 (D) |~c| ≤ 11

Answer and Comments: (B, C, D). The involvement of matrices is

only superficial. In essence the data gives a system of equations in the
five unknowns b2 , b3 , c1 , c2 , c3 . The single matrix equation gives three
equations, viz.

−b2 c3 + b3 c2 = 3 − c1 (1)
c3 − c1 b3 = 1 − c2 (2)
and − c2 + c1 b2 = −1 − c3 (3)

Since ~a = 3î + ĵ − k̂, and ~b = î + b2 ĵ + b3 k̂, the vanishing of their dot

product gives one more equation, viz.

3 + b2 − b3 = 0 (4)

So we have four equations. There is also the inequality b2 b3 > 0. But

it only says that b2 , b3 are non-zero and of the same sign. It is not as
strong as an equation. Moreover the equations are not all linear. So, it
is futile to find the values of the unknowns from them and then answer
which of the given options are true.
But sometimes, the data, even though insufficient to determine the
values of all unknowns, may be sufficient to answer what is asked.
As a trivial example, in the present problem, we can get the value of
b22 − 2b2 b3 + b23 from (4) even though we cannot determine b2 and b3
separately. Less trivially, if in our problem we were given any angle

56
(other than a right angle), between ~a and ~b, then since we know ~a we
can determine |~b| even though we cannot determine ~b.
In the present problem, if we multiply (1) by c1 , (2) by c2 and (3) by
c3 , and add, we get the equation

0 = (3c1 + c2 − c3 ) − (c21 + c22 + c23 ) (5)

i.e.

|~c|2 = ~u · ~c (6)
√ √
where ~u is the vector 3î + ĵ − k̂. |~u| = 9 + 1 + 1√
= 11. By Cauchy
-Schwarz inequality the R.H.S. is at most |~u||~c| = 11|~c|. Hence
√
|~c|2 ≤ 11|~c| (7)
√
which implies |~c| ≤ 11. Thus (D) is true. Also (A) is false as otherwise
(5) would imply that ~c is a null vector, which is not always the case.
Instead of c1 , c2 , c3 , if we multiply (1), (2), (3) by b1 (= 1), b2 and b3
respectively and add, then also the sum on the L.H.S. vanishes and we
get

0 = 3 − b1 c1 + b2 − b2 c2 − b3 − b3 c3 (8)

which, because of (4), gives b1 c1 + b2 c2 + b3 c3 = 0. Hence (B) is true.

In the vector ~b = b1 î + b2 ĵ + b3 k̂. b1 is given as 1 and because of (4),
we can express either one of b2 and b3 in terms of the other. We choose
b3 = 3 + b2 . Then

|~b|2 = 1 + b22 + (b2 + 3)2

= 10 + 2b22 + 6b2
= 10 + 2b2 (b2 + 3)
= 10 + 2b2 b3 (9)
√
As we are given b2 b3 > 0, we get |~b|2 > 10, whence |~b| < 10. That
makes (C) true too.
Summing up all options except (A) are true.

57
 We first took a linear combination of (1),(2),(3) with coefficients
c1 , c2 , c3 which, along with (4), enabled us to settle the options (A) and
(D). Later we took a linear combination with coefficients b1 , b2 , b3 and
that enabled us to settle (B) and then (C) (using b2 b3 > 0). In both the
cases the L.H.S. of the linear combination vanished. Both might seem
as clever tricks that somehow worked. But if we keep in mind that
c1 , c2 , c3 are the components of ~c while b1 , b2 , b3 are the components of
~b, then the same work can be cast in a more elegant form.
For this approach to work, we must cleverly recast the data. In
the present problem if we combine (1),(2),(3), together they give an
equation of column vectors.
   
−b2 c3 + b3 c2 3 − c1
 c3 − c1 b3  =  1 − c2  (10)
   

−c2 + c1 b2 −1 − c2

The
 column
 vector
 on
 the R.H.S. can be split into two column vectors
3 −c1
 1  and  −c2 . This is nothing but ~ a − ~c.
   

