Several Variable Dierential Calculus

1. Motivations
(a) Examples of Real valued functions of several variables
(b) Motivation with real world problems like temperature distribution on any 3 di-
mensional object . The temperature function at time t and at point (x, y) has 3
variables. For example, the temperature distribution in a plate, (unit square) with
zero temperature at the edges and initial temperature (at time t = 0)T
0
(x, y) =
sin xsin y, is T(t, x, y) = e

2
kt
sin xsin y.
(c) Sound waves and water waves: The function u(x, t) = Asin(kx t) represents
the traveling wave of the initial wave front sinkx.
(d) Optimal cost functions, for example a manufacturing company wants to optimize
the resources, for their produce, like man power, capital expenditure, raw materials
etc. The cost function depends on these variables.
2. Limits
(a) Level surfaces, open and closed sets in IR
2
, Denition of limit, sequence
characterization, uniqueness of limit
(b) Example of function which has dierent limits along dierent straight lines.
Consider the function f(x, y)
f(x, y) =

xy
x
2
+y
2
(x, y) (0, 0)
0 (x, y) (0, 0)
Then along the straight lines y = mx, we get f(x, mx) =
m
1+m
2
. Hence limit does
not exist.
(c) Example of function which has dierent limits along dierent curves say y = mx
2
Consider the function f(x, y):
f(x, y) =

x
4
y
2
x
4
+y
2
(x, y) (0, 0)
0 (x, y) (0, 0)
Then along the curves y = mx
2
, we get f(x, mx
2
) =
1m
2
1+m
2
. Hence limit does not
exist.
(d) Example of function which has limit and explain how to obtain for a given
1
Consider the function f(x, y):
f(x, y) =

4xy
2
x
2
+y
2
(x, y) (0, 0)
0 (x, y) (0, 0)
Let > 0. Then 4|xy
2
| 2(x
2
+y
2
)(

x
2
+ y
2
) = 2
3/2
. Therefore may be chosen
such that
3/2
<

2
. For such , |f(x, y)| < .
(e) Example function for which limit exists, through polar coordinates
Consider the function
f(x, y) =

x
3
x
2
+y
2
(x, y) (0, 0)
0 (x, y) (0, 0)
.
Taking x = r cos , y = r sin , we get
|f(r, )| = |r cos
3
| r 0 as r 0.
(f) Example function where polar coordinates seem to give wrong conclusions
Consider the function f(x, y):
f(x, y) =

2x
2
y
x
4
+y
2
(x, y) (0, 0)
0 (x, y) (0, 0)
.
Taking x = r cos , y = r sin , we get
f(r, ) =
2r cos
2
sin
r
2
cos
4
+ sin
2

.
For any r > 0, the denominator is > 0. Since | cos
2
sin | 1, we tend to think
for a while that this limit goes to zero as r 0. But taking the path, y = mx
2
, we
see that the limit does not exist at (0, 0). So if we take the path r sin = r
2
cos
2
,
f(r, ) =
2r cos
2
sin
2r
2
cos
4

= 1.
3. Continuity, Partial derivatives and Directional derivatives:
(a) Denitions, examples for continuous functions, Partial derivatives
Example 1: Consider the function
f(x, y) =

xy
x
2
+y
2
(x, y) (0, 0)
0 (x, y) (0, 0)
2
As noted earlier, this is not a continuous function, but
f
x
(0, 0) = lim
h0
f(h, 0) f(0, 0)
h
= lim
h0
0 0
h
= 0.
Similarly, we can show that f
y
(0, 0) exists.
Remark: Also for a continuous function, partial derivatives need not exist. For
example f(x, y) = |x| +|y|. This is a continuous function at (0, 0). Indeed, for any
> 0, we can take < /2. But partial derivatives do not exist at (0, 0)
Example 2: The function f(x, y) =

xy

x
2
+y
2
x
2
+ y
2
= 0
0 x = y = 0
is continuous at the
origin.
Let > 0. Then |f(x, y) 0| = |x|
|y|

