Você está na página 1de 81

EDUCATIVE COMMENTARY ON

JEE 2016 ADVANCED MATHEMATICS PAPERS


(First Draft uploaded June 4, 2016)

Contents
Paper 1

Paper 2

38

Concluding Remarks

80

The year 2013 represented a drastic departure in the history of the JEE.
Till 2012, the selection of the entrants to the IITs was entirely left to the
IITs and for more than half a century they did this through the JEE which
acquired a world wide reputation as one of the most challenging tests for entry
to an engineering programme. Having cleared the JEE was often a passport
for many lucrative positions in all walks of life (many of them having little to
do with engineering). It is no exaggeration to say that the coveted positions
of the IITs was due largely to the JEE system which was renowned not
only for its academic standards, but also its meticulous punctuality and its
unimpeachable integrity.
The picture began to change since 2013. The Ministry of Human Resources decided to have a common examination for not only the IITs, but
all NITs and other engineering colleges who would want to come under its
umbrella. This common test would be conducted by the CBSE. Serious concerns were raised that this would result in a loss of autonomy of the IITs
and eventually of their reputation. Finally a compromise was reached that
the common entrance test conducted by the CBSE would be called the JEE
(Main) and a certain number of top rankers in this examination would have
a chance to appear for another test, to be called JEE (Advanced), which
would be conducted solely by the IITs, exactly as they conducted their JEE
in the past.
So, in effect, the JEE (Advanced) from 2013 took the role of the JEE in the
past except that the candidates appearing for it are selected by a procedure

over which the IITs have no control. So, this arrangement is not quite the
same as the JEE in two tiers which prevailed for a few years. It was hoped
that now that the number of candidates appearing for the JEE (Advanced)
is manageable enough to permit evaluation by humans, the classic practice
of requiring the candidates to give justifications for their answers would be
revived at least from 2014, if not from 2013 (when there might not have been
sufficient time to make the switch-over). But this did not happen in 2015
even after a change of regime at the Union Government and has not happened
for 2016 either. A slight improvement is the provision of some partial credit
where a candidate marks some but not all correct answers provided that
no incorrect is marked. Another welcome feature is that the total number
of questions has been reduced and, more importantly, matching pairs has
been dropped. This often resulted in increasing the number of questions
tremendously.
As in the past, unless otherwise stated, all the references made are to the
authors book Educative JEE (Mathematics) published by Universities Press,
Hyderabad. The third edition of this book is now available in the market.
Because of the multiple choice format and many other constraints in papersetting, interesting questions in mathematics are getting rarer. The continuation of these annual commentaries has been possible largely because of
the keen interest shown by the readers. These commentaries are prepared
single-handedly and hence are prone to mistakes of spelling, grammar and occasionally, wrong symbols (but, hopefully, not mistakes of reasoning!). Many
alert readers in the past had pointed out some such errors. They were corrected and the corrected versions were uploaded from time to time. But by
that time their immediate relevance was reduced.
As an experiment, last year, a draft version of the commentary on both the
papers was uploaded. Those readers who noticed any errors in it were invited
to send an email to the author at kdjoshi314@gmail.com or send an SMS
to the author at 9819961036. Alternate solutions and any other comments
were also solicited. This really paid off. So this year too, the experiment
is repeated. The readers may send their comments to the author by e-mail
(kdjoshi314@gmail.com) or by short mobile messages (9819961036).
As in the past, although all candidates have to answer the same questions,
multiple codes are generated by permuting them. So the answer keys have
to be prepared keeping in mind these codes. Since the present commentary
gives not only the answers but detailed solutions, the code hardly matters.
But for easy reference, Paper 1 is in Code 1 while Paper 2 is in Code 9.
2

PAPER 1
Contents
Section - 1 (Only One Answer Correct Type)

Section - 2 (One or More than One Correct Choice Type)

11

Section - 3 (Single Digit Integer Answer Type)

30

SECTION - 1 (Maximum Marks : 15)


This section contains FIVE questions each of which has FOUR options
out of which ONLY ONE is correct.
Marking scheme :
+3 If only the bubble corresponding to the correct answer is darkened
0 If no bubble is darkened
1 In all other cases.

< < . Suppose 1 and 1 are the roots of the equation


6
12
x2 2x sec + 1 = 0 and 2 and 2 are the roots of the equation
x2 + 2x tan 1 = 0. If 1 > 1 and 2 > 2 , then 1 + 2 equals

Q.37 Let

(A)

2(sec tan ) (B) 2 sec

(C) 2 tan

(D) 0

Answer and Comments: (C).


By the quadratic formula, the roots
of the first equation are sec sec2 1 i.e. sec + tan and sec
tan . Since tan is negative, the first root is smaller than the second.
So we get
1 = sec tan and 1 = sec + tan

(1)

By a very similar calculation, the roots of the second quadratic are


tan sec . This time, however, sec is positive and so
2 = tan + sec and 2 = tan sec
From (1) and (2), 1 + 2 = 2 tan .
3

(2)

A very straightforward problem, requiring only the quadratic formula and the trigonometric identity sec2 = tan2 + 1. The only catch
is to determine which of the roots is larger in each case and this requires
that since lies in the fourth quadrant, sec is positive and tan negative. Using periodicity of trigonometric functions, the problem could
have been made to demand a little more work by giving the coefficients
of x in the quadratics as, say 2 sec(3) and 2 tan(2) and giving to
lie in some other interval. But that would not change the spirit of the
problem. The four given options correspond to all possible choices of
the larger roots. So, a candidate who makes a wrong choice will not be
alerted.
Q.38 A debate club consists of 6 girls and 4 boys. A team of 4 members is
to be selected from this club including the selection of a captain (from
among these 4 members) for the team. If the team has to include at
most one boy, then the number of ways of selecting the team is
(A) 380 (B) 320 (C) 260 (D) 95
Answer and Comments: (A). Another very straightforward problem. Let us first form only the teams (without worrying about the
captain). There are two types of teams,!those with no boys and those
6
= 15 teams of the first type.
with only one boy. Clearly, there are
2
For the second type, the boy can be chosen in 4 ways and
! for each such
6
= 20 ways. So
choice, the three girls in the team can be picked in
3
there are 80 teams of this type. Together there are 95 teams. For each
such team, the captain can be chosen in 4 ways. So, in all there are
95 4 = 380 teams, where two teams with the same members but with
different captains are to be distinguished. This is perhaps the only
testing point in an otherwise dull problem. But in verbal problems,
there is sometimes a risk that the correct interpretation may not strike
to a candidate because of language deficiency.
There is an alternate way to do the problem. First fix the captain.
This can be done in 10 ways. Now ask the captain to choose the
remaining team members. If the captain is a boy, all three remaining
members are girls and so the number of teams with his captaincy is
4

6
= 20. As there are 4 boys, there are 80 teams with a male captain.
3
!
5
= 10
If the captain is a girl, she can choose either three girls in
3
!
5
4 = 40 ways. So, for every girl,
ways or two girls and one boy in
2
there are 50 possible teams where she is the captain. As there are 6
girls in all, the number of teams with a female captain is 300. So the
total count is 80 + 300 = 380, the same answer as before.
The second solution is a little more complicated. Although for a
simple problem like this such a comparison is not of much value, the
approach taken in the second solution deserves to be commented. In
essence, the problem involves selection at two stages, once the selection
of 4 members from 10 (subject to the given restrictions) and then a
second selection, viz., that of the captain from those selected at the
first round. The second solution reverses this order. More generally,
suppose we have n objects from which k objects are to be chosen in
the first round and then from these k chosen objects, r objects are to
be chosen at the second round, where we assume that
r
! n. Then
!
k
n
while by

by the first method the number of selections is


r
k
!
!
nr
n
. Equating the two

the second method, the answer is


kr
r
counts we get a combinatorial proof of the binomial identity
!

n
k

n
k
=
r
r

nr
kr

(1)

(An algebraic proof is also possible by simply expanding both the sides.
But the combinatorial proof tells you the logic.)
An interesting special case of (1) arises when r = 1. (This is
also the case in our present problem, except that now there are some
restrictions on the selections.) In this case, we have
!

n1
n
=n
k
k1
k

(2)

from which it follows that if n is relatively prime to k, then it must


5

n
. (See also the last solution to the JEE 1998 problem in
divide
k
Comment No. 5 of Chapter 4.)

Q.39 Let S = {x (, ) : x 6= 0, }. The sum of all distinct solutions


2

of the equation 3 sec x + cosecx + 2(tan x cot x) = 0 in the set S is


(A)

7
9

(B)

2
9

(C) 0 (D)

5
9

Answer and Comments: (C). As we are given that x 6= 0, /2,


all the terms in the equation are defined for every x in the set S.
Multiplying the equation throughout by sin x cos x (which is non-zero
on S), we get an equivalent equation

3 sin x + cos x + 2(sin2 x cos2 x) = 0


(1)
Since both sin x and cos x are present, this cannot be regarded as a
quadratic equation either in sin x or in cos x. So, we have to solve it as
a trigonometric equation. We first rewrite it as

3 sin x + cos x = 2 cos 2x


(2)

3
1
and are familiar
If we divide both the sides by 2, and notice that
2
2
expressions in trigonometry, we get a clue to the solution. Indeed these
numbers are sin(/3) and sin(/6) respectively. So (2) reduces to
cos(x

) = cos 2x
3

(3)

)
3

(4)

whose general solution is


2x = 2n (x

where n is an integer. The two possibilities that arise are

3
2n
and x =
+
3
9
x = 2n

(5)
(6)

For x to lie in (, ), n must equal 0 in (5) and either 0 or 1 in


7
(6). So the given equation has four solutions in S, viz. , ,
and
3 9 9
5
. These add up to 0.
9
From the answer, it is tempting to think that there is a cleverer way
to get it if we can show that the solutions are symmetrically located
about 0, without actually identifying them. This, for example, would
have been the case if the L.H.S. of the given equation were an odd
function of x. As this is not the case, there is probably no easier way.
So this is not one of those interesting types of problems where you can
get the sum of the roots of an equation without actually identifying the
roots.
Q.40 A computer producing factory has only two plants T1 and T2 . Plant
T1 produces 20% and plant T2 produces 80% of the total computers
produced. 7% of the computers produced in the factory turn out to be
defective. It is known that
P(computer turns out to be defective given that it is produced in plant
T1 ) = 10 P(computer turns out to be defective given that it is produced
in Plant T2 ).
where P (E) denotes the probability of an event E. A computer produced in the factory is randomly selected and it does not turn out to
be defective. Then the probability that it is produced in Plant T2 is
(A)

36
73

(B)

47
79

(C)

78
93

(D)

75
83

Answer and Comments: (C). The very words given that suggest
that this is a problem on conditional probability. The paper-setters
deserve to be congratulated for asking a probability problem which does
not involve drawing balls from urns or throwing dice. Instead, they have
posed the problem in a setting that would easily appeal to the young
generation. Superficial as these garbs are, a faulty computer is surely
far less deadly than a cancerous tumour or a poisonous mushroom!
One wishes, however, that the paper-setters were also careful in
giving the fake answers. A layman often thinks of probabilities in terms
of percentages and this can also be justified mathematically. So, let us
7

say, that 100 computers are produced by the factory, with 20 coming
from plant T1 and 80 from T2 . We are given that 7 out of these 100
computers are faulty. The computer chosen at random is among the
remaining 93 ones. Out of these some, say x are from T1 and y from T2 .
y
The desired answer then is the ratio
. As only one of the answers
93
has a denominator which divides 93, it must be the right one. So even
without calculating y, a clever candidate can identify the answer. To
preclude such a sneaky cakewalk, at least one of the fake answers should
have had a denominator like 31.
For an honest solution, one has to calculate y. Let p be the
percentage of faulty computers in T2 . Then the percentage of faulty
computers in T1 is given to be 10p. So the numbers of faulty computers
in T1 and T2 are, respectively, 200p and 80p respectively. The data gives
the equation
280p = 7

(1)

1
So we get p =
. We have y = 80 80p = 78. Hence the correct
40
78
answer is
.
93
In a systematic (albeit somewhat pedantic) approach, the solution
can be paraphrased using Bayes theorem on conditional probability.
Let T1 and T2 also denote the events that the computer comes from
plants T1 and T2 respectively. These are complementary events with
P (T1 ) = 1/5 and P (T2 ) = 4/5. Let F be the event that a computer
from the factory is faulty. We are given that P (F ) = 0.07. Let F
be the complementary event of F . Then P (F ) = 0.93, The desired
probability is P (T2 /F ). By Bayes theorem, we have
P (T2 F )
P (T2 F )
=
P (T2 /F ) =
P (F )
0.93

(2)

Thus the problem is now reduced to calculating P (T2 F ). Since T2


is the disjunction of the mutually exclusive events T2 F and T2 F ,
and we are given P (T2 ), we get
P (T2 F ) =
8

4
P (T2 F )
5

(3)

This reduces the problem to calculating P (T2 F ). For this, we use


the data that
P (F/T1 ) = 10P (F/T2 )

(4)

By Bayes theorem again and the given values of P (T1 ) and P (T2 ), this
becomes
P (F T1 )
10P (F T2 )
=
1/5
4/5

(5)

5
P (F T1 ) = P (F T2 )
2

(6)

i.e.

But F is the disjunction of the mutually exclusive events F T1 and


F T2 . As we are given that P (F ) = 7/100, from (6) we get
7
7
= P (F T1 ) + P (F T2 ) = P (F T2 )
100
2

(7)

1
Thus we have finally obtained the value of P (F T2 ) as
. Putting
50
1
39
4
= . Putting this into (2),
this into (3) we get P (T2 F ) =
5 50
50
78

we get P (T2 /F ) = , the same answer as before.


93
While the second solution is the polished one, it is the first solution
which enables even a layman to solve the problem and often more
quickly. He can aid his thought process by thinking of two disjoint
piles of 20 and 80 computers, with the faulty computers marked with
some red tag. Once he correctly translates the data that the percentage
of such computers in the first pile is 10 times that in the second into
an equation (viz. (1)), not much is left in the problem.
Q.41 The least value of IR for which 4x2 +
(A)

1
64

(B)

1
32

(C)

1
27

1
1, for all x > 0 is
x

(D)

1
25

Answer and Comments: (C). This problem is in sharp contrast with


the last one. The last problem had a very lengthy statement but was
9

very easy to understand and even to solve for a layman. The present
problem has a one line statement. But it takes some patient thinking
to understand what is asked.
1
As a starter, the requirement that 4x2 + 1 for all x > 0 is
x
1
equivalent to saying that the minimum value of the function 4x2 +
x
over the interval (0, ) be at least 1. This is hardly profound. But it
gives the comforting clue that this is essentially a problem of finding
the minimum of a function over an interval. The catch is that this is a
different function for every real and so its minimum may change as
changes. The problem asks us to find the least value of for which
this minimum is at least 1. Clearly, has to be positive. For < 0,
1
1
4x2 + is negative for large x. For = 0, the function is simply
x
x
which is less than 1 for x > 1. So we assume > 0.
Once this is understood, the problem is routine. Let f (x) be the
1
function f (x) = 4x2 + . Differentiation gives
x
f (x) = 8x

1
x2

(1)

1
which vanishes only when 8x3 = 1, i.e. when x = 1/3 . At this
2
point the derivative changes sign from negative to positive and so the
1
function f (x) has its minimum at x = 1/3 . We denote the minimum
2
value by M to stress that it depends on . As cube roots are involved,
in the calculation of M , it helps to rewrite f (x) as
f (x) =

4x3 + 1
x

(2)

By a direct calculation,
1
)
21/3
4
+1
8
1 1/3

M = f (
=

= 31/3
10

(3)

1
. So
The requirement that M 1 is satisfied if and only if 3
27
1
is the least value of for which the given condition holds.
27
As it often happens, the A.M.-G.M. inequality can give a purely
algebraic solution to the problem of minimising a sum of two positive
functions. But as pointed out at the end of Comment No. 6 of Chapter
6, to apply this inequality sometimes one has to recast the sum as the
sum of three functions instead of two. (See also the end of Comment
No. 11 of Chapter 13 for minimisation of 2 cos x + sec2 x.) So, in our
problem, after it is established that > 0, one can rewrite f (x) as
1
1
4x2 +
+
and then the A.M.-G.M. inequality gives that f (x)
2x 2x
1
1
i.e. when x = 1/3 . This
31/3 with equality holding when 4x2 =
2x
2
leads to a quicker, albeit trickier, calculation of M . (The problem is
shockingly similar to the one mentioned at the end of Comment No. 6,
Chapter 6.)
In essence there are two minimisation problems here. The first
one is a bit tedious and prone to numerical errors. The second one is
easy. Fortunately, if a good student (who has understood the problem
correctly) makes a mistake in the first part, it is unlikely that his answer
will tally with any of the given ones and he will be alerted. Unlike the
last problem, there is no sneaky way to guess the answer. Because
cube roots are involved, a clever student might make a wild guess that
the correct answer is a perfect cube. But the paper-setters have taken
the precaution to include a fake answer which is also a perfect cube.
So this is a rare problem where the multiple choice format helps the
sincere students without rewarding the insincere ones.