−1 −c3
More skill is needed in recasting the L.H.S. Keeping in mind that b1 = 1,
it is simply the cross product of the column vectors ~c and ~b. Hence the
first part of the data can be paraphrased as a vector equation
~c × ~b = ~a − ~c (11)
If we take the dot product of the L.H.S. with either ~c or with ~b, we get
0. And that is exactly what we did earlier without using vectors.
It may, of course, be argued that recasting the column vector of the
L.H.S. of (10) as a cross product is as tricky, if not trickier, than the
earlier approach of taking suitable linear combinations of (1), (2) and
(3). This is a matter of unending debate and more a matter of opinion.
Q.14 For x ∈ IR, let the function y(x) be the solution of the differential
equation

dy π
+ 12y = cos( x) y(0) = 0.
dx 12
Then which of the following statements is/are true ?

58
(A) y(x) is an increasing function
(B) y(x) is a decreasing function
(C) There exists a real number β such that the line y = β intersects
the curve y = y(x) at infinitely many points
(D) y(x) is a periodic function.

Answer and Comments: (C). Back to the old game of combining

two (essentially unrelated) problems into a single question. The first is
to find a particular solution of a differential equation and the second,
testing the properties of the solution as a function of the independent
variable.
The first part is quite straight-forwarded. The d.e. is a non-homogeneous
linear d.e. of first order. Its integral factor is e12x . There is a ready-
made formula for its general solution. Finding the particular solution
from this and the initial condition is a mechanical task which we skip.
The final answer is
(12)2 π π π (12)3
 
y(x) = 12 cos x + sin x − e−12x (1)
(12)4 + π 2 12 12 12 (12)4 + π 2

It is the second part of the question that is more interesting. In fact,

the nice thing about it is that it asks only qulitative properties of y(x)
and so a minor slip of computation in the first part will not change the
answer. We can write y(x) as

y(x) = f1 (x) + f2 (x) (2)

where
π π
f1 (x) = A cos x + B sin x (3)
12 12
and f2 (x) = −Ce−12x (4)

where A, B, C are some positive constants. Then regardless of what

these constants are, f1 (x) is periodic (with period 24, although that is
not crucial) and hence bounded on IR. But f2 (x) is not bounded below
since it tends to −∞ as x → −∞. (Note that the exponent and the

59
coefficient −C are both negative.) Hence (D) is false because y(x) is
not even bounded.
For studying the increasing/decreasing behaviour of y(x), we take
πA π πB π
f1′ (x) = − sin x + cos x (5)
12 12 12 12
and f2′ (x) = 12Ce−12x (6)
Then y ′(x) = f1′ (x) + f2′ (x). f1′ is also periodic and bounded. If we
let M and m be, respectively its maximum and minimumon IR, then
M > 0, m < 0 and on any interval of length 24, f1′ (x) assumes all
values between m and M. f2′ (x) is positive for all x, but unbounded
on (−∞, 0) and bounded on (0, ∞). It tends to 0 as x → ∞. So,
for any R ∈ IR, there are intervals of length 24 in [R, ∞) on which
y ′(x) assumes positive as well as negative values since the contribution
from f2′ (x) can be made arbitrarily small. So y(x) cannot be either
increasing or decreasing on IR. Thus (A), (B) are both false.
As at least one option is correct, it has to be (C). But as in Q.11, we
give an honest proof of it. We have to show that there exists some
β such that the line y = β cuts the graph of y(x) in infinitely many
points. As a simplest choice we take β = 0. Then y(x) = 0 holds if and
only if f1 (x) = −f2 (x). It suffices to draw the graphs of y = f1 (x) and
y = −f2 (x) and check in how many points they intersect each other.
Both these graphs are shown in the figure below.

y= −f (x)
2

y=K
y = f 1 (x)

O x

y = −K

60
 It is clear that every (upper) arc of the graph y = f1 (x) after some
stage, (that is, after −f2 (x) becomes smaller than the height of the
arc) cuts the graph of y = −f2 (x) in two points. As there are infinitely
many such arcs, there are infiniely many points of intersection, which
also give points where y(x) = 0.

