MATHEMATICS II FOR DATA SCIENCE

GA-11 SOLUTIONS MAY 2024

1. Question
Consider the following function
f(x, y, z) = x 2 z + y 3 e xy + sin(xyz)
Which of the following option(s) is (are) true?

1.1. Answer:
• f xx = 2z + y 5 e xy + y 2 z 2 - y 2 z 2 sin(xyz)

1.2. Solution
Few useful formulae :
d
• ex = ex
dx
• Let h(x) = f(g(x)), then h'(x) = f'(g(x) × g'(x)
d
• (f(x) × g(x)) = f'(x) × g(x) + f(x) × g'(x)
dx

Let us find f xx

fxx 𝛿 𝛿
= x 2 z + y 3 e xy + sin(xyz)
𝛿x 𝛿x
𝛿
= 2xz + y 4 e xy + yz cos(xyz)
𝛿x
= 2z + y 5 e xy + y 2 z 2 - (yz) 2 sin(xyz)
Hence, option 1 is incorrect and option 2 is correct.

Let us find f zy

fzy 𝛿 𝛿
= x 2 z + y 3 e xy + sin(xyz)
𝛿y 𝛿z
 𝛿
= x 2 + xy cos(xyz)
𝛿y
= x cos(xyz) - x 2 yz sin(xyz)

Hence, option 3 is incorrect.

Let us find f xz

fxz 𝛿 𝛿
= x 2 z + y 3 e xy + sin(xyz)
𝛿z 𝛿x
𝛿
= 2xz + y 4 e xy + yz cos(xyz)
𝛿z
= 2x + y cos(xyz) - xy 2 z sin(xyz)

Hence, option 4 is incorrect.

2. Question
Number of critical points of the function f(x, y, z) = e x x 2 - y 2 - 2z 2 is

2.1. Answer : 2

2.2. Solution
Few useful formulae :
d x
• e = ex
dx
• Let h(x) = f(g(x)), then h'(x) = f'(g(x) × g'(x)
d
• (f(x) × g(x)) = f'(x) × g(x) + f(x) × g'(x)
dx

Let us find ∇f(x, y, z)

∇f(x, y, z) = e x x 2 - y 2 - 2z 2 + 2xe x , - 2ye x , - 4ze x

= e x x 2 - y 2 - 2z 2 + 2x, - 2y, - 4z

We can see that the gradient of f is defined for all the points in R 3 . Therefore, all the critical points will have
its gradient equal to (0, 0, 0)

Let us equate ∇f(x, y, z) with (0, 0, 0). Note, e x ≠ 0 for all x ∈ R.

∇f(x, y, z) = (0, 0, 0)
e x x 2 - y 2 - 2z 2 + 2x, - 2y, - 4z = (0, 0, 0)
x 2 - y 2 - 2z 2 + 2x, - 2y, - 4z = (0, 0, 0)
We get the following equations :

x 2 - y 2 - 2z 2 + 2x = 0
x 2 + 2x = 2z 2 + y 2 (1)

-2y = 0
y=0 (2)

-4z = 0
z=0 (3)

Substitute (2) and (3) in (1)

x 2 + 2x = 2(0) 2 + (0) 2
x 2 + 2x = 0
x( x + 2) = 0

We can see that x = 0 or x = 2.

The critical points of f are (0, 0, 0) and (2, 0, 0).

3. Question

Choose the correct statement(s) about f(x, y) = x2 + y2.

3.1. Answers :
• (0, 0) is the only critical point of f.
• (0, 0) is a local minimum of f.
• (0, 0) is a global minimum of f.
3.2. Solution
Few useful formulae :
 d 1
• x =
dx 2 x
• Let h(x) = f(g(x)), then h'(x) = f'(g(x) × g'(x)
d
• (f(x) × g(x)) = f'(x) × g(x) + f(x) × g'(x)
dx

Let us find the gradient of f

∇f(x, y) 2x 2y
= ,
2 x2 + y2 2 x2 + y2
x y
= ,
x2 + y2 x2 + y2

We can see that the gradient is not defined at (0, 0).