SECTION - 2 : (Maximum Marks : 32)


This section contains EIGHT questions.
Each question has FOUR options (A), (B), (C) and (D). ONE OR MORE
THAN ONE of these four option (s) is (are) correct.
Marking scheme :
+4 If only the bubble(s) corresponding to all the correct option(s) is (are)
darkened
+1 for darkening a bubble corresponding to each correct answer provided
11

NO incorrect option is darkened


0 if none of the bubbles is darkened
2 in all other cases.
Q.42 Consider a pyramid OP QRS located in the first octant (x 0, y
0, z 0) with O as the origin, and OP and OR along the x-axis and
the y-axis respectively. The base OP QR of the pyramid is a square with
OP = 3. The point S is directly above the midpoint T of diagonal OQ
such that T S = 3. Then

3
(B) the equation of the plane containing the triangle OQS is xy = 0

(A) the acute angle between OQ and OS is

(C) the length of the perpendicular from P to the plane containing


3
the triangle OQS is
2
(D) the perpendicular
distance from O to the straight line containing
s
15
RS is
2
Answer and Comments: (B, C, D). The first step is to visualise the
pyramid correctly. It is a solid figure and therefore only a rough sketch
is possible on a plane paper as shown below.
z

y
R

3
Q

3
P

12

But because of coordinates, if we can correctly identify the coordinates


of all the vertices, all questions can be answered using appropriate
formulas from coordinate geometry. We are already given that O =
(0, 0, 0), P = (3, 0, 0). As the base OP QR is a square in the first
octant and R lies on the y-axis, the base must lie in the xy-plane with
R = (0, 3, 0) and Q = (3, 3, 0). The midpoint T of OQ is (3/2, 3/2, 0).
All these points are in the xy-plane. The vertex S lies directly above
T and at a distance 3 from it. So S = (3/2, 3/2, 3).
With this spadework, we answer the questions one-by-one. For

3
3
(A), we can consider the vectors OQ= 3~i + 3~j and OS= ~i + ~j + 3~k
2
2
and find the angle between them by taking their dot product. But
it is much easier to do this by simple trigonometry. The angle, say
between OQ and OS is the same as the angle 6 SOT in the right

ST
3
ST
= 1
=
= 2.
angled triangle SOT . Hence tan =
OT
OQ
3 2/2
2

Since tan(/3) = 3 6= tan we see that (A) is false.

For (B) it is hardly necessary to find the equation of the plane


containing the triangle OQS. All we need to verify is whether the
coordinates of the three points O, Q and S satisfy the given equation.
They certainly do and so (B) is correct. For (C), we see from the
diagram that the perpendicular from P to the plane containing the
triangle OQS is P T . Also the triangle OT P is a right angled isosceles
3
triangle. So, P T = OT which we already calculated as . Hence (C)
2
is also correct.
For (D), we first find the area, say , of the triangle ORS. This
can be done by taking OR as a base of length 3 and calculating the
corresponding altitude. But vectors are more convenient, for we get

1
| OR OS |
2


~i ~j ~
k


1

=
0 3 0

2 3 3
3
2
2
1 ~ 9~
9
=
|9i k| =
5
2
2
4

13

(4)

9 9
27
3
+ +9=
=3
. By dividing 2 by
Next, we find RS =
4 4
2
2
RS we get
that
thesperpendicular distance of O from the line containing
9 5 2
15
=
RS is
. Hence (D) is true. An alternate approach is
2
2.3 3
to get the equations of the line RS. Since R = (0, 3, 0) and S =
(3/2, 3/2, 3), the parametric equations of the line RS are
y3
z0
x0
=
=
=t
3/2
3/2
3

(1)

Therefore a typical point X on the line RS is of the form


3
3
X = ( t, 3 t, 3t)
2
2

(2)

for some t. We want t for which OX RS. Taking the dot product of

the vectors RS and OX, this requirement comes out to be


3 3
3
3
( t) + ( )(3 t) + 3(3t) = 0
(3)
2 2
2
2
9
1
27
t = giving t = . Therefore, from (2) X =
which simplifies to
2
2
3
1 5
( , , 1). Hence the perpendicular distance OX of O from the line
2 2 s
s
s
1 25
30
15
RS is
+
+1 =
=
. Hence (D) is correct. Those
4
4
4
2
who remember the formula for the perpendicular distance of a point
from a parametrically given line can save some time here. Yet another
approach is to minimise the distance OX as a function of t. This
27 2
reduces to minimising the quadratic function
t 9t + 9. Taking
2
1
derivative, the minimum occurs at t = . The rest of the work remains
3
the same.
The problem is not as lengthy as it appears at first sight. Once a
diagram is drawn, the first three statements can be disposed off almost
by inspection. It is only Part (D) that demands some work. The papersetters this year have been merciful enough to give some partial credit
for a candidate who solves the first three parts correctly but leaves the
last one. This is a welcome change from the past.
14

Q.43 Let f : (0, ) IR be a differentiable function such that f (x) =


f (x)
2
for all x (0, ) and f (1) 6= 1. Then
x
1
(A) lim+ f ( ) = 1
x0
x
1
(B) lim+ xf ( ) = 1
x0
x
2
(C) lim+ x f (x) = 1
x0

(D) |f (x)| 2 for all x (0, 2)


Answer and Comments: (A). Let y = f (x). Then we are given that
y satisfies the differential equation
dy y
+ =2
dx x

(1)

Multiplying both the sides by x, this can be rewritten as


d
(xy) = 2x
dx

(2)

xy = x2 + c

(3)

whose general solution is

where c is an arbitrary constant. (This could also have been obtained by


noting
R dx that (1) is a linear differential equation with integrating factor
e x = eln x = x. But a direct approach, when feasible, is always
better.)
No intial condition is given and so the value of c cannot be determined.
From f (1) 6= 1 we can only infer that c 6= 0. As the function is given
to be defined only for x > 0, we can divide by x and get
y = f (x) = x +

c
x

(4)

The question asks which of the given four statements is(are) true for
this function. We tackle them one-by-one.
15

By direct differentiation of (4),


f (x) = 1

c
x2

(5)

1
for all x > 0. So f ( ) = 1 cx2 which tends to 1 as x tends to 0
x
1
1
from the right. So (A) is true. For (B), xf ( ) = x( + cx) = 1 + cx2
x
x
which tends to 1 as x 0+ . So (B) is false. In (C), x2 f (x) =
c
x2 (1 2 ) = x2 c which tends to c as x 0+ . As c 6= 0, (C) is
x
false. Finally for (D), from (4), the function f (x) is clearly unbounded
in a neighbourhood of 0. So (D) is false too.
The question is not difficult, but puzzling because its purpose is not
clear. It is a combination of two problems, solving a differential equation and then answering some arbitrary questions about the solution.
The first part is too familiar and the second is completely unrelated.
The problem would have been interesting if the properties of the function could be derived directly from the differential equation without
solving it. Combining two mediocre problems only results in a clumsy
and not in an interesting problem.

3 1 2


Q.44 Let P = 2 0
, where
3 5 0
such that P Q = kI where k IR, k
k
order 3. If q23 = and det(Q) =
8

IR. Suppose Q = [qij ] a matrix


6 0 and I is the identity matrix of
=
k2
, then
2

(A) = 0, k = 8
(B) 4 k + 8 = 0

(C) det(P adj(Q)) = 29


(D) det(Qadj(P )) = 213
Answer and Comments: (B, C). The matrix P has one unknown
entry, viz. . The given relationship between P and Q, viz.,
P Q = kI
16

(1)

introduces one more unknown, viz. k. Taking determinants of both


the sides,
det(P )det(Q) = k 3
(since I has order 3). But we are also given that det(Q) =
k 6= 0. Putting this into (1), we get
det(P ) = 2k

(2)
k2
and
2
(3)

By an easy direct calculation, det(P ) = 15 3 + 20 = 12 + 20. So


we now have one equation in the two unknowns and k, viz.
6 + 10 = k

(4)

We need one more equation in and k to determine their values. This


is given indirectly by specifying that the entry q23 of the matrix Q is
k
. To make out anything from this we need to know the entries of Q
8
from those of P . If we take the adjoint of P , then we have
P adj(P ) = det(P )I = (12 + 20)I

(5)

As k 6= 0, P is non-singular. Hence multiplying (1) and (5) by P 1 we


get
Q = kP 1

(6)

adj(P ) = (12 + 20)P 1

(7)

and

respectively. Combining these two equations we can express Q in terms


of adj(P ) as
Q=

k
adj(P )
12 + 20

(8)

It is easy to calculate the entries of adj(P ). We are interested only


in the entry in the second row and the third column of it. It is the
cofactor of the entry in the third row and the second column of P , i.e.
17

the cofactor of the entry 5 in the third row. To obtain it, we delete
the third row and the second column
of#P , take the determinant of the
"
3 2
the resulting 2 2 submatrix
, and multiply by (1)3+2 , i.e.
2
k
by 1. This comes out to be 3 4. So the data q23 = translates
8
into
k
k
=
(3 4)
(9)
8
12 + 20
As k 6= 0, we get
2(3 + 4) = 3 + 5

(10)

This determines uniquely as 1 and shows that option (A) is incorrect. Putting = 1 into (4) gives k = 4. With these values we see
that (B) is true.
The remaining two options deal with the determinants of the adjoints
of P and Q. From (3) we know that det(P ) = 8. Since P adj(P ) = 8I,
k2
we get det(adj(P )) = 82 = 26 . Since Q has determinant
= 8, we get
2
that the determinant of the adjoint of Q is also 26 . So, both P adj(Q)
and Qadj(P ) have the same determinants, viz. 8 26 = 29 , showing
that (C) is true and (D) false.
Problems on adjoints of matrices are rarely asked. Computation of
the entire adjoint of an n n matrix is a laborious job even for n as low
as 3. The present problem requires the calculation of only one entry in
the adjoint and so the work is reasonable. However, some candidates
are likely to confuse adjoints with Hermitian adjoints of matrices, which
are totally unrelated to each other.
Knowledge of the adjoint of a matrix is essential in answering (C)
and (D) because the very statement of these options involves adjoints.
However, in answering (B), knowledge of the adjoint was only instrumental and not essential. A direct calculation of is possible by a
k
brute force method using the data that q23 = . Since q23 appears
8
in the third colum of Q, we calculate the third
column
of the product

P Q and equate it with the column vector 0 . This gives a system


k
18

of three equations in q13 , q23 , q33 (out of which q23 is already known as
k
), viz.
8
k
+ 2q33 = 0
8
2q13 + q33 = 0
5k
and 3q13 +
= k
8
3q13 +

(11)
(12)
(13)

k
k
. Putting this into (12) gives q33 = . Putting
8
4
these into (11) we get
From (13), q13 =

3k k
k
+
=0
8
8 2

(14)

On simplification and cancellation of k (which is given to be non-zero,


this gives = 1. Combining this with (4) (which too, did not need
adjoints) we get k = 4 and so we are in a position to say that (B) is
true. Of course, this derivation is not as elegant as the earlier one based
on adjoints. But it is elementary and allows those who do not know
adjoints to pocket some partial credit (which has been introduced for
the first time) if they do not answer the remaining parts.
Q.45 In a triangle XY Z, let x, y, z be the lengths of the sides opposite to the
sx
sy
sz
angles X, Y, Z respectively and let 2s = x+y+z. If
=
=
4
3
2
8
, then
and the area of the incircle of the triangle XY Z is
3

(A) area of the triangle XY Z is 6 6


35
6.
(B) the radius of the circumcircle of the triangle XY Z is
6
X
Y
Z
4
(C) sin sin sin =
2
2
2
35


X
+
Y
3
(D) sin2
=
2
5
Answer and Comments: (A, C, D). A triangle XY Z is completely
determined by its three sides x, y, z. So, if we know all three sides
19

we can calculate any attribute, be it the angles, the circumradius, the


inradius or the area of the triangle. In the present problem, the sides
are not specified. The given system of equations, viz.
sy
sz
sx
=
=
4
3
2

(1)

can only determine the ratios of the sides to each other. That is, the triangle is uniquely determined upto similarity. And this is good enough
for answering (C) and (D). However, for (A) and (B) we need the additional piece of data that that the area of the incircle of the triangle is
8
which is a twisted way of saying that its inradius, generally denoted
3
s
8
. (Incidentally, area of a circle is the familiar but old
by r, equals
3
language. Nowadays, it is more customary to call it the area bounded
by a circle.)
Let us see first what we get by setting each of the ratio in (1) equal to
some k. Adding the three resulting equations we get 3s(x+y+z) = 9k
i.e. s = 9k. From this we get
x = 5k, y = 6k, z = 7k

(2)

This is sufficient to tackle (C) and (D) using some well-known formulas
for the sines and cosines of half the angles of a triangle in terms of its
sides. Specifically, for (C) we use
X
sin
=
2

(s y)(s z)
yz

(3)

Y
Z
and two similar formulas for sin and sin . If we multiply all these
2
2
three formulas, every factor in the numerator as well as in the denominator appears twice and so the radicals drop out. Also the k factor
cancels out. Thus
sin

X
Y
Z
sin sin
2
2
2

(s x)(s y)(s z)
xyz
432
=
567
4
=
35
=

20

(4)

X +Y
Z
= cos which,
2
2 s
s
s
s(s z)
92
3
by yet another well-known formula, equals
=
=
.
xy
56
5
Squaring, we see that (D) is also true.

Hence (C) is true. (D) is even easier because sin

As remarked earlier, for (A) and (B), we need the additional


information that
r=

8
3

(5)

To put this piece of information to use, we need to express r in terms


of the sides x, y, z. There are various formulas for this. For example
X
v 2
u
u (s y)(s z)
= (s x)t
s(s x)

r = (s x) tan

(s x)(s y)(s z)
s

(6)
(7)

With our present data, after squaring both the sides we have
8
4k 3k 2k
8
=
= k2
(8)
3
9k
3
which means k = 1. Hence we now have x = 5, y = 6, z = 7 and s = 9.
Herons formula for the area of a triangle, viz.
q

s(s a)(s b)(s c)

gives = 9 4 3 2 = 6 6. Thus (A) is correct.