Actually, many other values of β will work. The equation f1 (x) +

f2 (x) = β is equivalent to f1 (x) = g2 (x) where g2 (x) = −f2 (x)+β. The
graph of g2 (x) is obtained by raising the graph of y = −f2 (x) vertically
by β. It is obvious that any β in the interval (−K, K) will work where
K is the maximum of f1 (x).
The second part is an excellent test of theoretical calculus. It should
have been a full length question in a subjective type of examination.
Finding all values of β for which the line y = β cuts the graph of y(x)
in infinitely many points would itself be a good question.
It seems that the paper setters have a fascination for the number
π
12
and its trigonometric functions. It has appeared twice in Paper 1
(Q.2 and Q.18) and also in Q.12 above in this paper. In the present
π
occurrence, as well as in the occurrences in Paper 1, 12 does not arise
naturally (as it does in Q.12 above). It is essentially an arbitrary choice.

SECTION 3
This section has FOUR questions. Each has four options of which only
one is correct. There are 3 marks for choosing only the correct option, 0
marks if no option is chosen and −1 mark in all other cases.

Q.15 Consider 4 boxes, where each box contains 3 red balls and 2 blue balls.
Assume that all 20 balls are distinct. In how many different ways can
10 balls be chosen from these 4 boxes so that from each box at least
one red ball and one blue ball are chosen?

(A) 21816 (B) 85536 (C) 12096 (D) 156816

Answer and Comments: (A). Boxes and balls are a standard setting
for many combinatorial problems. Usually balls of the same colour are
to be regarded as indistinguishable from one another. In the present

61
problem, it is stipulated that all 20 balls are distinct. Still, by habit,
some candidates may make a mistake. If they do so, the number of
selections will be fairly small and since none of the options will match
it, there is an implied alert. But given the time constraint on the can-
didates, it is better to change the setting (which is superficial anyway)
to something like there are four teams each having 3 boys and 2 girls
and a team of 10 players is to be formed in which there is at least one
boy and at least one girl from each team. In that case no stipulation
is needed to the effect that all children are different.
Coming to the problem, there are three types of selections, either
4 blue and 6 red balls, 5 blue and 5 red balls, and finally 6 red and 4
blue balls. We calculate the numbers of each type separately and add.
In each case, we use the term privileged box to mean a box from which
more than two balls are taken.

(i) 6 blue and 4 red balls: Here there have to be two privileged
boxes for blue balls. They can be chosen in 6 ways. For each such
choice, only the non-privileged boxes have 2 choices each. As for
the red balls, there is no privileged box and each box has 3 choices.
So the number of selections of type (i) is 6 × 4 × 81 = 24 × 81.
(ii) 5 blue and 5 red balls: Here the privileged box for each colour
can be chosen in 4 ways. So there are 4 × 4 = 16 ways to identify
the pair of privileged boxes. Once the choice is made, for the blue
balls the privileged box has only one choice while the other three
2 choices each. So the 5 blue balls can be chosen in 8 ways. For
the red balls, each box, whether privileged or not, has 3 choices.
So there are 34 = 81 ways to choose 5 red balls. So the number of
selections of type (ii) is 16 × 8 × 81 = 128 × 81.
iii) 4 blue and 6 red balls: Here one blue ball is taken from the
two blue balls in each box. For each box, there are two choices.
So the number of ways to pick 4 blue balls is 24 = 16. As for
six red balls, the distribution can be 3 from one and 1 each from
the other 3 or 2 each from 2 and one each from the other two.
In the first case, the privileged box can be chosen in 4 ways. For
each such choice, all 3 red balls from the privileged box have to
be taken. For the remaining 3 boxes there are 3 choices for each.
So the number of selections of 6 red balls in which one box gives

62
 3 red balls is 4 × 3 × 3 × 3 = 108. In the second case, the two
privileged boxes can be chosen from the four boxes in 6 ways. For
each such choice, each box, whether it is privileged or not, has
3 ways ro contribute its share. So the number of selections of 6
red balls with two privileged boxes is 6 × 34 = 6 × 81 = 486. Put
together the selection of 6 red balls can be made in 108+486 = 594
ways. As there are 16 ways to chooe the blue balls, the number
of selections of type (iii) is 16 × 594 = 9504.