Let us find f x (0, 0) using the definition of partial derivative.

f(0 + h, 0) - f(0, 0) h2 - 0
lim = lim
h→0 h h→0 h
|h|
= lim
h→0 h

|h| = lim 1 |h| = lim - 1

lim h+ → 0 lim h- → 0
h →0+
h -
h →0 h
=1 = -1

|h| |h| |h|

Since, lim ≠ lim , we can say that lim does not exist.
h+ → 0 h -
h →0 h h→0 h

Therefore, f x (0, 0) does not exist. Hence, (0, 0) is a critical point.

x y
We can see that ∇f(x, y) = , will not be equal to (0, 0) for all
2 2 2 2
x +y x +y
(x, y) ∈ R 2 - {(0, 0)}.Therefore, the only critical point of f is (0, 0).

f(0, 0) = 0 2 + 0 2
=0

The range of f(x, y) is [0, ∞). Therefore, (0, 0) is the global minimum. Since it is a global minimum, it is
also a local minimum.
Hence, option 2, option 4 and option 5 are correct. Option 1 and option 3 are incorrect.

4. Question
Find the minimum value of the surface area of a rectangular box that has a volume of 1000 uint 3 .

Note: If x, y and z represent the lengths of the three sides of a rectangular box, then the surface area and
volume are given by 2(xy + yz + zx) and xyz respectively.

4.1. Answer : 600

4.2. Solution
Useful formula :
d 1 -n
• =
dx x n
x n+1

Let x, y and z represent the lengths of the three sides of the given rectangular box.
Given ⟹ the volume of the rectangular box is 1000 unit 3 .

xyz = 1000
z = 1000 / xy

We have to find the minimum value of the surface area of the rectangular box, provided the volume is
1000 unit 3 .
Surface Area = 2(xy + yz + zx)
1000 1000
= 2 xy + y +x
xy xy
2000 2000
= 2xy + +
x y

We can see that the surface area is a function of x and y. Therefore, Surface Area = f(x, y) where
2000 2000
f(x, y) = 2xy + + . Clearly, f is not defined at (0, 0).
x y

We have to find the minimum value of f. Let us find ∇f(x, y)

∇f(x, y) 2000 2000

= 2y - , 2x -
x2 y2

Let us find the critical points of f by equating ∇f(x, y) with (0, 0)

∇f(x, y) = (0, 0)
 2000 2000 = (0, 0)
2y - , 2x -
x2 y2

We get the following equations

2000 = 0 2000 = 0
2y - 2x -
x2 y2
2y 2000 2x 2000
= =
x2 y2
yx 2 = 1000 (1) x 1000
= (2)
y2

Substitute (2) in (1)

1000
2 = 1000
y
y2
(1000) 2 = 1000
y3
1000 = y 3
y = 10

Substitute y = 10 in (2)

x 1000
=
(10) 2
x = 10

Therefore, (10, 10) is a critical point of f.

Let us apply Hessian test on this point.

2000 2000
fx = 2y - and f y = 2x - . Let us find f xx , f xy , f yx and f yy .
x2 y2
fxx 𝛿 2000 fxy 𝛿 2000
= 2y - = 2y -
𝛿x x2 𝛿y x2
4000 =2
=
x3

fyx 𝛿 2000 fyy 𝛿 2000

= 2x - = 2x -
𝛿x y2 𝛿y y2
 =2 4000
=
y3
Hessian matrix of f is given by :

Hf(x, y) 4000
2
= x3
4000
2
y3
Let us find Hf(10, 10)

Hf(10, 10) 4000

2
(10) 3
=
4000
2
(10) 3
4 2
=
2 4
det(Hf(10, 10)) = 12. We can also see that fxx (10, 10) = 4. Since det(Hf(10, 10)) > 0 and
fxx (10, 10) > 0, we can say that (10, 10) is a local minimum.
f(10, 10) 2000 2000
= 2(10)(10) + +
10 10
= 200 + 200 + 200
= 600

When x = 10 and y = 10, the surface area becomes 600 unit 2 .