=

(9)

Finally, for (B), there are many formulas for the circumradius R
of a triangle. But as we already know the area , the best formula to
use is
xyz
(10)
R=
4
As all the values on the R.H.S. are known, we get
R =

567
35
=
46 6
4 6
21

(11)

So (B) is false.
The problem is a test of the knowledge of the umpteen number
of trigonometric formulas for a triangle. But more importantly, it is a
test of the ability to select the right formula when several formulas are
available.
Q.46 A solution curve of the differential equation
(x2 + xy + 4x + 2y + 4)

dy
y 2 = 0. x > 0
dx

passes through the point (1, 3). Then the solution curve
(A) intersects y = x + 2 exactly at one point
(B) intersects y = x + 2 exactly at two points
(C) intersects y = (x + 2)2
(D) does NOT intersect y = (x + 3)2
Answer and Comments: (A, D). The first three options all deal with
the curves involving x + 2. This is a built-in hint that the substitution
u = x+2 may simplify the problem. This is borne out by the coefficient
dy
of
in the given differential equation because it factors out as
dx
du
dy
dy
(x + 2)(x + 2 + y). Since
= 1,
is the same as
. So with this
dx
dx
du
substitution. the differential equation becomes
y2
dy
= 2
du
u + uy

(12)

The R.H.S. is the ratio of two homogeneous polynomials of equal degrees (viz. 2) and so we use the substitution y = vu.
dy
dv
=u +v
du
du

(13)

v 2 u2
v2
dv
+v = 2
=
du
u + u2 v
1+v

(14)

and (12) becomes


u

22

which simplifies to
u

dv
v2
v
=
v =
du
1+v
1+v

(15)

Separating the variables,


1
du
(1 + )dv =
v
u

(16)

Integrating we get the general solution as


v + ln v + ln u = c

(17)

where c is some constant. In terms of y this becomes


ln y +

y
=c
u

(18)

It is customary at this stage to go back to the original variables and


rewrite the solution as
y
ln y +
=c
(19)
x+2
However, since the options involve x + 2, it is better to keep (18) as
it is and work with y and u instead of with y and x. After all this
only amounts to shifting the origin to the point (0, 2) and so does not
affect the intrinsic geometric properties of any curve. In terms of the
new coordinates, the point (1, 3) becomes (3, 3) and we are given that
it lies on the curve (18). This gives ln 3 + 1 = c and so the equation of
the solution curve (in the new coordinates u and y) is
ln y +

y
= 1 + ln 3
u

(20)

We cannot solve this explicitly for y. So we have to resort to adhoc methods to test the various options. In (A) and (B), we have to
take the points of intersection of the curve (20) with the straight line
y = u. Solving simultaneously, we have ln y = ln 3, i.e. y = 3. So (20)
intersects the line y = u at the unique point (3, 3). Therefore (A) is
true and (B) is false. (The existence of the point of intersection was
already given in terms of the initial condition. But now we have proved
its uniqueness.)
23

In (C), we have to take the intersection of (20) with the parabola


y = u2 . At every such point of intersection we shall have
ln u2 + u = 1 + ln 3

(21)

The requirement that x > 0 translates into u > 2. Consequently, u2 > 4


and so the L.H.S. of (21) is bigger than ln 4 + 2 and hence bigger than
than the R.H.S. So equality can never hold in (21) for u > 2. Therefore
the solution curve and the part of the parabola y = u2 , u > 2 cannot
intersect. Thus (C) is false.
For (D), the method is similar but a little more complicated. The
points of intersection, if any, of (20) with y = (u + 1)2 must satisfy
2 ln(u + 1) +

(u + 1)2
= 1 + ln 3
u

(22)

1
= 1 + ln 3
u

(23)

which simplifies to
2 ln(u + 1) + u +

For u > 2, the first term is bigger than 2 ln 3 and hence than ln 3 (as ln 3
is positive). The two remaining terms of the L.H.S. are both positive
for u > 2 and hence certainly bigger than 1. So equality cannot hold
in (23) for u > 2. Hence the solution curve and the curve y = (u + 1)2
do not intersect. So (D) is true.
This problem is of the same spirit as Q.43 above. In both, the first
part was to solve a given differential equation and the second part was
to test the properties of the solution. But both the parts of the present
problem are considerably more complicated than their counterparts in
Q.43. It is ironic that both the questions are given the same credit.
Q.47 Let f : IR IR, g : IR IR and h : IR IR be differentiable
functions such that f (x) = x3 + 3x + 2, g(f (x)) = x and h(g(g(x))) = x
for all x IR. Then
(A) g (2) =

1
15

(B) h (1) = 666 (C) h(0) = 16 (D) h(g(3)) = 36

Answer and Comments: (B, C). The relationship g(f (x)) = x does
not automatically mean that the function g is the inverse function of
24

f . For this to happen, one must first show that the function f (x) =
x3 + 3x + 2 is one to one and onto. The first part follows by Lagranges
Mean Value Theorem, because f (x) = 3x2 + 3 > 0 for all x IR
which means that f (x) is strictly increasing and hence one to one on
IR. That it is onto follows from the fact that f (x) as x .
So, the function f indeed has an inverse function f 1 and the equality
g(f (x)) = x is sufficient to conclude that g = f 1 .
There is no easy explicit formula for g(x) because such a formula
would require solving the cubic equation x3 + 3x + 2 = y for every
y IR and there is no easy way to do this. However, by inspection we
see that f (0) = 2 and so g(2) = 0. We are asked g (2). From the chain
rule
g (f (0))f (0) = 1

(1)

1
= . So we see that (A) is false.
3
The remaining three statements are about the function h. Since g =
1
f it is a bijection (i.e. one-to-one and onto). Therefore the composite
function g g : IR IR is also a bijection. Since h(g(g(x))) = x for
all x IR, we get that h is the inverse function of g g. Since g 1 = f ,
by the formula for the inverse of a composite function we get

Hence g (2) = g (f (0)) =

f (0)

h = (g g)1 = g 1 g 1 = f f

(2)

Hence by the chain rule


h (1) = f (f (1))f (1)

(3)

Since we have f (1) = 6 and f (6) = 3 36 + 3 = 111 from (3) we see


that h (1) = 666. So (B) is true.
The truth of (C) is even easier to test. Since h(0) = f (f (0)) =
f (2) = 8 + 6 + 2 = 16, (C) is true.
Finally, for (D), there is no formula for g(3). However, we can use
(2) to get
h g = (f f ) g
= f (f g)
= f
25

(4)
(5)
(6)

where we have used the associativity of the composition of functions


in going from (4) to (5) and the fact that f and g are inverses of each
other in going from (5) to (6). Therefore h(g(3)) = f (3) = 27 + 9 + 2 =
38 6= 36. So (D) is false.
A very interesting and well-designed problem requiring only simple
properties of composition of functions, the inverse functions and the
chain rule. Unfortunately, the Multiple Choice format masks the difference between those who rigorously prove that f has an inverse and
those who simply assume it.

The paper-setters have also done a commendable educative job


through this problem. At the undergraduate level (and definitely more
so at the JEE level) a function is invariably mistaken with the formula
for it. This happens primarily because nearly all the problems deal
with real valued functions of a real variable defined by some formulas.
Non-numerical functions such as the father function rarely figure in any
problems. Actually, many concepts about functions can be illustrated
through such real life examples. For example monogamy is equivalent
to saying that the husband function (defined on the set of married
women) is one-to-one. But such questions are rarely asked in JEE.
Even though all the three functions f, g, h in the present problem
are from IR to IR, their nature is such that there is no explicit formula
for g(x). So the derivations of (2) and (6) would have been very complicated if the functions were tied down to their formulas rather than
being treated as independent entities.
Q.48 The circle C1 : x2 + y 2 = 3, with centre at O intersects the parabola
x2 = 2y at the point P in the first quadrant. Let the tangent to the
circle C1 at P touch two other circles C2 and C3 at thepoints R2 and
R3 respectively. Suppose C2 and C3 have equal radii 2 3 and centres
Q2 and Q3 respectively. If Q2 and Q3 lie on the y-axis, then
(A) Q2 Q3 = 12

(B) R2 R3 = 4 6

(C) area of the triangle OR2 R3 is 6 2

(D) area of the triangle P Q2 Q3 is 4 2

26

Answer and Comments: (A, B, C).


y

C2

.Q

R2

Q3
C3

Before solving the problem, its long-winded statement deserves


some rebuttal. It reminds one of the unending narratives of illiterate
villagers. What is the great idea in specifying the point P as the point
2
of intersection of the circle x2 +y 2 = 3 and the parabola
x = 2y lying in
the first quadrant, instead of giving it directly as ( 2, 1)? A candidate
who has successfully cleared JEE Main can be presumed to know how
to find this point. (If not, that would raise serious doubts about the
validity of JEE Main!). Nothing is gained by putting such mediocre
roadblocks (which are irrelevant to the main core of the problem) right
at the start, except to increase the chances of a wrong start because of a
numerical slip and, at any rate, the loss of some precious time. Sounds
like a person saying that he met his mothers husbands daughter when
what he means is that he met his sister.
Also, it would be best to describe all three circles together in
a single statement as C1 , C2 , C3 are circles with their centres at the
pointsO, Q
on the y-axis, with O = (0, 0) and having
2 , Q3 respectively

radii 3, 2 3 and 2 3 respectively. They have a common tangent

27


which touches them at P, R2 and R3 respectively with P = ( 2, 1).
This covers all the data but in a more succinct and systematic manner.
Now
point P can be easily identified
coming to the solution itself, the
2
as ( 2, 1) by solving the equations x + y 2 = 3 and x2 = 2y simultaneously and choosing the solution for which both
x, y are positive. The
equation of the tangent to C1 at this point ( 2, 1) is

2x + y = 3
(1)

If a circle with centre at a point (0, b) (say) on the y-axis and radius 2 3
is to touch this line, then equating the radius with the perpendicular
distance from this point, we get


b 3




3

=2 3

(2)

which simplifies to |b 3| = 6 and has two solutions, b = 9 and b = 3.


We take Q2 as (0, 9) and Q3 as (0, 3). Then the distance Q2 Q3 is 12
and so (A) is correct.
Similarly we can determine the distance R2 R3 by identifying the
coordinates of the points R2 and R3 . But there is a better way. Let M
be the midpoint of Q2 Q3 . Then M is also the midpoint of R2 R3 and
so
q

R2 R3 = 2MR2 = 2 MQ22 R2 Q22 = 2 36 12 = 4 6


(3)

Hence (B) is true. Now that we know R1 R2 , to get the area of the
triangle OR2 R3 we only need the perpendicular distance of O from
R2 R2 . This distance is the radius of C1 because R2 R3 touches C1 and
C1 is centred at O. Therefore

1
area of OR2 R3 = 4 6 3 = 6 2
(4)
2
which shows that (C) is also correct.
Finally, for the triangle P Q2 Q3 too, we already know Q2 Q3 = 12.
We need the perpendicular distance of P from Q2 Q3 , But as the line
Q2 Q3 is the y-axis, this distance is simply the x-coordinate of P , viz.

2. Hence

1
(5)
area of P Q2 Q3 = 12 2 = 6 2
2
28

which shows that (D) is not correct.


Like Q.42, although the problem can be done using coordinates,
pure geometry methods work better. Of course, the success of these
methods hinges on drawing a good diagram. In the golden days of JEE,
sometimes the question would specifically ask the candidates to draw
a diagram. And even when it did not, the marking scheme would often
allow some partial credit if a diagram was drawn correctly but some
mistake made later. Like the ability to express ones thoughts verbally,
the equally important ability to correctly draw a diagram is a casualty
of the Multiple Choice format.
Q.49 Let RS be the diameter of the circle x2 + y 2 = 1, where S is the point
(1, 0). Let P be a variable point (other than R and S on the circle)
and let the tangents to the circle at S and P meet at the point Q. The
normal to the circle at P intersects a line drawn through Q parallel to
RS at a point E. Then the locus of E passes through the point(s)
(A)

1 1
,
3 3

(B)

1 1
,
4 2

(C)

1
1
,
3
3

(D)

1 1
,
4 2

Answer and Comments: (A, C). Once again, a good diagram makes
you not only understand the
y
problem correctly but may also
inspire solutions. Take P as
(cos , sin ). Then the equaP
tions of the tangent and the
Q
E
normal to the circle at P are

x
S
O
R
x cos + y sin = 1 and y =
tan x respectively while the
tangent at S is simply the line
x = 1.
Solving x cos + y sin = 1 and x = 1 simultaneously, we get
Q = (1,

1 cos
) = (1, cosec cot )
sin

(1)

Hence the equation of the line through Q parallel to RS is


y = cosec cot
29

(2)

Solving this simultaneously with y = tan x gives the point E as


E = (cosec cot cot2 , cosec cot )

(3)

To get the locus of E we need to eliminate between the equations


x = cosec cot cot2
and y = cosec cot

(4)
(5)

To do this, we rewrite (5) as cosec = y + cot . Squaring both the


sides we get
1 = y 2 + 2 cot y = y 2 + 2 cot cosec 2 cot2 = y 2 + 2x

(6)

So the equation of the locus of the point E is y 2 + 2x = 1. By direct


1
1
substitution this is satisfied if x = and y = . So, both (A) and
3
3
1
1
(C) are on the locus. However, when x = , we get y = . Hence
4
2
the points in (B) and (D) are not on it.

SECTION - 3 : (Maximum Marks : 15)


This section contains FIVE questions, each having a single correct answer
which is an integer ranging from 0 to 9 both inclusive.
Marking scheme:
+3 If only the correct bubble is darkened
0 In all other cases.
Q.50 The
total number of distinct x IR for which

x
x2
1 + x3


2
1 + 8x3 = 10 is
2x 4x


3x 9x2 1 + 27x3

Answer and Comments: 2. Clearly the question is about the number of real solutions of a polynomial equation p(x) = 10 where the
polynomial p(x) is specified as a determinant, viz.
x x2
1 + x3
2
p(x) = 2x 4x 1 + 8x3
3x 9x2 1 + 27x3






30

(1)

So the first task is to evaluate this determinant. Taking out x and x2


as common factors from the first and the second columns respectively
we have
p(x) = x3 q(x)

(2)

where
1 1 1 + x3
q(x) = 2 4 1 + 8x3
3 9 1 + 27x3






(3)

Subtracting suitable multiples of the first row from the remaining rows,
q(x) =

1 1 1 + x3
0 2 6x3 1
0 6 24x3 2

= 48x3 4 36x3 + 6
= 12x3 + 2

(4)

Hence the equation to be solved is x3 (12x3 + 2) = 10 which reduces to


6x6 + x3 5 = 0

(5)

Factoring the L.H.S. as (6x3 5)(x3 + 1) we see that the only solutions
5
are those where x3 = or x3 = 1. Each of these equations has only
6
one real solution, Hence the given equation has two real solutions.
A clever student can save some time by dispensing with the actual
roots of (5). Treating it as a quadratic in x3 , as the discriminant is
positive, there are two distinct real roots. Each of these has only one
real cube root and hence (5) has exactly 2 solutions.
The starting step in the expansion of the determinant (1) is too
simple to be missed by anybody. However, although the evaluation
of the determinant in (3) as given here is simple enough, an alternate
solution is to split the third column and write the determinant as a
sum of two determinants, viz.
1 1 1 1 1 x3

q(x) = 2 4 1 + 2 4 8x3

3 9 1 3 9 27x3






31

(6)

If we cyclically permute the columns of the first determinant, it becomes


a Vandermonde determinant (see Exercise (3.26)). If we take out the
factor x3 from the third column of the second determinant and the
factors 2 and 3 from its second and third rows respectively, it also
becomes a Vandermonde determinant. Thus,

1 1 1
1 1 1


3
q(x) = 1 2 4 + 6x 1 2 4


1 3 9
1 3 9

(7)

These two Vandermonde determinants are the same and their common
value is (21)(31)(32) = 2. This gives an alternate derivation
of (4). But the direct derivation is faster.
A fairly simple problem on determinants.
Q.51 Let m be the smallest positive integer such that the coefficient of x2 in
the expansion of (1 + x)2 + (1 + x)3 + . . . + (1 + x)49 + (1 + mx)50 is
(3n + 1) 51 C3 for some positive integer n. Then the value of n is
Answer and Comments: 5. Combinatorial identities have faded out
from the JEE papers after the multiple choice format came in. But in
the present problem the paper-setters have managed to squeeze in one
such identity. The identity itself is not new. Indeed it is exactly the
same identity that appears as the first hint to the Main Problem of
Chapter 5, viz. that for any positive integers m and n with m n,
!

n+1
m
m+1
n2
n1
n
=
+
+ ...+
+
+
m+1
m
m
m
m
m

(1)

the present problem, we let S be the coefficient of x2 in the given


polynomial. Then by the binomial theorem,
!