Adding, the total number of selections is 24 × 81 + 128 × 81 + 9504 =

152 × 81 + 9504 = 12312 + 9504 which comes to 21816, which tallies
with (A).

The problem is basically a simple counting problem. But the number

of cases is rather large and the calculations in each case are repetitious
and liable to errors if done by hand. The answer can be verified by
other methods. But they also run into a large number of cases. So it
is hardly practicable in a severely time constrained test.
Nevertheless, we mention a method in which the work done above
can be translated algebraically. The idea is to associate to each box a
polynomial in two variables, viz. (2x + x2 )(3y + 3y 2 + y 3 ). Here the
coefficient of any power, say xr is the number of ways r blue balls can
be chosen (subject to the restrictions imposed by the problem). Thus
the constant term, i.e. coefficient of x0 is 0 because at least one blue
ball from the box is to be chosen. That of x is 2 because there are two
distinct blue balls in the box and the number of ways to select one of
them is 2. The coefficient of x2 is 1 because there is only one way to
pick two balls, viz. take them both. In the factor 3y + 3y 2 + y 3, the
coefficient of y r is the number of ways for choosing r red balls. Thus
the coefficient of y 2 is 3 because out of the three distinct red balls, two
can be chosen in 3 ways. As there is only one way to choose 3 balls (as
there are only three balls), we have the term y 3 . This polynomial gives
a complete coding of the possible choices of balls from the box and is
called the enumerator for the box.
Enumerators are especially useful when several boxes are involved
because the enumerator for the two boxes together is simply the product
of their enumerators. If the restrictions on the boxes are different, their

63
enumerators will differ too. In the present problem, all four boxes have
the same restrictions and hence the same enumerators. Therefore, for
the given problem the enumerator is

E(x, y) = (2x + x2 )4 (3y + 3y 2 + y 3 )4 (1)

The way this works is as follows. Suppose we want the number of

ways to pick 6 blue and 4 red balls. This is equivalent to finding the
coefficient of x6 y 4. If we simplify E(x, y) it becomes

E(x, y) = x4 y 4 (2 + x)4 (3y + 3y 2 + y 3)4 (2)

The term x4 y 4 is a reflection of the fact that at least four blue and at
least four red balls are present in each selection. So the coefficient of
x6 y 4 is the same as the coefficient of x2 in

f (x, y) = (2 + x)4 (3 + 3y + y 2 )4 (3)

The first factor is (2+x)(2+x)(2+x)(2+x). When fully expanded, it is

the sum of 16 terms, which are then grouped together according to the
powers of x they contain. That is precisely the binomial expansion. The
only way x2 can appear in this product is that we select x in two of these
four factors and the constant term, viz. 2 in the remaining two. (This
corresponds to saying that we pick two blue balls, i.e. one additional
blue ball from two each of the four boxes and no additional ball from
the
  remaining two.) The two factors contributing x can be chosen in
4
2
ways. The product of the constant terms from the remaining two
factors is 4. We must further multiply this by the constant term from
the other factor (3 + 3y + y 2)4 . This is 34 = 81. Hence the total number
of selections of 6 blue and 4 red balls is 6 × 4 × 81 = 24 × 81. This is
exactly the same count we got for (i) above.
It can be similarly be checked that selections with 5 blue and 5 red
balls correspond to the coefficient of x5 y 5 in E(x, y) and hence to the
coefficient of xy in f (x, y) and similarly, in (iii) the answer is the coef-
ficient of y 2 in f (x, y). (In (iii), in the expansion of (3 + 3y + y 2 )4 , y 2
can occur in two ways: either by taking y 2 from one of the four factors
and the constant term from the other three or by taking 3y from two
of the six factors and the constant term from the remaining two. The