For sufficiently large n, consider the rectangles with boundaries at

• x = 10 -n and x = 10 n
• y = 10 -n and y = 10 n

2000 2000
f(x, y) = 2xy + + is a continuous function in this region. (10, 10) is the local minimum in
x y
this rectangle.
 10 10

(10, 10)

10 -10
10 -10 10 10

By choosing n appropriately, we can cover any arbitrary point in the domain of the function and construct a
bounded rectangle that contains that point. In that bounded region, f attains a local minimum at (10, 10).
Hence, option 4 is correct.

5. Question
Consider a function f(x, y, z) = x 3 + 3y 3 + z 3 - 4x - 6y - 4z. If A is the Hessian matrix of f(x, y, z) at
(0, 1, 1), then which of the following option(s) is(are) true?

5.1. Answers :
• det(A) = 0
• Rank(A) = 2
• Nullity(A) = 1

5.2. Solution
𝛿f 𝛿f 𝛿f
Let us find , and .
𝛿x 𝛿y 𝛿z
 𝛿f = 3x 2 - 4 𝛿f = 9y 2 - 6 𝛿f = 3z 2 - 4
𝛿x 𝛿y 𝛿z
𝛿2f 𝛿2f 𝛿2f
Let us find , and
𝛿x 2 𝛿y 2 𝛿z 2
𝛿2f 𝛿 𝛿 𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= = =
𝛿x 2 𝛿x 𝛿x 𝛿y 2 𝛿y 𝛿y 𝛿z 2 𝛿z 𝛿z
𝛿 𝛿 𝛿
= 3x 2 - 4 = 9y 2 - 6 = 3z 2 - 4
𝛿x 𝛿y 𝛿z
= 6x = 18y = 6z

𝛿2f 𝛿2f
Let us find and
𝛿x𝛿y 𝛿x𝛿z
𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿x𝛿y 𝛿x 𝛿y 𝛿x𝛿z 𝛿x 𝛿z
𝛿 𝛿
= 9y 2 - 6 = 3z 2 - 4
𝛿x 𝛿x
=0 =0

𝛿2f 𝛿2f
Let us find and
𝛿y𝛿x 𝛿y𝛿z
𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿y𝛿x 𝛿y 𝛿x 𝛿y𝛿z 𝛿y 𝛿z
𝛿 𝛿
= 3x 2 - 4 = 3z 2 - 4
𝛿y 𝛿y
=0 =0

𝛿2f 𝛿2f
Let us find and
𝛿z𝛿x 𝛿z𝛿y
𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿z𝛿x 𝛿z 𝛿x 𝛿z𝛿y 𝛿z 𝛿y
𝛿 𝛿
= 3x 2 - 4 = 9y 2 - 6
𝛿z 𝛿z
=0 =0

Hessian matrix Hf(x, y, z) is given by

 𝛿2f 𝛿2f 𝛿2f
𝛿x 2 𝛿x𝛿y 𝛿x𝛿z
𝛿2f 𝛿2f 𝛿2f 6x 0 0
Hf(x, y, z) = = 0 18y 0
𝛿y𝛿x 𝛿y 2 𝛿y𝛿z 0 0 6z
𝛿2f 𝛿2f 𝛿2f
𝛿z𝛿x 𝛿z𝛿y 𝛿z 2
A at (0, 1, 1) is given by :
0 0 0
A = Hf(0, 1, 1) = 0 18 0
0 0 6

Clearly, det(A) = 0

The reduced row echelon form of this matrix is

0 1 0
0 0 1
0 0 0
There are only two columns with leading entries, therefore, Rank(A) = 2.
By Rank-Nullity theorem of matrices, we can say that Nullity(A) = 1.
Hence, option 1, option 2 and option 3 are correct. Option 4 and option 5 are incorrect.