50
49
3
2
m2
+
+ ...+
+
S=
2
2
2
2

(2)

The last term in the sum is irregular. But the remaining terms can be
summed by writing them in the reverse order and applying (1) with
n = 49 and m = 2. Hence
!

50
50
m2
+
S=
2
3
32

(3)

Thus the data reduces to the equation


!

51
50
50
m2 = (3n + 1)
+
3
2
3

(4)

We now expand the binomial coefficients and cancel the factor 50 49


which appears in all numerators to get
8+

m2
17
= (3n + 1)
2
2

(5)

m2 = 51n + 1

(6)

which reduces to

The problem is now reduced to a problem in number theory. We are


given that m is the smallest positive integer for which there exists some
positive integer n such that the equation above holds true. Simply
stated, we want the smallest positive integer n for which 51n + 1 is a
perfected square. This can be done by trial. For n = 1, 2, 3 or 4 we do
not get a perfect square. But for n = 5, 51n + 1 = 256 = (16)2 . So the
correct answer is 5.
A slightly different approach (although not entirely free of trial and
error) is to rewrite (6) as
(m + 1)(m 1) = 51n

(7)

The problem now is to find the least positive integer n for which 51n
can be factored as the product of two numbers that differ from each
other by 2. If 51 had too many factors then the search would take a
lot of time. But fortunately 51 has only two prime factors, viz. 17 and
3. Rewriting 51n as 17 3n we see that n = 5 fits the table.

This is a good problem and an easy one once you hit upon the right
combinatorial identity to apply. In this respect the problem resembles
a trigonometric problem which unfolds itself once you pull the right
trigonometric identity. The difference is that the world of trigonometric
identities useful at the JEE level is limited and well trodden. That can
hardly be said of combinatorial identities. They are used less frequently
and so while it is reasonable to give the candidate a binomial identity
33

and ask him to prove it, it is a little unreasonable to expect him to


pull it from his memory kit. Doing so brings mathematics closer to
chemistry where you simply have to remember what cassiterite is an
ore of. No amount of logical thinking will help you.
Q.52 The total number of distinct x [0, 1] for which
is

t2
dt = 2x 1
1 + t4

Answer and Comments: 1. Consider the difference function


f (x) =

t2
dx 2x + 1
1 + t4

(8)

defined for x [0, 1]. Our job is to find out how many zeros f (x) has
in the interval (0, 1]. As in Q.50, our interest is only in the number
of zeros and not in identifying them. So there may be an easier way
without evaluating the integral explicitly. And indeed there is. Clearly
Z 1
t2
1 < 0 because the interval of
f (0) = 1 > 0 while f (1) =
0 1 + t4
integration has length 1 and the integrand is less than 1 at all its points.
As functions defined by integrals are continuous, f (x) is continuous.
Therefore by the Intermediate Value Property, f (x) vanishes at least
once in the interval (0, 1).
If we could prove that f (x) is strictly decreasing in the interval
[0, 1] that would mean that there are no more zeros. The most standard
method to test this is to check the sign of f (x). By the second form
of Fundamental Theorem of Calculus, we have
f (x) =

x2
2
1 + x4

(9)

As already noted, the first term is always less than 1 for all x [0, 1].
So f (x) < 0 throughout (0, 1) and hence by Lagranges Mean Value
Theorem, f (x) is strictly decreasing in (0, 1). Hence there cannot be
more than one zeros of f (x) in [0, 1]. As we already know that there is
at least one zero, the correct count is 1.
Problems where the number of zeros of a strictly monotonic function
have to be determined using Intermediate Value Property and the Lagrange Mean Value theorem are fairly common in JEE. The additional
34

feature in the present problem is that the derivative is obtained using


the second form of the fundamental theorem of calculus. That too is
quite common.
x2 sin x
= 1. Then 6( + ) equals
x0 x sin x

Q.53 Let , IR be such that lim

0
Answer and Comments: 7. The limit in question is of the form.
0
So application of lHopitals rule is indicated. It gives
2x sin x + x2 cos x
x0
cos x

1 = lim

(1)

Here the denominator tends to 1 while the numerator tends to 0


as x 0. If 1 6= 0, then the limit would be 0, a contradiction. So
we get = 1.
To determine , we could apply lHopitals rule again. But there
is a better way out. The denominator 1 cos x equals 2 sin2 (x/2) But
near 0, sin x is of the same order as x near 0 in the sense that the ratio
sin x
tends to 1 as x 0. So when = 1, the denominator of the
x
x2
function on the R.H.S. of (1) is of the same order as 2(x/2)2 = . As
2
for the numerator, sin x is of the order of x while cos x tends to
1. So the numerator is of the order of 2x2 + x2 = 3x2 . Cancelling
3
the factor x2 the limit on the R.H.S. of (1) is
= 6. So we get
1/2
6( + ) = 6 + 1 = 7.
The argument can be presented rigorously by rewriting the ratio
on the R.H.S. of (1) (with = 1). We divide both the numerator and
the denominator by x2 . Then the ratio equals
2 sinxx + cos x
)2
(1/2)( sin(x/2)
x/2

(2)

Then as x 0, each ratio in this expression tends to 1 and so does


cos x. Therefore this ratio tends to 6.

35

For those familiar with the infinite series expansion of sin x an


argument can be given by rewriting the ratio as
3 x3 5 x5
x2 x
+
...
3!
5!
x3 x5
( 1)x +

+ ...
3!
5!

For the ratio to tend to a finite non-zero limit, the dominating powers
in the denominator and the numerator must be the same. This forces
= 1, and, afterwards, the limit equals the ratio of the dominating
terms, which is 6.
A fairly simple problem. Although a rigorous justification is time
consuming, in a multiple choice format it is only the numerical work
that matters and the computations involved are light.

1 + 3i
where i = 1 and r, s {1, 2, 3}. Let P =
Q.54 Let z =
#2
"
(z)r z 2s
and I be the identity matrix of order 2. Then the total
z 2s
zr
number of ordered pairs (r, s) for which P 2 = I is
Answer and Comments: 1. It is more customary to denote z by .
It is called the complex cube root of unity and in the polar form it equals
e2i/3 . Because of the factorisation of z 3 1, it satisfies z 2 + z + 1 = 0.
Also powers of z recur with a cycle of 3. That is, z 1 = z 2 = z 5 = z 8 =
. . . etc. Most of the problems about complex cube roots of unity can
be handled with these two basic properties. (See Comment No. 12 in
Chapter 7 for a few such problems. Also see Part (C) of Q.59 in the
first paper of JEE Advanced 2015.)
In the present case, a direct computation gives
#"

P2 =

"

(z)r z 2s
z 2s
zr

"

z 2r + z 4s
(1)r z r+2s + z r+2s
(1)r z r+2s + z r+2s
z 4s + z 2r

36

(z)r z 2s
z 2s
zr

(1)

Equating this matrix with I results into a system of two equations,


viz.
z 2r + z 4s = 1
and (1)r z r+2s + z r+2s = 0

(2)
(3)

Since z and its powers are all non-zero, the second equation can hold
only when (1)r = 1, i.e. when r is odd. In the given set {1, 2, 3},
r has to be 1 or 3. We determine the possible values of s in each
case separately from the first equation.. When r = 1, we must have
1 + z 2 + z 4s = 0. This means z 4s = z. So, 4s 1 is divisible by 3. In
the set {1, 2, 3} this is possible only for s = 1. So (1, 1) is a possible
pair.
In the other possibility, viz. r = 3 we have z 2r = z 6 = 1 and so by
equation (2) above z 4s = 2. But this is not possible for any s because
|z| = 1 and so all powers of z are also of absolute value 1.

So there is only one pair. The problem is much simpler and hence
less interesting than last years problem referred above. Apparently
the question was motivated by a perceived compulsion to ask some
question about complex numbers.

37

PAPER 2
Contents
Section - 1 (Only One Correct Answer Type)

38

Section - 2 (One or More than One Correct Answers Type)

55

Section - 3 (Paragraph Type)

75

SECTION - 1 : (Maximum Marks : 18)


This section contains SIX questions, each having FOUR answers only one
of which is correct.
Marking scheme :
+3 If only the bubble corresponding to the correct answer is darkened.
0 If no bubble is darkened
1 in all other cases.

1 0 0

Q.37 Let P = 4 1 0 and I be the identity matrix of order 3. If


16 4 1
q31 + q32
Q = [qij ] is a matrix such that P 50 Q = I, then
equals
q21
(A) 52 (B) 103 (C) 201 (D) 205
Answer and Comments: (B). Whatever be the other criticism levelled upon the JEE paper-setters, even their worst critic would not
accuse them of being so inhuman as to require the candidates to calculate the fiftieth power of a 3 3 matrix P by actual multiplication 49
times. Multiplication of even one pair of matrices is such a horrendous
job. So brute force is not the right method. There are two standard
ways to proceed to find such a high power. One is to calculate some
lower powers by actual multiplication, notice some pattern in their entries, make a guess about the n-th power and finally prove the guess
by induction on n (or by some other method). Since JEE doesnt care
about proofs any more, the last step is irrelevant. So he who makes a
38

correct guess gets the reward.


If we follow this strategy then by a direct calculation, the first three
powers come out to be

1 0 0

P = 4 1 0
16 4 1

1
0 0
8
1 0
=

16 + 32 8 1

P2

1
0 0

12
1 0
=

16 + 32 + 48 12 1

and P 3

(1)

The pattern is sufficiently clear. Only three entries change. Two of


these are equal and are in an A.P. The third one, sitting in the bottom
left corner is 16 times the sum of the first n integers, for which we have
a formula. So we guess that for every positive integer n,
Pn

1
0 0

4n
1 0
=

8n(n + 1) 4n 1

(2)

An honest student would try to prove this by induction as follows. The


truth for n = 1 is evident from the very definition of P . As for the
inductive step,
n
P n+1 = P
P

1 0 0
1
0 0

4n
1 0 4 1 0
=

16 4 1
8n(n + 1) 4n 1

1
0
0

4n + 4
1
0
=

8n(n + 1) + 16n + 16 4n + 4 1

1
0
0
4(n
+
1)
1
0
=

8(n + 1)(n + 2) 4(n + 1) 1


39

(3)

This completes the proof of the inductive step. Hence we have now
established that (2) holds for every positive integer n. Putting n = 50
we get
P 50

1
0 0

200
1 0
=

400 51 200 1

(4)

To finish the solution we are given that

Q = [qij ] = P 50 + I

(5)

Mercifully, unlike matrix multiplication, matrix addition is done entrywise. To make the things even simpler, the entries of Q we are
interested in are all outside the diagonal and all non-diagonal entries
of the identity matrix are 0. So we simply read out the corresponding
entries of P 50 and get
q31 = 400 51
q32 = 200
and q21 = 200

(6)

q31 + q32
400 51 + 200
=
= 102 + 1 = 103.
q21
200
Another method of finding high powers of a matrix without brute
force is to relate it to some other matrices whose high powers are known
from definition or because of their relationship with some other entities
such as complex numbers (see Comment No. 20 of Chapter 2). Two
classes of such matrices are (i) nilpotent matrices and (ii) idempotent
matrices. By definition, an n n matrix A is nilpotent if Ar is the
zero matrix O for some positive integer r while it is called idempotent
if A2 = A. Note that if A is nilpotent then all powers of A after some
stage are O while if A is idempotent then all powers are equal to A
itself.
So, by sheer arithmetic,

In the present problem, we write P as I + A where


I is the identity

0 0 0

matrix of order 3 and A is the matrix 4 0 0 . By a direct calcu16 4 0


3
lation, we see that A = O, the zero matrix. So A is nilpotent. (This
40

is true of every n n matrix which has zero entries on and above its
main diagonal. But the proof is a bit tedious. An elegant proof can be
given using what are called linear transformations. But that is well
beyond our scope.)
Since P = I + A, P 50 = (I + A)50 . We have to be wary in hastily
expanding this by the binomial theorem. The binomial theorem is valid
for the power of a sum of two real (or complex) numbers because the
multiplication for real or complex numbers is commutative. In fact, the
most direct proof is by writing (a+ b)n as a sum of 2n terms of the form
x1 x2 . . . xn where each xi equals a or b and then grouping together the
like terms using the fact because of commutativity of multiplication,
every term in which exactly r factors are
! a and the remaining n r
n
terms of this type for every
are b equals ar bnr , there being in all
r
r = 0, 1, 2, . . . , n. But matrix multiplication is not commutative. So, if
A and B are two square matrices of the same order then we can expand
(A + B)2 only as (A + B)(A + B) = A2 + AB + BA + B 2 . However,
if A and B commute with each other, then we can write (A + B)2
as A2 + 2AB + B 2 . More generally, the binomial theorem holds for
(A + B)n for any positive integer n if the matrices A and B commute
with each other.
Coming back to P 50 = (I + A)50 , as the identity matrix commutes
with every matrix we can apply the binomial theorem to conclude
P

50

=I

50

50 48 2
50 49
I A + . . . + I nr Ar + . . . + A50
I A+
+
2
1

(7)

All powers of I equal I. As for powers of A, all powers after A2 vanish.