64
 first corresponds to selection with one box giving 3 red balls and the
other to the selection in which two boxes give two red balls each.
It will be seen that the work involved is the same. But we are work-
ing now with the mathematical essence stripping off verbal details like
boxes, balls and colours.
5 3
" #
Q.16 If M = 2 2 , then which of the following matrices is equal to
− 23 − 12
M 2022 ?
" # " #
3034 3033 3034 −3033
(A) (B)
"
−3033 −3032 # "
3033 −3032 #
3033 3032 3032 3031
(C) (D)
−3032 −3031 −3031 −3030

Answer and Comments: (A). Obviously, it is unthinkable to calcu-

late M 2022 by direct multiplication. We look for short cuts. Note that
|M| = 1. So the determinant of all powers of M is also 1. The paper-
setters have been careful to ensure that the matrix in every option has
determinant 1. So, none of them can be eliminated on this preliminary
ground.
There are very few types of matrices whose powers can be written down
without much computation. All diagonal matrices are of this kind.
Special cases of this are the zero matrix O and the identity matrix
I. For a 2 × 2 matrix M," if it corresponds
# to some complex number,
x −y
i.e. if it is of the form , where z = x + iy is a complex
y x
number, then M n corresponds z n , whose real and imaginary parts can
be written down by writing z in the polar form, as r cos θ + ir sin θ. By
DeMoivre’s rule z n comes out to be r n cos nθ + ir n sin nθ. But in the
present problem, M is not of this type.
Then there are some matrices whose very definitions are in terms of
some equality of their powers. Two common examples are the idempo-
tent matrices and the nilpotent matrices. A matrix A is idempotent
if A2 = A and nilpoent if for some positive integer k (not necessarily
the order of the matrix A), Ak = O, the zero matrix. In the first case,
all powers of A equal A, in the second case, they all vanish after some

65
stage. As a result if M is of the form I +A, then its powers can be writ-
ten down very easily using the binomial theorem (which is, in general
not valid for matrices, but is valid here since the matrix I commutes
with any matrix). Indeed
! !
n n 2 n 3
(I + A) = I + nA + A + A + . . . + nAn−1 + An (1)
1 2

If A is idempotent, the sum of the R.H.S. is simply I + (2n − 1)A. If A

is nilpotent, all terms after the k-term vanish.
To see if either of these is applicable here, we put
5 3
" # " #
2 2
1 0
A = M −I = −
− 23 − 12 0 1
3 3
" #
= 2 2 (2)
− 32 − 32

By a direct computation, A is nilpotent. In fact, A2 = O, the zero

matrix. Hence applying (1) with n = 2022 and k = 2, we get

M 2022 = "I + 2022A

3 3
# " #
1 0 2 2
= + 2022
0 1 − 32 − 32
" # " #
1 0 3 3
= + 1011
0 1 −3 −3
" #
3034 3033
= (3)
−3033 −3032

Hence (A) is true.

The essence of the solution above was to express a high power of M

in terms of M. This worked only for some special marices. There is a
general result, called the Cayley Hamilton theorem which says that
a square matrix of order M satisfies a polynomial equation of degree n.
By repeated applications of this result, we can express any power M k
for k ≥ n in terms of only the powers M 0 (= I), M, M 2 , . . . , M n−1 . This
is usually very complicated. For matrices of order 2, this polynomial
can be written down from the trace and the determinant of the matrix

66
M. In the present problem, we already saw that the determinant of M
is 1. Its trace is 52 − 21 = 2. The Cayley Hamilton theorem gives

M 2 − 2M + I = O (4)

Hence M 2 = 2M − I. This is like a recurrecne relation for M n and can

be solved to express M n as an M + bn I for suitable constants an and
bn . But this is a lot more complicated than the solution above (which
became simple because of the strong nilpotency property of the matrix
M − I).
For those who are familiar with the Cayley Hamilton theorem, its
failure in the present problem can be analysed further. The equation
(4) is called the characteristic equation of M and its roots charac-
teristic roots or eigenvalues of M. In the present problem, 1 is a
repeated eigenvalue. If the two eigenvalues, say λ1 and λ2 are distinct,
then it can be shown that M is diagonalisable, i.e. there is some
non-singular 2 × 2 matrix U such that
" #
−1 λ1 0
U MU = (5)
0 λ2