6. Question

6.1. Answers :
 e - e 2 -e 2
• Hessian matrix of f at point (1,1) is
-e 2 e - e 2
• f has a saddle point

6.2. Solution
Few useful formulae :
d 1
• x =
dx 2 x
• Let h(x) = f(g(x)), then h'(x) = f'(g(x) × g'(x)
d
• (f(x) × g(x)) = f'(x) × g(x) + f(x) × g'(x)
dx

𝛿f 𝛿f
Let us find and
𝛿x 𝛿y
𝛿f = e x - e x+y 𝛿f = e y - e x+y
𝛿x 𝛿y

𝛿2f 𝛿2f
Let us find and
𝛿x 2 𝛿y 2
𝛿2f 𝛿 𝛿 𝛿2f 𝛿 𝛿f
= =
𝛿x 2 𝛿x 𝛿x 𝛿y 2 𝛿y 𝛿y
𝛿 x 𝛿 y
= e - e x+y = e - e x+y
𝛿x 𝛿y
= e x - e x+y = e y - e x+y

𝛿2f 𝛿2f
Let us find and
𝛿x𝛿y 𝛿y𝛿x
𝛿2f 𝛿 𝛿 𝛿2f 𝛿 𝛿
= =
𝛿x𝛿y 𝛿x 𝛿y 𝛿y𝛿x 𝛿y 𝛿x
𝛿 𝛿
= e y - e x+y = e x - e x+y
𝛿x 𝛿y
= -e x+y = -e x+y
Hessian matrix Hf(x, y) is given by
𝛿2f 𝛿2f
𝛿x 2 𝛿x𝛿y e x - e x+y -e x+y
Hf(x, y) = =
𝛿2f 𝛿2f -e x+y e y - e x+y
𝛿y𝛿x 𝛿y 2
 𝛿f 𝛿f
We can see that both and are defined for all (x, y) ∈ R 2 . Let us find a critical points of f by equating
𝛿x 𝛿y
𝛿f 𝛿f
with 0 and with 0. Note, e m > 0 for all m ∈ R
𝛿x 𝛿y
𝛿f = 0 𝛿f = 0
𝛿x 𝛿y
x
e -e x+y =0 y
e -e x+y =0
ex 1 - ey =0 ey 1 - ex =0
1 = ey 1 = ex
ln(1) = ln e y ln(1) = ln e x
0=y 0=x

From this, we can see (0, 0) is the only critical point of f(x, y)

Let us find det(Hf(x, y))

det(Hf(x, y)) = e x - e x+y × e y - e x+y - -e x+y × -e x+y

= e xy - e 2x+y - e x+2y + e 2x+2y - e 2x+2y
= e xy - e 2x+y - e x+2y

Let us find det(Hf(0, 0))

det(Hf(0, 0)) = e 0×0 - e 2(0)+(0) - e 0+2(0)

= 1-1-1
= -1

Since det(Hf(0, 0)) < 0 we can say that (0, 0) is a saddle point. There are no more critical points.
Therefore, f does not have a maximum.

Hence, option 2 and option 4 are correct. Option 1 and option 3 are incorrect.

7. Question
Choose the correct statements.
7.1. Answers :

• If f : R 3 → R is a linear transformation then the Hessian matrix H f (x, y, z) is same for all
(x, y, z) ∈ R.
• For each n ∈ N there exists a map f : R n → R such that the rank of the Hessian matrix
Hf (x1 , x2 ,..., xn ) is n for some (x1 , x2 ,..., xn ) ∈ R n
• For each n ∈ N there exists a map f : R n → R such that the rank of the Hessian matrix
Hf (x1 , x2 ,..., xn ) is symmetric for all (x1 , x2 ,..., xn ) ∈ R n
7.2. Solution
Option 1 is incorrect. Let f(x, y, z) = x 4 + y 2 + z 2 .
𝛿2f
We can see that
2
= 12x 2 . As we change x, this will vary. In this example, Hessian matrix Hf (x, y, z)
𝛿x
will not be the same for all (x, y, z). Therefore, the first option is not always true.