So only the first three three terms in the expansion above matter and
give
P 50 = I + 50A + 1225A2

(8)

Q = P 50 + I = 2I + 50A + 1225A2

(9)

Hence
By a direct calculation,

0 0 0
0 0 0
0 0 0

A2 = 4 0 0 4 0 0 = 0 0 0
16 0 0
16 4 0
16 4 0
41

(10)

We substitute these in (9). As before, we get q31 , q32 and q21 by adding
the corresponding entries on the R.H.S. This gives q31 = 0 + 50.16 +
1225.16 = 1275.16 = 400 51, q32 = 0 + 50.4 + 0 = 200 and q21 =
0 + 50.4 = 200. These are the same values that we got in (6). So the
answer is the same.
A good problem about matrices. But a bit above the JEE level.
Nilpotent matrices are studied only in degree courses. So most candidates will be forced to try only the first approach and sadly one again,
the scrupulous candidates who not only guess but prove (2) by induction on n will lose in terms of time. The question therefore deserves a
better place in the conventional examination as a question on a proof
by induction.
In passing we mention two other instances where the high powers
of a matrix can be calculated without brute force. First, analougously
to (7), if P = I + A where A is an idempotent matrix, then from the
binomial theorem and the fact that Ar = A for all r 1, we get
P n = (I + A)n = I + (2n 1)A

(11)
"

x y
,
As the other example, if P is a 2 2 matrix of the form
y x
then P corresponds to the complex number z = x + iy. By writing z
as r e i, we get (using DeMoivres rule) that
n

P =r

"

cos n sin n
sin n cos n

(12)

Dont be surprised if problems based on (11) or (12) are asked in the


coming JEEs ! But now that the secret is out, maybe they wont be
asked!
Q.38 Area of the region {(x, y) IR2 : y
equal to
(A)

1
6

(B)

4
3

(C)

3
2

|x + 3|, 5y x + 9 15} is
(D)

5
3

Answer and Comments: (C). A typical problem of finding the area


between two curves of the form y = f (x) and y = g(x) by integrating
42

the difference function |f (x) g(x)|. When the vertical boundaries of


the region are specified in the form x = a and x = b for some a, b IR,
with a < b, the integration is over the interval (a, b). But when this is
not the case, the first task is to find the points of intersection of the two
given curves because the x-coordinates of these points of intersection
give us the interval over which to integrate.
y

B
F

D
C

The present problem falls in the latter category. The upper boundary
of the region is the straight line (shown by L in the figure above)
5y = x + 9

(1)

while the lower boundary is given as the curve


y=

|x + 3|

(2)

Because of the absolute values sign, we have to consider two cases: (i)
where x + 3 > 0, i.e. x 3 and (ii) where x + 3 < 0, i.e. x 3.
For (i), i.e. for x 3, (2) can be rewritten as
y2 = x + 3

(3)

which is a parabola with vertex at E = (3, 0). We are interested only


in the upper half of this parabola. It cuts the line (1) at points where
y 2 5y + 6 = 0, i.e. where y = 2 or y = 3. The corresponding values
of x are x = 1 and x = 9. So the points of intersection of (1) and the
43

upper half of (3) are (2, 1) and (9, 3). These are shown as B and H
in the figure below. Note that in between these two points, the line
lies below the parabola and so the portion between them is not inour
region. For x > 9, the line again lies above the parabola. This is true
for all x > 9 and would result in an unbounded region. But we are
specifically given that x + 6 < 15 and that precludes any part of this
unbounded region.
Summing up, the subregion for which x 3 shown as the shaded
part bounded by E B F E has area A1 where
A1


x+9
=
x + 3 dx
5
3
"
#
1
(x + 9)2 2

=
(x + 3)3/2
3
10
3
16
100 36 16

=
=
10
10
3
15
Z

(4)

Let us now turn to the other subregion (bounded by E F D E).


Here the calculations are very similar. The upper boundary is the same
line L. But the lower boundary is the upper half of the parabola
y 2 = x 3

(5)

which is the mirror image of the parabola (3). It has the same vertex,
viz. E = (3, 0). To find out where it cuts the line (1), we have to
solve the equation y + 5y = 6 which has two roots, 1 and 6. We ignore
the second value as it will give a point of intersection which lies on the
lower half of the parabola. The first value gives (4, 1) as the (unique)
point of intersection of the line (1) with the upper half of the parabola
(5). It is shown as D in the figure above. So, the area of the second
subregion, say A2 is
A2


x+9
x 3 dx
=
5
4
#
"
3
(x + 9)2 2

+ (x 3)3/2
=
4
10
3
36 25 2
13
=

=
10 10 3
30
Z

44

(6)

3
45
= .
30
2
The integration part of the problem is too straightforward. The real
task is to identify the region correctly. So essentially, this is a problem
in coordinate geometry. The paper-setters have been careful enough
to stipulate the condition x + 9 15 which may appear irrelevant at
first sight. (of course, it would have looked more natural to give this
condition as x 6.) Also even though square roots are involved in
the definition of the function, the numerical data has been carefully
chosen to avoid any surds in the answer. This reduces the chances of
numerical error.
Adding (4) and (6) we get the total area of the region as

Q.39 The value of

13
X

k=1

(A) 3

1
sin

3 (B) 2(3

(k1)
6

sin

k
6

is equal to

3) (C) 2( 3 1) (D) 2(2 + 3)

Answer and Comments: (C). Just as finding the fiftieth power of a


matrix is a torture if done directly, so is adding 13 terms, each of which
is the reciprocal of the product of two factors. The angles appearing in
the problem are in an A.P. and there are formulas for the sum of their
sines (see Comment No. 6 of Chapter 7). But these are inapplicable
here. So here we apply a general rule of thumb. When there is no
obvious way to find the sum of a series of a large number of terms,
recast the series as a telescopic series. Problems based on this trick
are popular with paper-setters of competitive examinations and have
appeared several times in the JEE too. See for example, the JEE 1999
problem in Comment No. 11 of Chapter 5. Also see the comments on
Q.28 of Paper I of JEE 2010 and Q.47 of Paper I of JEE 2013.
So our job is to express the k-th term of the series in the form
Ak Bk in such a way that Ak+1 will equal Bk for k = 1, 2, . . . , 12. In
the present problem the k-th term is the reciprocal of the product of
two sines, i.e. of the form sin sin . It is tempting to rewrite this as
1
half the difference of two cosines, viz. as (cos()cos(+)). And
2
this would have worked had the term sin sin been in the numerator
(with the denominator a constant). But unfortunately, it is in the
denominator.
45

So, we look for identities that have the product of two sines in the
denominator. One such identity is
cot + cot =

cos cos
sin( + )
+
=
sin
sin
sin sin

(1)

But as our interest is to express the general term of the series as a


difference and not as a sum of two terms, we replace by in (1)
and get
cot cot =

cos cos
sin( )

=
sin
sin
sin sin

(2)

So, we apply this identity with = (k 1) + and = k + .


6
4
6
4

Then is simply which is independent of k. Dividing both the


6
1
sides of (2) by sin( ) which equals , we get
2
1
sin

(k1)
6

sin

k
6

= 2 cot (k 1)

2 cot k
6
6


(3)

The job is almost done. Now it is only a matter of adding these terms
for k = 1 to k = 13 and noting that the R.H.S. forms a telescopic series.
Thus if we denote the given sum by S, we have
S =
=

13
X

1


(k1)
6

k=1

sin

13
X

2 cot((k 1)

k=1

= 2 cot
4
= 2 cot

sin

k
6



+ ) 2 cot(k + )
6
4
6
4


13
2 cot
+
6
4
29
2 cot
12

(4)

To finish the solution we only need to calculate the values of these two
cotangents. The first one is 1. For the second one, we need to work a
29
little. First, since cot is periodic with period , cot
is the same as
12
46

5
cot . In terms of degrees, this equals cot 75 or equivalently, tan 15 .
12
This value is not needed as frequently as tan of some other angles such
as 30 or 45 . To calculate it, we let u = tan 15 . Then the identity for
tan 2 gives
1
2u
tan 30 = =
1 u2
3

(5)

This gives a quadratic for u, viz.

u2 + 2 3u 1 = 0
(6)

whose (positive) root


is 2 3. Putting these values into (4) we finally
get S = 2 2(2 3) = 2( 3 1).
A tricky but otherwise an easy problem once you get the trick. One
wishes that the numerical data was so arranged as not to require the
value of tan 15 . This value is irrelevant to the main theme. Having to
calculate it gives an advantage to those who remember such rarely used
values listed in some tables over those who prefer to remember only a
few standard values and derive the rest as and when needed. So, like
Q.51 in Paper I, the last part of the question brings it closer in spirit
to a question in chemistry.

Q.40 Let bi > 0 for i = 1, 2, . . . , 101. Suppose log b1 , log b2 , . . . , log b101 are in
Arithmetic Progression (A.P.) with common difference loge 2. Suppose
a1 , a2 , . . . , a101 are in A.P. such that a1 = b1 and a51 = b51 . If t =
b1 + b2 + . . . + b51 and s = a1 + a2 + . . . + a51 , then
(A) s > t and a101 > b101
(A) s < t and a101 > b101

(B) s > t and a101 < b101


(A) s < t and a101 < b101

Answer and Comments: (B). The first statement is an indirect


way of specifying that the numbers b1 , b2 , . . . , b101 are in a geometric
progression (G.P.) with common ratio 2. Any candidate who knows
the most basic property of logarithms, viz.
y
ln y ln x = ln
(1)
x
will know this. And once again, candidates who have cleared the JEE
Main can be presumed to belong to this category. So, one fails to
47

know what is gained by giving this twist to the data. Or is it that the
paper-setters are allergic to geometric progressions?
Anyway, coming to the problem, the question is about the relative
rates of growths of geometric progressions and arithmetic progressions.
An A.P. is like a car which runs at a uniform speed, so that it covers equal distances in time intervals of equal durations. A geometric
progression (with common ratio greater than 1), on the other hand, is
like a car which keeps accelerating. Suppose we track two such cars,
A and B respectively by recording their positions at regular time intervals. Suppose that at the start both the cars are together. (That
does not mean that they are at rest. They are both moving but pass
the same roadmark simultaneously.) Suppose we notice that half an
hour later they are together again. What do we make out? Evidently,
their average speeds over this half hour are the same. But in the case
of A, this is also its uniform speed every moment. For B, on the other
hand, intially the speed is less than this average speed and so the car
is lagging behind A. But later its speed picks up and the distance between them begins to decrease and becomes 0 half an hour after the
start. Obviously, at this point the speed of B is much higher than that
of the (constant) speed of A. So, after this half an hour, B will beat A
throughout and the gap between them will go on increasing forever.
All this is clear by sheer common sense. And a candidate whose
common sense is sound can easily tell that (B) is the correct option. In
fact, he can predict that till n = 50, an > bn (the car B lagging behind
A). So not only s > t but even term by term, the summands of s are
bigger than the corresponding summands of t, except the first and the
last summands which are equal for both. In other words, s beats t not
only in the aggregate but in each term. And for all n > 51, an < bn
(car B is ahead of car A).
The present problem asks you to put this common sense into a
mathematical form. an and bn are like the positions of the cars at the
end of the n-th time interval (or to be fussy, at the end of the (n 1)th time time interval from the start, because the progressions begin
with a1 and b1 rather than with a0 and b0 which would have been more
natural). Let d be the common difference of the A.P. a1 , a2 , . . . , a100 .
(In terms of the analogy above d is like the uniform speed of the car

48

A.) Then
an = a1 + (n 1)d
while bn = b1 2n1 = a1 2n1

(2)
(3)

for all n = 1, 2, . . . , 101. Equating a51 with b51 gives the value of d as
d = a1

250 1
50

(4)

With this, (2) becomes


an = a1 + a1 (n 1)

250 1
50

(5)

We now attempt to prove that an > bn for n < 51 while an < bn for all
n > 51. Since a1 > 0, it cancels out and these comparisons reduce to
showing that
250 1
2n1 1
>
for 2 n 50
50
n1
250 1
2n1 1
and
<
for n > 51
50
n1

(6)
(7)

Note that equality would hold in both (6) and (7) for n = 51. This is
a consequence of the data that a1 = b1 and a51 = b51 . If we can prove
these inequalities, that will surely imply (B) because s = a1 + a2 + . . . +
a50 + a51 while t = b1 + b2 + . . . + b51 .
Unfortunately, a direct proof of these inequalities is not easy. The
paper-setters have apparently realised this and made the problem less
ambitious by requiring to prove merely that s > t and a101 < b101 . This
is a lot easier because of the formulas for the sums of the terms in an
A.P. or G.P. Indeed, from the formula for the sum of the terms in an
A.P., we get
s = a1 + a2 + . . . + a51 = 51a1 +
250 1
25 51a1
= 51a1 +
50
250 + 1
= 51a1
2
49

50.51
d
2

(8)

As for t, it is the sum of the first 51 terms of a G.P. with common ratio
2 and the first term a1 . So
t = a1 (251 1)

(9)

Clearly, t < a1 251 while from (8), s > a1 249 51 > a1 251 because surely
51 > 4. Put together, s > t.
The comparison of a101 with b101 is much easier. Indeed, from (5),
a101 = a1 + 2a1 (250 1) = a1 (251 1) < a1 251 which is a lot smaller
than a1 2100 which is precisely b101 .
Although that completes the solution, one nagging question remains.
Why is it so difficult to prove the inequalities (6) and (7) when their
counterparts in the analogy given above are so obvious by common
sense, viz. that during the first half an hour, the accelerating car B
always lags behind the car A while thereafter B leads A?
The explanation is that there is a suble flaw in the analogy. The
motion of a car is continuous. Its position is a function of time t which
is a continuous variable. On the other hand, progressions, and more
generally all sequences are functions of a discrete variable n. As a
result, the speed of a moving car has no analogue for progressions,
since speed is defined in terms of derivatives and derivatives (and more
generally limits) are meaningless for functions of discrete variables. If
mangos are falling off a tree, the instantaneous rate of their fall has
no meaning. But if these mangos are squeezed and made into a pulp
and the pulp is pored from a jar then the instantaneous rate of flow
of the pulp is a well-defined entity. This happens because the number
of mangos fallen is a discrete variable while the volume of the mango
pulp is a continuous variable.
However, sometimes we can extend a discrete variable by allowing it to assume all real values (or all non-negative real values) when
the nature of a function is such that it makes sense for a continuous
variable as well. Doing so allows us to apply the powerful machinery
of continuous mathematics such as theorems like the Lagranges Mean
Value Theorem. And then it may be possible to prove some results
about sequences which would not be easy to prove directly.
We illustrate this technique by giving a proof of the inequality
(6). The proof of (7) is similar and left to the reader. Let us first get
50

rid of a1 which has no role. So we take a1 = 1. Also let us take the


progressions with n starting from 0 instead of 1 (which is more natural
for many purposes.) In effect this means that we call n 1 as our new
n. So instead of (3) we now have bn = 2n . Now, we replace the discrete
variable n (taking values 0, 1, 2, . . .) by a continuous variable x taking
all non-negative real values. The function 2x makes sense for all such
values. We call it f (x). Thus,
f (x) = 2x

(10)

The inequality (6) now translates as


f (50) f (0)
f (x) f (0)
>
for 0 < x < 50
50 0
x0

(11)

To prove this, we note that the function f (x) = 2x is strictly concave


upwards on [0, ) because f (x) = (ln 2)2 2x > 0 for all x 0. Now
consider the points P1 = (0, f (0)), P2 = (50, f (50)) and P3 = (x, f (x))
on the graph of y = f (x). Then because of the concave upwards nature
of the function, the slope of the chord P1 P3 is less than the slope of the
chord P1 P2 (see Comment No. 18 of Chapter 13). But thats precisely
(11).
Where did we need Lagranges MVT? As shown in Comment
No. 11 of Chapter 16, the property of concave upwards functions just
mentioned is a consequence of LMVT. So, we see that an ostensibly
simple inequality about progressions had to be derived using methods
from calculus.
This is an excellent problem which rewards candidates who have
good intuition and sound common sense. In fact such candidates will
get the answer without doing any numerical work. For a rigorous solution, numerical work is indeed needed. The paper-setters have taken
care to see to it that this work is manageable by making the problem
less ambitious than it could have been.
The enormity of the geometric growth as compared to arithmetic
growth is quite well-known. Suppose a person B makes an offer to a
person A that he will give A Rs. 10000 to start and then every day
will give him 1000 more rupees than on the previous day and that in
return all A has to do is to give B one match stick at the start and then
51

every day twice as many matchsticks as on the previous day. From As


point of view this offer looks incredibly lucrative for the first week or
so. But within less than a month he will go bankrupt. This happens
because As earnings form an A.P. but his liability increases in a G.P.
At a more serious level, the Malthusisn economic theory holds that the
resources grow in an arithmetic progression while the population grows
in a geometric progression.
Q.41 The value of the integral
(A)

/2

/2

x2 cos x
dx equals
1 + ex

2
2
2 (B)
+ 2 (C) 2 e/2
4
4

(D) 2 e/2

Answer and Comments: (A). Call the integral as I. There is no


obvious antiderivative of the integrand. So this integral has to be evaluated by special methods by splitting it into two bad integrals, which,
after some suitable transformation, combine nicely into a good integral.
The interval of integration, viz. [/2, /2] is symmetric about the
origin. Had the integrand been an even function of x, the integral over
each half would have been the same. Here that is not
But
Z 0the case.
x2 cos x
that is precisely what leads to the solution. Let us call
dx
/2 1 + ex
Z /2 2
x cos x
dx as I1 and I2 respectively. Then
and
1 + ex
0
I = I1 + I2

(1)

We leave I2 as it is. But we transform I1 with the substitution y = x


Z /2 2
y cos y
into the integral
dy. Replacing the dummy variable y by
1 + ey
0
Z /2 2
x cos x
x, this is the same as
dx. So, from (1) we get
1 + ex
0
/2 x2 cos x
x2 cos x
dx
+
dx
1 + ex
1 + ex
0
0

Z /2 
1
1
+
x2 cos x dx
=
1 + ex 1 + ex
0

I =

/2

52

(2)

Further simplification would be possible if the sum inside the parentheses can be simplified. This is indeed possible, because a straight
2 + ex + ex
computation shows that this sum equals
which is
(1 + ex )(1 + ex )
simply 1 because the denominator expands to 1 + ex + ex + 1 thanks
to the fact that ex ex = 1.
We have now removed the troublesome part of I and have reduced
it to a manageable integral, viz.
I=

/2

x2 cos x dx

(3)

which can be evaluated by two applications of integration by parts.