Moreover, once we know the eigenvalues, finding the matrix U does

not take much work. Once we do that any power of M can be calcu-
lated using the fact that any power of a diagonal matrix can be easily
calculated. Indeed, we can rewrite (5) as
" #
λ1 0
M =U U −1 (6)
0 λ2

Note that
" # " #
2 λ1 0 −1 λ1 0
M = (U U )(U U −1 )
0 λ2 0 λ2
" #2
λ1 0
= U U −1 (7)
0 λ2

and more generally, for every positive integer k,

67
 " #k
k λ1 0
M = U U −1
0 λ2
" #
λk1 0
= U U −1 (8)
0 λk2

Since U, λ1 and λ2 are already known, we know M k . As said earlier,

in the present problem, this is not applicable because the eigenvalues
of M are not distinct. More generally, we have shown that the powers
of any diagonalisable matrix can be calculated easily, if we know its
eigenvaluves.
An alert reader might notice that I2 , the identity matrix of order
2, also has trace 2 and determinant 1 and hence the same eigenvalues
as our matrix M. But I2 is diagonalisable, indeed it is itself a diagonal
matrix. The difference between their behaviour can be explained by
considering what is called the geometric multiplicity of an eigen-
value. But it will take us too far afield.

Q.17 Suppose that

Box-I contains 8 red, 3 blue and 5 green balls,
Box-II contains 24 red, 9 blue and 15 green balls,
Box-III contains 1 blue, 12 green and 3 yellow balls,
Box-IV contains 10 green, 16 orange and 6 white balls.
A ball is chosen randomly from Box-I; call this ball b. If b is red
then a ball is chosen randomly from Box-II, if b is blue then a ball is
chosen randomly from Box-III, and if b is green the a ball is chosen
randomly from Box-IV. The conditional probability of the event ‘one
of the chosen balls is white’ given that the event ‘at least one of the
chosen balls is green’ has happened, is equal to
15 3 5 1
(A) 256
(B) 16
(C) 52
(D) 8

Answer and Comments: (C). There are two rounds of selection

and for a layman selection would mean selection at the final round
after going through the earlier ones. And then the outcome of the first
selection does not matter. Indeed it is often forgotten. (Who cares
about JEE Main scores once the results of JEE Advanced are out?) So

68
the problem would then seem to ask the conditional probability that
the ball selected was white, given that the green ball was selected. But
that would be absurd, for such a probability is always 0.
The catch is that the statement ‘at least one of the chosen balls is
green’ has to be interpreted to mean that a green ball has been chosen
at either one of the two rounds. Now, a white ball can be chosen only at
the second round provided a green ball was chosen at the first. Hence
the conditional probability that is asked is that of the event ‘white ball
ball was selected, given that a green ball was selected at either the first
or at the second round.
Once this point is understood, the problem is easy. The event of se-
lecting a white ball (which can only occur at the second round) is the
concatenation of the event of selecting a green ball at the first round
followed by selecting a white ball (from Box IV) at the second. Hence
its probability is
5 6 15
P (G ∩ W ) = P (G)P (W ) = × = (1)
16 32 256
The selection of a green ball can occur at the first round (with proba-
5
bility 16 ) or at the second after selecting a red ball or blue ball in first
16
round. Their probabilities are 32 × 15
48
5
= 32 and 163
× 34 = 64
9
respectively
obtained by taking the appropriate concatenations. Therefore
5 9 5 39
P (G) = + + = (2)
32 64 16 64
Hence the desired conditional probability is
P (G ∩ W )
P (W |G) =
P (G)
15/256 15
= =
39/64 39 × 4
5
= (3)
52
which tallies with (C).

The problem is more a test of correct interpretation of the event.

Once that is done, the rest of the work is simple and the calculations
reasonable.