Option 2 is correct.
Any linear transformation f : R 3 → R will be of the form f(x, y, z) = ax + by + cz, where a, b, c ∈ R.
Differentiating a linear function twice will result in 0. Therefore, all the second order partial derivatives of f
will be 0. In this case, the Hessian matrix of f will be as follows :
0 0 0
0 0 0
0 0 0
Therefore, the hessian matrix of any linear transformation f : R 3 → R will be the matrix above.

Option 3 and option 4 are correct.

Let f : R n → R be defined as follows :
f(x1 , x2 ,..., xn ) = a1 x12 + a2 x22 + ... + an xn2

where a 1 , a 2 ,..., a n ∈ R

We can see the following :

 𝛿2f 𝛿2f 𝛿2f
= a1 , = a2 ,..., = an
𝛿x12 𝛿x22 𝛿xn2
We can also see that all the second order mixed partial derivatives will be 0. Therefore, the Hessian matrix
Hf (x1 , x2 ,..., xn ) will be
a1 0 ⋱ 0 0
0 a2 0 ⋱ 0
Hf (x1 , x2 ,..., xn ) = ⋱ 0 a3 0 ⋱
0 ⋱ 0 ⋱ 0
0 0 ⋱ 0 an
This is an n × n diagonal matrix. If a 1 , a 2 ,..., a n ∈ R - { 0 }, then all the diagonal entries will be non-zero.
The reduced row echelon form of such a matrix is an n × n identity matrix. Therefore, in such a case, the
rank of H f (x 1 , x 2 ,..., x n ) will be n.

Notice, the above matrix is also a symmetric matrix.

8. Question
The maximum value of the function -2x 2 + 2xy - y 2 + 12x - 4y - 7.

8.1. Answer : 13

8.2. Solution

Let f(x, y) = - 2x 2 + 2xy - y 2 + 12x - 4y - 7.

Let us find the gradient of f

∇f(x, y) = (-4x + 2y + 12, 2x - 4y)

We can clearly see that the gradient is defined for all (x, y) ∈ R 2 . Therefore, we will find the critical points
of f by equating it with (0, 0).

∇f(x, y) = (0, 0)
(-4x + 2y + 12, +2x - 4y) = (0, 0)
We get the following equations.

-4x + 2y + 12 = 0 (1)

2x - 4y = 0
x = 2y (2)

Substitute x = 2y in (1)

-4(2y) + 2y + 12 = 0
-8y + 2y + 12 = 0
 -6y + 12 = 0
y=2

From (2) we can see that y = 2 will imply x = 4. Therefore, (4, 2) is the only critical point of f.

Let us find all the second order derivatives of f.

𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿x 2 𝛿x 𝛿x 𝛿x𝛿y 𝛿x 𝛿y
𝛿 𝛿
= (-4x + 2y + 12) = (2x - 4y)
𝛿x 𝛿x
= -4 =2

𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿y𝛿x 𝛿y 𝛿x 𝛿y 2 𝛿y 𝛿y
𝛿 𝛿
= (-4x + 2y + 12) = (2x - 4y)
𝛿y 𝛿y
=2 = -4

The hessian matrix of f will be :

-4 2
Hf(x, y) =
2 -4

Hf(4, 2) will also be the same as the above matrix.

𝛿2f
det(Hf(4, 2)) = 12 and = - 4.
𝛿x 2
(4,2)

𝛿2f
Since det(Hf(4, 2)) > 0 and < 0, we can say that (4, 2) is a local maximum.
𝛿x 2
(4,2)

Let us look at f(x, y)

Note : (x - y - z) 2 = x 2 + y 2 + z 2 - 2xy - 2xz + 2yz

f(x, y) = -2x 2 + 2xy - y 2 + 12x - 4y - 7

= -x 2 - x 2 - y 2 + 2xy + 8x + 4x - 4y - 4 - 3
= -x 2 + 8x - 3 - x 2 - y 2 + 2xy + 4x - 4y - 4
= - x 2 - 8x + 3 - x 2 + y 2 - 2xy - 2(2)x + 2(2)y + 4
 = - x 2 - 8x + 16 - 16 + 3 - x 2 + y 2 - 2xy - 2(2)x + 2(2)y + 2 2
= - x 2 - 8x + 16 - 13 - (x - y - 2) 2
= - x 2 - 8x + 16 + 13 - (x - y - 2) 2
= -(x - 4) 2 - (x - y - 2) 2 + 13