Specifically,
I =

i /2

x2 sin x

/2
0

2x sin x dx

/2
2

+ (2x cos x)
=
0
4
2
/2


=
+ 0 (2 sin x)
0
4
2

2
=
4

/2

2 cos x dx

(4)

Unlike the last question, there is nothing intellectually inspiring in this


problem. The reduction of I to (2) is a fairly common strategy and
could have been partial credit in a conventional examination. Further
success depends on the ability to simplify the expression in the parentheses. Problems like this reward those who are good at drill. And
that is why such problems are soon forgotten.
Q.42 Let P be the image of the point (3, 1, 7) with respect the plane x y +
z = 3. Then the equation of the plane through P and containing the
x
y
z
straight line = = is
1
2
1
(A) x + y 3z = 0
(C) x 4y + 7z = 0

(B) 3x + z = 0
(D) 2x y = 0
53

Answer and Comments: (C). This is a combination of two problems.


First, identifying the point P and, once it is known, identifying the
plane spanned by it and the given line. Both the parts are routine.
For the first part, call the point (3, 1, 7) as Q. Then the line P Q
is perpendicular to the plane x y + z = 3 and therefore has direction
numbers 1, 1 and 1. Hence the point P is of the form
P = (3 + , 1 , 7 + )

(1)

for some real number . To find its value, we note that the midpoint of

P Q, i.e. the point (3 + , 1 , 7 + ) lies on the plane x y + z = 3.


2
2
2
A direct substitution gives
1 1 1
3 1 + 7 + ( + + ) = 3
(2)
2 2 2
which means = 4. Putting this into (1), we get
P = (1, 5, 3)

(3)

In the second part of the problem, we have to find the equation of the
y
z
x
plane passing through this point P and containing the line = = .
1
2
1
For this we need a vector n perpendicular to this plane. This vector

can be taken to be P A P B where A and B are any two points on this

line. We choose A as (0, 0, 0) and B as (1, 2, 1). Then P A= i 5j 3k

and P B= 2i 3j 2k. Hence


n =

i j
k
1 5 3
2 3 2

= i 4j + 7k

(4)

To find the equation of the plane, instead of P we choose A = (0, 0, 0)


which also lies on it. Then the equation of the plane is
(x 0) 4(y 0) + 7(z 0) = 0

i.e. x 4y + 7z = 0.

(5)

An extremely routine problem. Suitable for JEE Main but hardly


so for JEE advanced. The last problem had at least some excitement
1
1
because of the simplification of the expression
+
. The
x
1+e
1 + ex
present problem is absolutely dull.
54

SECTION - 2 : (Maximum Marks : 32)


This section contains EIGHT questions. Each questions has FOUR options (A), (B), (C) and (D). ONE OR MORE THAN ONE of these four
option(s) is(are) correct.
Marking scheme :
+4 if fully correct
+1 for darkening a bubble corresponding to each correct answer provided
NO incorrect option is darkened
0 if no bubble is darkened
2 In all other cases.
Q.43 Let a, b IR and f : IR IR be defined by f (x) = a cos(|x3 x|) +
b|x| sin(|x3 + x|). Then f is
(A) differentiable at x = 0 if a = 0 and b = 1
(B) differentiable at x = 1 if a = 1 and b = 0
(C) NOT differentiable at x = 0 if a = 1 and b = 0
(D) NOT differentiable at x = 1 if a = 1 and b = 1
Answer and Comments: (A, B). Normally, one checks differentiability of a function at a point by calculating its left and right handed
derivatives at that point and seeing if they are equal. But before doing
so, it pays to simplify the function. In the present problem, cosine is an
even function and therefore regardless of whether |x3 x| equals x3 x
or x x3 , cos |x3 x| will always equal cos(x3 x) which is differentiable everywhere, being the composite of two differentiable functions.
As for the second function, viz. |x| sin(|x3 + x|) we note first of all that
x3 + x = x(x2 + 1). The second factor is always positive. So |x3 + x|
equals x3 + x if x > 0 and x3 x if x 0. The sine function is an
odd function. So, for x > 0 |x| sin(|x3 + x|) = x sin(x3 + x), For x 0,
|x| sin(|x3 + x|) = x sin(x3 x) = x sin(x3 + x). So, regardless of
the sign of x, the given function equals a cos(x3 x) + bx sin(x3 + x).
This is differentiable everywhere. So (A) and (B) are true.
The constants a and b have no role in this problem. So the
question is somewhat mischievous. Its focal point is something else,
viz. composites and products of even and odd functions.
55

Q.44 Let f (x) = lim

Then

nn (x + n)(x + n2 ) . . . (x + nn )
2
2
n!(x2 + n2 )(x2 + n4 ) . . . (x2 + nn2 )

1
(A) f ( ) f (1)
2
(C) f (2) 0

!x/n

, for all x > 0.

1
2
(B) f ( ) f ( )
3
3
f (3)
f (2)
(D)

f (3)
f (2)

Answer and Comments: (B, C). The very sight of the function f (x)
is frightening. The first simplification we do is to take ln f (x) and apply
the fact that the limit of the log of a function is the log of the limit of
that function. (The justification for this comes from the continuity of
the logarithm function. But that is hardly expected at the JEE level.)
x
ln gn (x)
n n

ln f (x) = lim

(1)

nn (x + n)(x + n2 ) . . . (x + nn )
2
2
n!(x2 + n2 )(x2 + n4 ) . . . (x2 + nn2 )

(2)

where
gn (x) =

Note that here x is a fixed positive real number and the limit in (1) is
taken as the variable n tends to . As x is independent of n we can
pull it out in the R.H.S. of (1) and write
ln f (x) = x lim

1
ln gn (x)
n

(3)

1
in (3) gives a clue.
n
1
is the
If the interval [0, 1] is partitioned into n equal parts, then
n
width of each subinterval and the points of subdivision are of the form
r
for r = 0, 1, 2, . . . , n. Therefore if we could recast ln gn (x) in the
n
n
X
r
h( ) for some suitable function h(t) of a variable t (which is
form
n
r=1
independent of x), then the coefficient of x in the R.H.S. of (3) will
be the limit of a Riemann sum of h(t) over this partition as the mesh
Although this is hardly profound, the coefficient

56

of the partition tends to 0. Therefore, it will be the integral

1
0

h(t)dt

and that will take us into familiar waters. (We deliberately choose t
rather than x as the independent variable for the function h because
x is already booked. Of course any other variable such as y, z, u etc.
except x itself will do. Note alsoZthat the function h(t) may have x as
1

h(t)dt will, in general, not be some


a parameter and so the integral
0
number but will be some function of x. Perhaps it is better to denote
the function h(t) by h(x, t) and keep in mind that when it is integrated
w.r.t. t, the variable x is to be treated as a constant. But we stick to
h(t).)
The crucial task now is how to recast ln gn (x) in this form. Keeping
in mind that the logarithm of a product of n factors is the sum of the
logarithms of these n factors, we would be through if we could write
r
gn (x) as the product of n factors of the form u( ) for some suitable
n
function u(t) of t [0, 1] for once we do this, our magic function h(t)
will be simply ln(u(t)).
n
Y

r
u( ) we first ignore the
n
r=1
leading factors nn and n! in the numerator and the denominator of
gn (x) respectively and work with the ratio of their r-th factors, viz.
1
)
(x +
n
(x + r )
(r/n)
. If the
for a typical r. We rewrite this as
2
1
(x2 + nr2 )
(x2 +
)
(r/n)2
n
nagging factors n and n!, which we have ignored so far were not there,
(x + 1t )
then we could take u(t) as 2 1 .
(x + t2 )
To recast gn (x) in the desired form

But we cant simply ignore these factors nn and n!, however inconvenient they may be. So we need to do some further recasting so
that these unwanted factors will get cancelled out. This is done by one
more rewriting. Multiplying both the numerator and the denominator

57

by (r/n)2 , we get
1
)
1 + x(r/n) r
(r/n)
=
1
1 + x2 (r/n)2 n
)
(x2 +
(r/n)2
(x +

(4)

We now take the product of n such ratios, with r running from 1


n
Y
r
n!
to n. Then the product
is simply n which nicely cancels the
n
r=1 n
nn
in gn (x).
unwanted factor
n!
Thus we have finally recast gn (x) in the form we wanted, viz.
gn (x) =

n
Y

1 + x(r/n)
2
2
r=1 1 + x (r/n)

(5)

Taking logarithms,
n
X

1 + x(r/n)
ln
ln gn (x) =
1 + x2 (r/n)2
r=1

(6)

We now let
h(t) = ln

1 + xt
1 + x2 t2

(7)

Then
ln gn (x) =

n
X

r
h( )
n
r=1

(8)

and hence, finally from (3),


ln f (x) = x lim

n
X

1 r
h( )
n
r=1 n

(9)
Z

h(t) dt, i.e.


As already noted the limit on the R.H.S. is the integral
0

Z 1 
1 + xt
ln
the integral
dt. So, at long last we get an expression
1 + x2 t2
0
for ln f (x) which we can manage, viz.
ln f (x) = x

1 + xt
ln
1 + x2 t2

58

dt

(10)

It is tempting at this stage to evaluate the integral on the R.H.S. and


then take the exponentials of both the sides to get an explicit formula
for f (x). But that would be laborious. Further, it may not even be
needed. The question does not ask for specific values of f at any point
but only for comparisons of values at various points and this is often
possible by looking at the sign of the derivative (and using Lagranges
MVT).
So, we try to find f (x) from (10).
f (x)
=
f (x)

1 + tx
ln
1 + t2 x2


d
dt + x
dx

Z

1 + xt
dt
ln
1 + x2 t2


(11)

In the second term on the R.H.S., we are differentiating the integral


w.r.t. the parameter x. But the integration itself is w.r.t. t. It can be
shown that under suitable conditions of smoothness, the derivative of
the integral is the integral (w.r.t. t) of the derivative of the integrand
(w.r.t. x). This is known as differentiation under the integral
sign. It is of the same spirit as the result that the derivative of a
sum is the sum of the derivatives, because, after all, integration is the
limiting form of finite summation. But the proof is far from easy. Even
if we apply it, the result will be a very complicated integral.
Fortunately, there is an easier way to differentiate (10). Suppose we
try to evaluate the integral on the right using the substitution u = tx.
Then keeping
in mind 
that x is a constant, du = xdt and the integral
Z x 
1+u
becomes
ln
du. Putting this into (10) gives
1 + u2
0
ln f (x) =

1+u
ln
1 + u2


du

(12)

The R.H.S. can be differentiated using the second fundamental theorem


of calculus. Thus we get
1+x
f (x)
= ln
f (x)
1 + x2


(13)

We are now ready to dispose off the given options. Note that f (x) > 0

(as otherwise

 ln f (x) is undefined). So the sign of f (x) is also the sign
1+x
2
of ln 1+x
2 . This is positive when 1 + x > 1 + x , i.e. when 0 < x < 1
59

and negative when 1 + x < 1 + x2 i.e. when x > 1. At x = 1, f


vanishes. So we conclude, using Lagranges MVT that f (x) is strictly
increasing on (0, 1] and strictly decreasing on [1, ). So (A) is false
while (B) is true. Similarly, (C) is true since 2 > 1. It only remains to
check (D). Here, for the first time, we apply (13) with specific values
f (3)
4
f (2)
3
4
3
of x. Then
= ln( ) while
= ln( ). Since
< and
f (3)
10
f (2)
5
10
5
f (2)
f (3)
<
.
the logarithm function is strictly decreasing, we have
f (3)
f (2)
Therefore (D) is false.
The principal idea in this problem is to convert the limit of a
sequence whose n-th term is the product of n factors to the limit of a
sequence whose n-th term can be recognised as a Riemann sum of a
suitable function. Problems based on this idea are not uncommon, see
e.g. Exercise (17.9)(c). But in this problem, the terms of the sequence
involve a real parameter x and so the resulting limit is a function of x.
After this (i.e. after reaching (10)), the problem takes a different turn.
The data of the problem is such that this function cannot be evaluated
in an explicit form. So questions about it have to be answered using
derivatives. Here too, calculation of the derivative is fairly tricky. It
is not difficult to go from (10) to (12). But to realise that thats what
is needed to reach (13) is tricky. Those who dont get this trick would
have wasted the time they spent in deriving (10) (which, too, is made
fairly complicated by the factors nn and n! in the numerator and the
denominator of (2) respectively.) A far more reasonable problem would
have been to give (10) as the data and then ask the options. That way,
those who hit the idea of converting (10) to (12) will be able to go
further. Others will not waste the precious time they spent in deriving
(10).
So, as it stands, this is a very difficult, tricky and lengthy problem.
In the past, occasionally such challenging problems were asked. But
that time there was a provision for partial credit. In fact, many times
the candidates would be asked to reach the crucial intermediate step
(such as (10) in the present problem) and then to complete the problem
using that step or otherwise. (Hence or otherwise was a frequent
phrase seen those days.) Even when such an intermediate step was
not specifically asked, the marking scheme would provide some partial
60

credit for it. Also people would talk about such problems and keep a
count of how many candidates answered them fully correctly. There
have been problems for which the numbers of such complete solvers
were single digit numbers. And for the second part of the 1992 JEE
problem given in Comment No. 10 of Chapter 10, nobody got it! Such
discussions and statistics would provide valuable guidance to future
paper-setters and that made the JEE a sort of tradition. With the
multiple choice format all we know is how many candidates answered
a particular question correctly. But there is no way to tell how many
got it through real hard work, how many got it through clever sneaky
work, how many got it by fluke and for how many it was a shot in the
dark.
Q.45 Let f : IR (0, ) and g : IR IR be twice differentiable functions
such that f and g are continuous functions on IR. Suppose f (2) =
f (x)g(x)
g(2) =, f (2) 6= 0 and g (2) 6= 0. If lim
= 1, then
x2 f (x)g (x)
(A) f has a local minimum at x = 2
(B) f has a local maximum at x = 2
(C) f (2) > f (2)
(D) f (x) f (x) = 0 for at least one x IR
Answer and Comments: (A, D). The paper-setters have not given
the functions f (x) and g(x) explicitly. Instead, thet have given as
data a few properties of them from which some conclusions are to be
drawn. When a function is given explicitly, it has so many properties
and a candidate may get lost in deciding which of them are relevant.
(Sometimes, this is the very thing to test, as was the case in Q.43
above.)
Because of the data that g(2) and f (2) both vanish, the limit
f (x)g(x)
0
lim
is
of
the
indeterminate
form
. The most standard (and
x2 f (x)g (x)
0
popular) method to evaluate such limits is to apply lHopitals rule. So
from the last piece of data, we get
f (x)g(x) + f (x)g (x)
=1
x2 f (x)g (x) + f (x)g (x)
lim