69
Q.18 For positive integer n, define
16 + 5n − 3n2 32 + n − 3n2 48 − 3n − 3n2 25n − 7n2
f (n) = n+ + + +. . .+
4n + 3n2 8n + 3n2 12n + 3n2 7n2
Then the value of n→∞
lim f (n) is equal to

(A) 3 + 43 loge 7 (B) 4 − 34 loge ( 37 )

(C) 4 − 43 loge ( 37 ) (D) 3 + 43 loge 7

Answer and Comments: (B). We first simplify f (n). Ignoring the

starting term, there is some pattern in the numerators as well as in the
denominators of the terms. But the last term does not fit these patterns
as it stands. To make it so, we write its denominator as 4n2 +3n2 . Now
it is clear that there are n terms and for r = 1, 2, . . . , n, the denominator
of the r-th term is 4rn + 3n2 . To make its numerator fit into a pattern
we write it as 25n−4n2 −3n2 and further as 16n+(9−4n)n−3n2 . Now
the numerator of the r-th term can be written as 16r + (9 − 4r)n − 3n2 .
Hence
n
X 16r + (9 − 4r)n − 3n2
f (n) = n +
r=1 4rn + 3n2
n
X 16r + 9
= n ( 2
− 1)
r=1 4rn + 3n
n
X 16r + 9n
= 2
(1)
r=1 4rn + 3n

This is still not in a form where we can calculate the limit. But we see
that the n-th term is of the order of n1 . So, if we take a factor n out
from the numerator and n2 from the denominator, the ratio simplifies
to
Xn
16 nr + 9 1
f (n) = r × (2)
r=1 4 n + 3 n
If we notice that n1 is the length of each subinterval when the unit
interval [0, 1] is divided into n equal parts and nr is a node of this
partition for each r = 1, 2, . . . , n, then f (n) is precisely a Riemann sum
of the function
16x + 9
g(x) = (3)
4x + 3

70
As n → ∞, this tends to the Riemann integral, say,
1 16x + 9
Z
I= dx (4)
0 4x + 3
The evaluation of the integral is routine, especially after writing the
3
integrand as 4 − .
4x + 3
A straightforward integration gives
1 3 1
I = 4x − (ln(4x + 3)
 
0 4 0
3
= 4 − (loge 7 − loge 3)
4
3 7
= 4 − loge ( ) (5)
4 3
which tallies with (B).

The problem is a combination of two problems, the recasting of a

given sum as a Riemann sum of a suitable function and the evaluation
of an integral. In the present problem, the second part is easy. So
the real work is involved in the first part. There is only a faint hint
that integrals are involved as the difference of two logs rarely appears
except in integrals. This problem is a good test of the ability to take
such veiled hints.

71
 CONCLUDING REMARKS

The general trend of combining several unrelated parts into a single ques-
tion continues this year too. There are hardly any questions in number theory
or combinatorial identities. There is a fair number of problems on complex
numbers but none involving complex roots of unity. (Complex cube roots
of unity are useful in identifying the roots in Q.5 of Paper 1, but that is
not asked.) Of the two counting problems, Q.7, based on complementary
counting is interesting. Q.15 in Paper 2 is laborious.
The greatest integer function appears only superficially, to make an option
have a single digit as its answer. On the other hand, logrithms figure in a
disproportionately large number of questions.
The only question in Paper 1 that needs some novel thinking is Q.11 of
Paper 1 (about the edges of a tetrahedron). Q.7 (about tangents to hyper-
bolas) is poorly drafted and will be easy for those who know a relatively less
known property of tangent to a hyperbola. Q.1 is a simple but well designed
trigonometric question. Q.17 about a system of three linear equations is a
novel one.
Comparatively, Paper 2 has more good questions. Here too, the very first
question in trigonometry is well designed. Q.3 is about evaluating a definite
integral using an unusual substitution. Q.11 deserves a mention because
very rarely questions asking for conditions for some complex numbers to
have integral real and imaginary parts have been asked. Both the problems
on differential equations are so-so although the second part of Q.14 has the
potential to be asked as a question on theoretical calculus. Q.13 is a question
where the cross product is applicable but with no hint of it. Q.15 will be
easy for those familiar with applications of enumerators in counting problems.
Q.16 is easy for those who know nilpotent matrices. Q.17 is more a test of
interpreting the data correctly while Q.18 is a good test of a candidate’s
ability to take subtle hints as to what the problem probably involves.

72