Both -(x - 4) 2 and -(x - y - 2) 2 will always be lesser than or equal to zero. f(x, y) will be maximum when
both -(x - 4) 2 and -(x - y - 2) 2 become zero. This is possible at (4, 2). Therefore, (4, 2) is a global
maximum.

f(4, 2) = -(4 - 4) 2 - (4 - 2 - 2) 2 + 13
= 13

The maximum value of the function -2x 2 + 2xy - y 2 + 12x - 4y - 7 is 13

9. Question

9.1. Answer :
• f is continuous at (0, 0)
• f x (0, 0) = f y (0, 0)
• f is differentiable at (0, 0)
9.2. Solution
Let us find lim f(x, y)
(x,y) → (0,0)

lim f(x, y) 1
(x,y) → (0,0) = lim (x - y) 2 sin
(x,y) → (0,0) x-y

We can see that substituting (0, 0) in f(x, y) is not possible. Let us approach (0, 0) along
Notice, the range of sine function is [-1, 1].

3 1
In our case, for example, if x = and y = , we get
𝜋 𝜋
1 1 𝜋
sin = sin = sin =1
3
-𝜋
1 2 2
𝜋 𝜋

1 3
if x = and y = , we get
𝜋 𝜋
1 1 -𝜋
sin = sin = sin = -1
1
-
3 -2 2
𝜋 𝜋 𝜋

Therefore,

1
-1 ⩽ sin ⩽1
x-y
1
-(x - y) 2 ⩽ (x - y) 2 sin ⩽ ( x - y) 2
x-y
1
lim - ( x - y) 2 ⩽ lim (x - y) 2 sin ⩽ lim (x - y)
(x,y) → (0,0) (x,y) → (0,0) x-y (x,y) → (0,0)

1
0⩽ lim (x - y) 2 sin ⩽0
(x,y) → (0,0) x-y

By applying Sandwich Theorem, we get :

1
lim (x - y) 2 sin =0
(x,y) → (0,0) x-y

Notice, f(0, 0) = 0.

1
Since, lim (x - y) 2 sin exists and is equal to f(0, 0), we can say that f is continuous at
(x,y) → (0,0) x-y
(0, 0). Hence, option 1 is correct.

Partial derivative of f with respect to x at (0, 0)

𝛿f f(h, 0) - f(0, 0)
= lim
𝛿x
h→0 h
(0,0)
 1
(h - 0) 2 sin h-0
-0
= lim
h→0 h
1
h 2 sin h
= lim
h→0 h
1
= lim h sin
h→0 h

Again, the range of sine function is [-1, 1]

2 1 𝜋 2
In our case, for example, when h = , sin = sin = 1 and if h = - then
𝜋 h 2 𝜋
1 -𝜋
sin = sin = - 1. Therefore,
h 2

1
-1 ⩽ sin ⩽1
h
1
-h ⩽ h sin ⩽h
h
1
lim - h ⩽ lim h sin ⩽ lim h
h→0 h→0 h h→0

1
0 ⩽ lim h sin ⩽0
h→0 h

By applying Sandwich Theorem, we get :

𝛿f = 0
𝛿x

Let us find partial derivative of f with respect to y at (0, 0). Note, sin(-𝜃) = - sin(𝜃)

𝛿f f(0, h) - f(0, 0)
= lim
𝛿y
h→0 h
(0,0)
1
(0 - h) 2 sin 0-h
-0
= lim
h→0 h
-1
h 2 sin h
= lim
h→0 h
 1
= lim - h sin
h→0 h
1
= - lim h sin
h→0 h
= - ( 0)
=0

Therefore, f x (0, 0) = f y (0, 0) = 0. Hence, option 2 is incorrect and option 3 is correct.