61

(1)

Because of continuity of the functions and their first and second derivatives, this limit can be obtained by mere substitution provided it is not
0
of the
form. This is so, because with the substitution x = 2, the
0
denominator becomes f (2)g (2) which is non-zero from the data. So
f (2)
the limit is simply
which is given to be 1. So, f (2) = f (2).
f (2)
This immediately implies that (C) is false and (D) is true. Since f is
given to take only non-negative values, f (2) and hence f (2) are both
positive. So at x = 2, f vanishes while f > 0, Therefore f has a
(strict) local minimum at 2 and cannot have a local maximum at 2.
An extremely simple problem. In total contrast with the last
one. But the paper-setters do not have the freedom to allot credit
proportional to the degree of difficulty. So a mediocre student who
strategically omits problems like the last one and comfortably answers
questions like the present one is at an advantage over an intelligent
student who spends considerable time in solving the last problem partly
but later gives up. That makes JEE a game of strategy.
1

Q.46 Let u = u1i + u2j + k be a unit vector in IR3 and w = (i + j + 2k).


6
Given that there exists a vector ~v in IR3 such that |
v ~v | = 1 and
w (
u ~v ) = 1, which of the following statements are correct?
(A) There is exactly one choice for such ~v
(B) There are infinitely many choices for such ~v
(C) If u lies in the xy-plane then |u1| = |u2 |

(D) If u lies in the xz-plane then |u1 | = |u3|.


Answer and Comments: (B, C). The vector u is given to be a unit
vector whereas w is also a unit vector by direct calculation (or from
the notationwhich is generally used only for unit vectors). We are not
given ~v . But we are given that
|
u ~v | = 1

(1)

w
(
u ~v ) = 1

(2)

and

62

As both |w|
= 1 and |
u ~v | = 1, the only way (2) can hold is when
the angle between w and u ~v is 0, whence they are equal too (as they
have equal lengths. So we get
w = u ~v

(3)

u and w are unit vectors. Also by (3) they are perpendicular to each
other. So there exists a unique unit vector a such that the triple
(
u, w,
a
) forms an orthonormal, right-handed system. (In fact, a
has to
be u w.)
We are looking for all vectors ~v (not necessarily of length
1) such that (1) and (3) hold. Since u (a) = a u = w,
we see that
~v may be taken as a. But as it is not required to be a unit vector,
~v = a +
u will also fit the table for any real . So there are infinitely
many choices for ~v. Thus, (B) is correct and (A) is false.
For the remaining two options we express u ~v in terms of the
components. Thus
i j k

u ~v =
u1 u2 u3
v1 v2 v3
= (u2 v3 u3 v2 )i + (u3 v1 u1 v3 )j + (u1 v2 u2 v1 )k

(4)

From (3) we have a system of three equations :

and

1
u2 v3 u3 v2 =
6
1
u3 v1 u1 v3 =
6
2
u1 v2 u2 v1 =
6

(5)
(6)
(7)

In (C) we have u3 = 0. Then from (5) and (6), u2 v3 = u1 v3 . Also


v3 6= 0 from (5). So, u1 = u2 and (C) holds. In (D), on the other
1
2
hand, we have u2 = 0. So we get u1 v2 = and u3 v2 = . Hence
6
6
|u3| = 2|u1 |. So (D) is false.

A simple problem, not really suitable for an advanced test. It would


have been more interesting to ask to characterise all vectors ~v for which
63

u ~v = w.
They are precisely vectors of the form w u +
u for any
real . But then. such questions are suitable only for conventional
examinations.
Q.47 Let P be the point on the parabola y 2 = 4x which is at the shortest
distance from the centre S of the circle x2 + y 2 4x 16y + 64 = 0. Let
Q be the point on the circle dividing the line segment SP internally.
Then

(A) SP = 2 5

(B) SQ : QP = ( 5 + 1) : 2
(C) the x-intercept of the normal to the parabola at P is 6
1
(D) the slope of the tangent to the circle at Q is
2
Answer and Comments: (A, C, D). Another bunch of problems of
different types, clubbed together with a common data. Reminds one
of passengers sitting in the waiting room at a railway station. Their
destinations are different and they have little relationship with each
other. But just to kill time, they talk on some common theme (usually,
sky-rocketing prices and rampant corruption)!
y

.S

C2

X
P

To begin the solution, by completing squares, we get that the centre


S of the given circle, say C1 , is at (2, 8) and its radius is 2. If P is the
64

point on the parabola y 2 = 4x which is closest to S, then the circle,


say C2 , centred at S with radius CP must touch the parabola at P . So
both C2 and the parabola have the same tangent and hence the same
normal at P . This makes it easy to identify P . Taking the parametric
equations of the parabola y 2 = 4x, vix. x = t2 , y = 2t, P is of the form
P = (t2 , 2t)

(1)

for some t IR. The slope of the normal to the parabola at P is


dx
t. (This can be seen by writing the slope of the normal as
=
dy
2t
dx/dt
= .) The normal to the circle C2 at P is along MP and

dy/dt
2
8 2t
. Equating these two slopes we get
so has slope
2 t2
8 2t
= t
2 t2

(2)

which gives t3 = 2 or t = 2. Hence P = (4, 4).


In this reasoning, we appealed to intuition to determine P . Consider
a variable point X on the parabola and draw a circle with centre S and
radius SX. In general, this circle will meet the parabola at some other
point, say Y too. But then the points on the parabola lying in between
X and Y will be closer to S than X (and Y too). So, for X to be the
closest point to S, the circle centred at S and passing through X must
meet it only at X and hence muct touch the parabola at X.
A more rigorous method to locate P is by actually calculating
the distance SP as a function of the parameter t and minimising it.
Clearly, it suffices to minimise (SP )2 which is a little easier. With P
as in (1), we have
(SP )2 = (t2 2)2 + (2t 8)2 = t4 32t + 68

(3)

Let us call this as f (t). Then f (t) = 4t3 32 which vanishes only at
t = 2. Also, it changes sign from + to as t passes through 2. Hence
the distance of P from S is minimum when t = 2, i.e. when P = (4, 4),
the same answer as before.
The use of the parametric coordinates considerably simplified the
determination of P . To stress this, let us also try to find P without
65

them, i.e. directly from the equation of the parabola y 2 = 4x. Let the
point P be (x0 , y0 ). Then the equation of the tangent to it at P is
yy0 = 2(x + x0 )

(4)

So the slope of the tangent to the parabola at P0 is

2
and hence that
y0

y0
of the normal at P is
. Hence this is also the slope of the line SP
2
because the radius through P is normal to the tangent to the circle C2
at P . Therefore we get
y0
y0 8
=
2
x0 2

(5)

This gives one equation in the unknowns x0 and y0 . The other equation
is y02 = 4x0 . Solving these two equations simultaneously we can get the
values of x0 and y0 . We do not do this. We mention it just to illustrate
that when a particular approach which is quite direct to start with,
later leads to complicated equations, it is time to pause and look for
an alternative. With parametric equations, we had to solve only one
equation in one unknown (viz. t) and that is surely easier than solving
a system of two equations in two unknowns.
A direct computation now gives
SP =

(4 2)2 + (4 8)2 =

4 + 16 = 2 5

(6)

So (A) is true. For (B) it is not necessary determine Q as we already


know SQ = 2, the radius of C1 . As S, Q, P are collinear with Q lying
between S and P , we have

(7)
QP = SP SQ = 2 5 2 = 2( 5 1)
Hence
SQ : QP =
After rationalisation, this becomes

66

1
51

5+1
. So (B) is false.
4

(8)

For (C), note that the normal to the parabola at P is also the
normal to the circle C2 at P as C2 and the parabola touch each other
at P . Indeed this is how we found P to begin with. Moreover, the
normal to C2 at P is the line SP whose equation is
y4=

4
(x 4)
2

(9)

i.e. y + 2x = 12. Hence its intercept with the x-axis is 6. So (C) is


true.
Finally, for (D), note that the tangent to the circle C1 at Q is
parallel to the tangent to C2 at P . Hence they have the same slopes.
The slope of the latter tangent is obtained from that of the normal
which we already know from (7) as 2. So the tangent to C1 at Q has
1
slope . Thus (D) is true too.
2
All the parts of the problem are easy if done in the right way.
Apparently, the problem is designed to test the ability of a candidate
to take short cuts. Thus, in (A) deciding that the circle C2 must touch
the parabola is done quicky by intuition. Then the use of parametric
equations considerably simplified the determination of P . Finally, although (B) and (D) involve the point Q, nowhere we have to determine
Q. This is, therefore, a good problem which rewards candidates who
are good at noticing such short cuts.
2
2
Q.48 Let a,
 b IR and a + b 6= 0. Suppose 

1
, t IR, t 6= 0 , where i = 1. If z = x+iy
S = z C| : z =
a + ibt
and z S, then (x, y) lies on

1
1
and centre
, 0 for a > 0, b 6= 0
2a
2a


1
1
and centre , 0 for a < 0, b 6= 0
(B) the circle with radius
2a
2a
(C) the x-axis for a 6= 0, b = 0


(A) the circle with radius

(D) the y-axis for a = 0, b 6= 0

Answer and Comments: (A, C, D). As a, b, t are real numbers,


a + ibt is a complex number with real part a and imaginary part bt.
67

The complex number z is given as the reciprocal of this complex number


a + ibt. But it does not ask us to find the locus of z. It only asks us to
check whether z lies on some given curves under the given conditions.
As a result, (C) and (D) are too easily seen to be true because in (C) the
number a + ibt is a real number and the reciprocal of a real number is a
real number. Similarly, in (D), the number a + ibt is purely imaginary
and hence its reciprocal is also purely imaginary.
It is only the first two options that need a little work. For that, we
express x and y in terms of t as follows.
x + iy =

1
a ibt
a ibt
=
= 2
a + ibt
(a + ibt)(a ibt)
a + b2 t2

(1)

equating the real and imaginary parts of the two sides we get
a
a2 + b2 t2
bt
and y = 2
a + b2 t2
x =

(2)
(3)

We eliminate t in these two equations. Dividing ((3) by (2) we have


y
bt
=
x
a

(4)

ay
. (Note that we are assuming that b 6= 0.) We can
bx
put this either in (2) or in (3). Obviously, the first oprion is better and
gives

which gives t =

x=

a
ax2
=
a2 + (a2 y 2 /x2 )
a2 (x2 + y 2)

(5)

Since a 6= 0 in both (A) and (B), x is also non-zero by (2). So (5)


simplifies to
1
x2 + y 2 x = 0
a

(6)

Completing the squares, this becomes


(x

1 2
1
) + y2 = 2
2a
4a
68

(7)

1
regardless of whether a > 0 or a < 0. So
2a
1
(B) is wrong. As for (A) the radius is
as a > 0. So (A) is correct.
2a
There is not much about complex numbers in this problem except fining the real and imaginary parts of the reciprocal of a complex
number. As mentiomed earlier, (C) and (D) can be answered without
doing any computation. And this may prove beneficial in view of the
new policy of giving credit for each correct answer if the other parts
are not answered. But there is a catch in (B). A candidate might think
that since (B) is obtained from (A) by replacing a by a, whenever
(A) is true, so is (B). But while this is indeed so for the radius of the
circle, it is not so for the centre. And this mistake will prove costly,
because then the credit earned by answering (C) and (D) correctly will
be washed out.
This is a circle centred at

Q.49 Let a, , IR. Consider the system of linear equations


ax + 2y =
3y 2x =
Which of the following statement(s) is(are) correct?
(A) If a = 3 then the system has infinitely many solutions for all
values of and
(B) If a 6= 3, then the system has a unique solution for all values of
and
(C) If + = 0, then the system has infinitely many solutions for
a = 3

(D) If + 6= 0, then the system has no solution for a = 3.

Answer and Comments: (B, C, D). The problem comes under the
theory of solutions of systems of linear equations discussed in Comments 15 to 17 of Chapter 2. But in the present problem, as the
number of unknowns as well as the number of equations is only 2, a
direct approach is possible and indeed, preferable, except possibly for
(B). Note that we are not actually asked to find the solutions, but only
whether they exist and how many of them exist. The determinant, say
69

a 2
of the coefficient matrix, viz.
is 2(a + 3). In (B), this
3 2
comes out to be non-zero and hence the system has a unique solution
regardless of what and are. (It is not hard to identify the solution.
But that is not asked.)
In all other options, vanishes. Evidently, truth of (A) will imply
that of (C). A clever student knows that generally when this happens,
the weaker statement is true and the stronger one false. And, in the
present case the gamble will pay off. Still, for an honest answer, when
a = 3, adding the two equations gives + = 0. So, if at all there
exists a solution, then + must vanish. This rules out (A) and also
establishes (D) (by taking contrapositive). As for (C), when = ,
the two equations are multiples of each other and have the same set of
solutions. But each equation has infinitely many solutions, because x
can be given any real value and then y can be determined from it. So,
in (C), the system has infinitely many solutions.
The problem is too trivial to comment upon. Still, it is worthwhile
to look at it geometrically. If x and y denote the cartesian coordinates
of a point in the plane, then each equation represents a straight line.
And a solution of the system is a point common to both the lines. In
(B), the lines are neither identical, nor parallel and hence they intersect
at a unique point. In (C), the lines are identical and so all points on
them are common to both. In (D), on the other hand, they are parallel
and hence have no point in common.
The problem could have been made less trivial by giving a system of
three equations in three unknowns, say x, y, z. In that case the solution
sets would be subsets of IR3 . Q.22 of Paper 1 of JEE 2007 is an excellent
example of an imaginative problem of this type.
Q.50 Let f : [1/2, 2] IR and g : [1/2, 2] IR be the functions
defined by f (x) = [x2 3] and g(x) = |x|f (x) + |4x 7|f (x), where [y]
denotes the greatest integer less than or equal to y for y IR. Then
(A) f is discontinuous exactly at three points in [/2, 2]
(B) f is discontinuous exactly at four points in [1/2, 2]
(C) g is NOT differentiable at exactly four points in (1/2, 2)
70

(D) g is NOT differentiable exactly at five points in (1/2, 2).