∇f(0, 0) = (fx (0, 0), fy (0, 0))

= (0, 0)

Differentiability of f
A scalar-valued multivariable function f : R n → R is said to be differentiable at v ∈ R n if ∇f(v) exists and
f(v + h) - f(v) - h · ∇f(v)
lim = 0. Here, n > 1.
h→0 ||h||
We know that the gradient of f at (0, 0) exists.

We have to see if the following limit is equal to 0. Let h = (a, b)

f((0, 0) + h) - f(0, 0) - h · ∇f(0, 0)

lim
h→0 ||(a, b)||

f(a, b)
lim
(a,b) → (0,0) ||(a, b)||

1
(a - b) 2 sin a-b
lim
(a,b) → (0,0)
a2 + b2

( a - b) 2 1
lim sin (1)
(a,b) → (0,0)
a2 + b2 a-b

Before solving this limit, let us solve the following

( a - b) 2
lim
(a,b) → (0,0)
a2 + b2
 a 2 + b 2 - 2ab
lim
(a,b) → (0,0)
a2 + b2

2
a2 + b2 - 2ab
lim
(a,b) → (0,0)
a2 + b2

2
a2 + b2 2ab
lim -
(a,b) → (0,0)
a2 + b2 a2 + b2

2ab
lim a2 + b2 -
(a,b) → (0,0)
a2 + b2

A. M - G. M inequality :

2|ab| ⩽ a 2 + b 2

Proof :

0 ⩽ ( a - b) 2
0 ⩽ a 2 - 2ab + b 2
2ab ⩽ a 2 + b 2
2|ab| ⩽ a 2 + b 2

Therefore,

-ab ⩽ |ab|

-2ab ⩽ 2|ab|

From 2|ab| < a 2 + b 2 ,

-2ab ⩽ 2|ab| ⩽ a 2 + b 2

-2ab ⩽ a 2 + b 2

-2ab a2 + b2
⩽
a2 + b2 a2 + b2
 2
-2ab a2 + b2
⩽
2 2
a +b a2 + b2

-2ab
⩽ a2 + b2
a2 + b2

-2ab
a2 + b2 + ⩽ a2 + b2 + a2 + b2
a2 + b2

-2ab
a2 + b2 + ⩽2 a2 + b2
a2 + b2

From the above we can say that the following is true

-2ab
-2 a2 + b2 ⩽ a2 + b2 +
a2 + b2

Combining both, we get

-2ab
-2 a2 + b2 ⩽ a2 + b2 + ⩽2 a2 + b2
a2 + b2

-2ab
lim - 2 a 2 + b 2 ⩽ lim a2 + b2 + ⩽ lim 2 a2 + b2
(a,b) → 0 (a,b) → 0 (a,b) → 0
a2 + b2

-2ab
0 ⩽ lim a2 + b2 + ⩽0
(a,b) → 0 2 2
a +b
By Sandwich Theorem, we get :

2 2
-2ab ( a - b) 2
lim a +b + = lim =0 (2)
(a,b) → 0 2 2 (a,b) → (0,0) 2 2
a +b a +b

We know that the following is true.

1
-1 ⩽ sin ⩽1
a-b
 ( a - b) 2 ( a - b) 2 1 ( a - b) 2
- ⩽ sin ⩽
2
a +b 2
a +b2 2 a-b a2 + b2

- ( a - b) 2 ( a - b) 2 1 ( a - b) 2
lim ⩽ lim sin ⩽ lim
(a,b) → (0,0) 2
a +b 2 (a,b) → (0,0) 2
a +b 2 a-b (a,b) → (0,0)
a2 + b2
From (2)

( a - b) 2 1
0⩽ lim sin ⩽0
(a,b) → (0,0)
a2 + b2 a-b

From this we can conclude that f is differentiable at (0, 0). Hence, option 4 is correct.