Answer and Comments: (B,C). The greatest integer function [x]
and the absolute value functions |x| are important in real life for they
give, respectively, an idea of approximately how big x is when rounded
(from below) to an integer and how big its true size (stripped of its
sign) is. (The function [x] is nowadays also often called the floor function and denoted by x. The related function defined as the smallest
integer greater than or equal to x is called the ceiling function and
denoted by x. Clearly, x x x for all real x. For interesting
applications of these apparently trivial functions, see Chapter 3 of the
book Concrete Mathematics - A Foundation for Computer Science by
Graham, Knuth and Patashnik. The book is full of interesting problems and humorous comments.)
In elementary courses in calculus, [x] and |x| figure as are standard
examples, respectively, of step functions and functions that are continuous but not differentiable. There was an unwritten rule that every year
the JEE paper in mathematics ought to include at least one question
based on the continuity and differentiability of these functions. But
recently there have been years without such questions. This year the
paper-setters have revived them.
Let us first tackle f (x) = [x2 3]. Since 3 is an integer, [y 3] is the
same as [y] 3 for any y. So, f (x) = [x2 ] 3. As constant functions
are continuous everywhere, we need only test the continuity of [x2 ].
It is discontinuous precisely when x2 is an integer. In the interval

[1/2, 2] the values of x for which this happens are x = 0, 1, 2, 3
and 2. So, these are the possible points of discontinuity. However,
at 0, x2 is positive on both the sides of 0 and so [x2 ] = 0 for x on
both the sides of 0 in a small neighbourhood of 0. So 0 is not a point
of discontinuity of [x2 ] even though it is a point of discontinuity of
[x]. Another point to note is that since 2 is the right end-point of the
interval [1/2,
2], continuity at it means only continuity from the left.
For x ( 3, 2), f (x) = [x2 ] 3 = 0 and hence lim f (x) = 0. But
x2

f (2) = [4] 3 = 1 6= 0. So, f is indeed discontinuous at 2 too. So (B)


holds and (A) is false. It is a little unfortunate that those who miss
the point about continuity at 2 will not get penalised. Perhaps, the left
71

end point of the domain interval should have been given as an integer
to test this point.
For (C) and (D), the end-points of the interval are carefully excluded.
The function g(x) factors as
g(x) = (|x| + |4x 7|)f (x)

(1)

The first factor is continuous everywhere and differentiable everywhere


7
except at 0 and . As for the second factor, f (x) = [x2 ] 3, we already
4
identified its points of dicontinuity. So it cannot be differentiable at
these points. However, in between any two consecutive points of discontinuity the function is constant and hence differentiable. Taking the
union of the points of non-differentiability, it is tempting to say that
the product function viz., g(x), fails to be differentiable at the points

7
0, 1, 2, 3 and in (1/2, 2). But there is a catch here. Sometimes
4
the irregularity of one factor at a point may get masked or cured by
the behaviour of the other function. For example, even though |x| is
not differentiable at 0, its product with x2 , i.e. the function |x|3 is
differentiable at 0 and has derivative 0.
So we have to be wary how the function behaves in a small neigh7
bourhood of the points 0 and . For x (, ), f (x) is a constant
4
function 3 and so (|x| + |4x 7|)f (x) will be 3|x| + 3|4x 7| in
this neighbourhood which is not differentiable at 0. However, in a suf7
7
7
ficiently small neighbourhood ( , + ) of , x2 is slightly greater
4
4
4
7
2
than 3 (since > 3) and so f (x) = [x 3] = 0. Consequently, g(x)
4
vanishes identically in this neighbourhood and hence is differentiable
7
everywhere in it including at . Thus, the only points where g(x) is
4
not differentiable are 0, 1, 2 and 3. Thus (C) is true and (D) false.
This is a well designed problem which penalises those who blindly
apply theorems like the product of two differentiable functions is differentiable. This theorem is true. But it is silent as to what happens
at points where either of the two factors fails to be differentiable.
With sums, this type of curing cannot occur. That is, suppose
w(x) = u(x)+v(x) where u(x) and v(x) are defined in a neighbourhood
72

of a point c and u is differentiable at c but v is not. Then we can


definitely say that w is not differentiable at c because if it were then so
would be the difference w u which equals v. An analogous reasoning
for a product w = uv would be valid only if the factor u were non-zero
w(x)
in a neighbourhood of c, for then we could write v(x) as
and
u(x)
claim that it is differentiable at c, being the ratio of two differentiable
functions with the denominator non-vanishing.

SECTION - 3: (Maximum Marks : 12)


This section contains TWO paragraph. Based on each paragraph, there
will be TWO questions Each question has FOUR options (A), (B), (C) and
(D). ONE OR MORE THAN ONE of these four option(s) is(are) correct For
each question, darken the bubble(s) corresponding to all the correct option(s)
in the ORS.
Marking scheme :
+3 If only the bubble(s) corresponding to all the correct option(s) is(are)
darkened
0 In all other cases.
PARAGRAPH 1
Football teams T1 and T2 have to play two games against each other.
It is assumed that the outcomes of the two games are independent. The
1 1
probabilities of T1 winning, drawing and losing a game against T2 are ,
2 6
1
and , respectively. Each team gets 3 points for a win, 1 point for a draw
3
and 0 point for a loss in a game. Let X and Y denote the total points scored
by teams T1 and T2 respectively after two games.
Q.51 P (X > Y ) is
(A)

1
4

(B) (A)

5
12

(C) (A)

1
2

(D) (A)

7
12

Answer and Comments: (B). There are three mutually exclusive


possibilities for each game. Their probabilities add up to 1 as they
73

ought to. The paper-setters could have withheld one of these probabilities and forced the candidate to find it out. In fact, that would be
more in order in a paragraph type question.
Let W1 , D1 and L1 be the events of the first team winning, drawing
and losing the first game. Let W2 , D2 and L2 denote the corresponding
events for the second game. We are given that each of the first three
events is independent of each of the last three. Further it is given that
1
2
1
P (D1 ) = P (D2 ) =
6
1
P (L1 ) = P (L2 ) =
3

P (W1 ) = P (W2 ) =

(1)
(2)
(3)

The outcome of the match is one of the nine events of the form A B
where A {W1 , D1 , L1 } and B {W2 , D2 , L2 }. These nine events are
mutually exclusive. The total score X of T1 will exceed Y only in the
cases W1 W2 , W1 D2 and D1 W2 , because in order that T1 beats
T2 , it has to win at least one game.
Because of independence of the ecvents, P (A B) = P (A)P (B) for all
A, B. Therefore, adding the three probabilities,
P (X > Y ) = P (W1 )P (W2 ) + +P (W1 )P (D2 ) + P (D1 )P (W2)
1
1
1
+
+
=
4 12 12
5
=
(4)
12
Q.52 P (X = Y ) is
(A)

11
36

(B)

1
3

(C)

13
36

(A)

1
2

Answer and Comments: (C). The logic is very similar to that in


the last question. The two scores will be equal when each team wins
one game or when both the games are draws. Therefore
P (X = Y ) = P (W1 L2 ) + P (L1 W2 ) + P (D1 D2 )
74

2
1
=
+
6
13
=
36

1 1 1 1 1
+ +
3 3 2 6 6
1
1
+
6 36
(5)

Both the parts are extremely simple, both in terms of reasoning


and calculations. The problem is ideally suited for a multiple choice
format. The reasoning is simple to conceive but time consuming to
express in words. And there is no sneaky path as there was in Q.40 of
Paper 1.
A more challenging question would have been to give a match with
n games (the other data remaining the same) and study P (X > Y ) as
1
5
a function of n. (Its values are and
for n = 1 and 2 respectively.)
2
12
PARAGRAPH 2
Let F1 = (x1 , 0) and F2 = (x2 , 0) for x1 < 0 and x2 > 0 be the foci of the
x2 y 2
+
= 1. Suppose a parabola having its vertex at the origin and
ellipse
9
8
its focus at F2 intersects the ellipse at a point M in the first quadrant and
at a point N in the fourth quadrant.
Q.53 The orthocentre of the triangle F1 MN is
(A)

9
,0
10

(B)

2
,0
3

(C)

9
,0
10

(D)

2
, 6
3

Answer and Comments: (A). Another geometry problem where we


have to determine one thing after another. The foci of the ellipse are
at the points (3e, 0) where e is the eccentricity of the ellipse. It is
given by
8 = 9(1 e2 )

(1)

y 2 = 4x

(2)

1
and hence F1 = (1, 0) and F2 = (1, 0). As the
3
focus of the parabola is at (1, 0) and its vertex at (0, 0), its equation is

which gives e as

75

Therefore the x-coordinates of its points of intersection with the given


ellipse satisfy the equation
x2 x
+ =1
9
2

(3)

q
9
+ 9. We
i.e. x2 + x 9 = 0. This is a quadratic with roots 94 81
16
2
discard the negative root as the parabola lies entirely on the right of
15 9
3
the y-axis. The positive root is
= . Hence the points M and
4
4
2
N are

3
, 6
M =
2


3
, 6
and N =
2


(4)
(5)

F2

As the triangle F1 MN is symmetric about the x-axis, its orthocentre


lies on the x-axis. To locate
need the altitude through M (or N).
it we
2 6
6
=
. Hence the slope of the altitude
The slope of F1 N is
5/2
5
5
through M is . Hence its equation is
2 6
y

3
5
6 = (x )
2
2 6
76

(6)

3
12
5
3
It intersects the x-axis when 6 = (x which gives x =
2
2  5
2 6

3 12
9
9
and hence x =
= . So the orthocentre is at , 0 .
2
5
10
10
A long, and essentially arbitrary chain of mediocre problems.
Q.54 If the tangents to the ellipse at M and N meet at R and the normal
to the parabola at M meets the x-axis at Q, then the ratio of the area
of the triangle MQR to the area of the quadrilateral MF1 NF2 is
(A) 3 : 4 (B) 4 : 5 (C) 5 : 8 (D) 2 : 3
Answer and Comments: (C). A continuation of the journey in the
last problem. The only intelligent part is that since both the parabola
and the ellipse are symmetric about the x-axis, so are the tangents to
the ellipse at M and N and therefore, to find R it suffices to find either
one of them and see where it cuts the x-axis.
y

F2

3
The equation of the tangent to the ellipse at the point M = ( , 6) is
2

x
6
+
=1
(7)
6
8
When y = 0, x = 6. So we get
R = (6, 0)
77

(8)

We now determine Q. The slope of the normal to the parabola y 2 = 4x


3
at the point M = ( , 6) can be obtained by writing the equations
2

6
. Then the slope of
parametrically as x = t2 , y = 2t and taking t =
2

6
. Hence its equation is
the normal is t =
2

6
3
y 6=
(x )
(9)
2
2
When y = 0, x = 2 +

7
3
= . Therefore
2
2
7
Q = ( , 0)
2

(10)

We are asked the ratio of the area of the triangle MQR to that of the
quadrilateral MF1 NF2 . We can do this without actually calculating
these areas. By symmetry about the x-axis, the area of the quadrilateral MF1 NF2 is twice that of the triangle MF1 F2 . Now the triangles
MQR and MF1 F2 have the same altitude. So the ratio of their areas is
the same as the ratio of their bases QR and F1 F2 . From (8) and (10),
5
7
QR = 6 = . Also F1 F2 = 2. So
2
2
MQR : MF1 F2 =

5
:2=5:4
2

(11)

Since the area of the quadrilateral MF1 NF2 is twice the area of the
triangle MF1 MF2 , the desired ratio is 5 : 8.
Not a very exciting problem, except for the short cut at the end.
The paper-setters have carefully arranged the numerical data in such
a way that even though square roots appear in the slopes sometimes,
the answers to both the problems in this paragraph come in terms of
whole numbers or simple fractions. This really helps the candidate in
a severely time limited test.
But the real objection to paragraph type questions voiced in the
past holds this year too. Such questions are meant to test if a candidate
can understand some unseen material. Instead what we see is a bunch
78

of ordinary questions sharing a common data. With this definition,


many other questions could have been posed as paragraph type, e.g.
Q.47 above or Q.42 in Paper 1.

79

CONCLUDING REMARKS
As mentioned at the beginning, there are some welcome changes this year.
The number of questions has been reduced to a reasonable level. Matching
the pairs which often resulted in clubbing together a large number of unrelated questions has been dropped. Further, in questions where more than
one options may be correct, candidates can get some partial credit unlike in
the past.
The paper-setters have also generally been careful in designing the numerical data to make the calculations simple. Some questions seem to be
designed to reward candidates who can take intelligent short cuts.
By far, the question that stands out as the best question is Q. 40 of Paper
2 about the relative growth of geometric and arithmetic progressions. Also
noteworthy are a few problems which seem to be designed with a specific
purpose. These include Q.42 of Paper 1 (about a pyramid) which tests the
ability to visualise solid geometric data and also the ability to quickly get an
answer using pure geometry rather than the drudgery of coordinates, Q.47
of Paper 1 (about composite functions and inverse functions) which tests the
ability to handle functions as entities in their own rights instead of being
0
tied down to formulas, Q.53 of Paper 1 (about a limit of the form) which
0
tests the ability to compare orders of magnitude, Q.37 of Paper 2 (about the
fiftieth power of a matrix) which tests the ability to experiment and recognise
patterns), Q.43 of Paper 2 (about the continuity and differentiability of a
linear combination of two functions) which tests the ability to focus on the
relevant part of the data, Q.47 of Paper 2 (which tests the ability to take
short cuts) and Q.50 of Paper 2 (about the differentiability of the product
of two functions) which tests the ability to keenly study what happens at an
irregular point).
As regards the distribution of topics, the paper-setters have revived combinatorial identities within the limits of a multiple choice question and also
paid a lip service to number theory in Q.51 of Paper 1. The problems about
combinatorics (Q.38 of Paper 1) and probability (Q. 40 of Paper 1 and Paragraph 1 of Paper 2) are too simple as compared with many problems in the
past few years (even after the JEE became totally MCQ). So are the problems
on complex numbers (Q. 54 of Paper 1 and Q.48 of Paper 2). Inequalities
figure in only indirectly. The trick of recasting a trigonometric series as a
telescopic series has been pulled in Q.39 of Paper 2.
80

Coordinate geometry problems usually have long winded statements and


seem to lack any purpose. Parabolas seem to be favourite of the paper-setters
as they have appeared in no less than five questions (Q. 48 of Paper 1 and
Questions 38, 47, 53 and 54 of Paper 2). As if this was not enough, the locus
in Q.49 of Paper 1 also turns out to be a parabola (although that is not
needed in the solution). One wishes that some of these problems could have
been replaced to make room for, say, hyperbolas and pairs of straight lines.
The two questions on differential equations (Q. 43 and Q.46 of Paper 1)
are of the same spirit and one wonders what is gained by this duplication,
especially within the same paper. At least one of these could have been
replaced by a question on applications of differential equations.
Q.44 of Paper 2 (about a function defined as the limit of a sequence of
functions) is clearly the toughest question in both the papers together. It
could have been diluted as suggested in the comments about it.
Many of the questions were suitable for JEE Main rather than JEE advanced. Their preponderence blurs the distinction between the two rounds.
Also a candidate who picks them first has a better chance than someone who
struggles with a difficult question.
But there is really no way to change this picture in the present set-up.
That would require drastic changes in the JEE structure as were pointed out
in detail in last years commentary at the end. So, as in the last year, this
commentary concludes with a prayer that sometime in the future JEE will
regain its status.

81

Você também pode gostar