10. Question
A company produces two types of chocolates, one that sells for Rs.3 and another at Rs.2. The total
revenue, in lakhs, from the sale of x lakhs chocolates at Rs.2 each and y lakhs at Rs.3 each is given by
F(x, y) = 2x + 3y
The total cost in lakhs for producing x lakhs of the Rs.2 chocolates and y lakhs of the Rs.3 chocolates is
given by
G(x, y) = 2x 2 - 2xy + y 2 - 9x + 6y + 7
Then the maximum profit in lakhs is

10.1. Answer : 11.25

10.2. Solution
Let profit be P(x, y). We know profit = revenue - cost.

P(x, y) = F(x, y) - G(x, y)

= (2x + 3y) - 2x 2 - 2xy + y 2 - 9x + 6y + 7
= -2x 2 + 2xy - y 2 + 11x - 3y - 7

Let us find ∇f(x, y)

∇f(x, y) = (-4x + 2y + 11, 2x - 2y - 3)

We can see that ∇f(x, y) is defined for all (x, y) ∈ R. Let us find critical points by equating ∇f(x, y) with
(0, 0)
∇f(x, y) = (0, 0)
(-4x + 2y + 11, 2x - 2y - 3) = (0, 0)
We get the following equations.
 -4x + 2y + 11 = 0 (1)

2x - 2y - 3 = 0
2x = 2y + 3
x 2y + 3
= (2)
2
Substitute (2) in (1)

2y + 3 =0
-4 + 2y + 11
2
-4y - 6 + 2y + 11 =0
-2y = -5
y = 2.5

Substitute y = 2.5 in (2)

x 2(2.5) + 3
= (3)
2
x 8
=
2
x=4

Therefore, the only critical point of f is (4, 2.5)

Let us find all the second order derivatives of f.

𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿x 2 𝛿x 𝛿x 𝛿x𝛿y 𝛿x 𝛿y
𝛿 𝛿
= (-4x + 2y + 11) = (2x - 2y - 3)
𝛿x 𝛿x
= -4 =2

𝛿2f 𝛿 𝛿f 𝛿2f 𝛿 𝛿f
= =
𝛿y𝛿x 𝛿y 𝛿x 𝛿y 2 𝛿y 𝛿y
𝛿 𝛿
= (-4x + 2y + 11) = (2x - 2y - 3)
𝛿y 𝛿y
=2 = -2

The hessian matrix of f will be :

-4 2
Hf(x, y) =
2 -2
Hf(4, 2.5) will also be the same as the above matrix.

𝛿2f
det(Hf(4, 2.5)) = 4 and = -4
𝛿x 2
(4,2.5)

𝛿2f
Since det(Hf(4, 2.5)) > 0 and < 0, we can say that (4, 2.5) is a local maximum.
𝛿x 2
(4,2.5)

Let us observe P(x, y).

Note : (x - y - z) 2 = x 2 + y 2 + z 2 - 2xy - 2xz + 2yz

P(x, y) = -2x 2 + 2xy - y 2 + 11x - 3y - 7

= -x 2 - x 2 - y 2 + 2xy + 8x + 3x - 3y - 2.25 - 4.75
= -x 2 + 8x - 4.75 - x 2 - y 2 + 2xy + 3x - 3y - 2.25
= - x 2 - 8x + 4.75 - x 2 + y 2 - 2xy - 2(1.5)x + 2(1.5)y + 2.25
= - x 2 - 8x + 16 - 16 + 4.75 - x 2 + y 2 - 2xy - 2(1.5)x + 2(1.5)y + (1.5) 2
= - x 2 - 8x + 16 - 11.25 - (x - y - 1.5) 2
= - x 2 - 8x + 16 + 11.25 - (x - y - 1.5) 2
= -(x - 4) 2 - (x - y - 1.5) 2 + 11.25

Both -(x - 4) 2 and -(x - y - 1.5) 2 will always be lesser than or equal to zero. P(x, y) will be maximum
when both -(x - 4) 2 and -(x - y - 1.5) 2 become zero. This is possible at (4, 2.5). Therefore, (4, 2.5) is a
global maximum.

P(4, 2.5) = -(4 - 4) 2 - (4 - 2.5 - 1.5) 2 + 11.25

= 11.25

Therefore, the maximum profit in lakhs(is Rs.) is 11.25.