x
2
+y
2
|x|. So if we choose = , then
|f(x, y)| .
(b) Sucient condition for continuity:
Theorem: Suppose one of the partial derivatives exist at (a, b) and the other par-
tial derivative is bounded in a neighborhood of (a, b). Then f(x, y) is continuous
at (a, b).
Proof: Let f
y
exists at (a, b). Then
f(a, b + k) f(a, b) = kf
y
(a, b) +
1
k,
where
1
0 as k 0. Since f
x
exists and bounded in a neighborhood of at (a, b),
f(a + h, b + k) f(a, b) =f(a + h, b + k) f(a, b + k) + f(a, b + k) f(a, b)
=hf
x
(a + h, b + k) + kf
y
(a, b) +
1
k
hM + k|f
y
(a, b)| +
1
k
0 as h, k 0.
(c) Directional derivatives, Denition and examples
Let p = p
1

i + p
2

j be any unit vector. Then the directional derivative of f(x, y)

at (a, b) in the direction of p is
D
p
f(a, b) = lim
s0
f(a + sp
1
, b + sp
2
) f(a, b)
s
.
Example f(x, y) = x
2
+xy at P(1, 2) in the direction of unit vector u =
1

i+
1

j.
3
D
p
f(1, 2) = lim
s0
f(1 +
s

2
, 2 +
s

2
) f(1, 2)
s
= lim
s0
1
s
(
s
2
+ s(2

2 +
1

2
)
)
= 2

2 +
1

2
(d) Existence of partial derivatives does not guarantee the existence of directional
derivatives in all directions. For example take
f(x, y) =

xy
x
2
+y
2
x
2
+ y
2
= 0
0 x = y = 0
.
Let

p = (p
1
, p
2
) such that p
2
1
+ p
2
2
= 1. Then the directional derivative along p is
D
p
f(0, 0) = lim
h0
f(hp
1
, hp
2
) f(0, 0)
h
= lim
h0
p
1
p
2
h(p
2
1
+ p
2
2
)
exist if and only if p
1
= 0 or p
2
= 0.
4. Dierentiability
(a) Dierentiability, examples of dierentiable and non dierentiable functions
Denition: Let D be an open subset of IR
2
. A function f(x, y) : D IR is
dierentiable at a point (a, b) of D if there exists
1
= (h, k),
2
=
2
(h, k) such
that
f(a + hb + k) f(a, b) = hf
x
(a, b) + kf
y
(a, b) + h
1
+ k
2
,
where
1
,
2
0 as (h, k) (0, 0).
Notations:
i. f = f(a + hb + k) f(a, b), the total variation of f
ii. df = hf
x
(a, b) + kf
y
(a, b), the total dierential of f.
iii. =

h
2
+ k
2
Equivalent condition for dierentiability:
Theorem: f is dierentiable at (a, b) lim
0
f df

= 0.
Proof. Suppose f(x, y) is dierentiable. Then, there exists
1
,
2
such that
f(a + hb + k) f(a, b) = hf
x
(a, b) + kf
y
(a, b) + h
1
+ k
2
,
where
1
,
2
0 as (h, k) (0, 0).
4
Therefore, we can write
f df

=
1
(
h

) +
2
(
k

),
where df = hf
x
(a, b) + kf
y
(a, b) and =

h
2
+ k
2
. Now since |
h

| 1,
k

1 we
get
lim
0
f df

= 0.
On the other hand, if lim
0
fdf

= 0, then
f = df + , 0 as h, k 0.
Now write =
1
h +
2
k, for some
1
,
2
. Then using the inequality
1

2
(|h| +|k|)

h
2
+ k
2
(|h| +|k|)
we see that

2
(|h| +|k|)
1
h +
2
k (|h| +|k|)
This implies that
1
,
2
0 as h, k 0.
Example: Consider f(x, y) =

x
2
y
x
2
+y
2
x
2
+ y
2
= 0
0 x = y = 0
.
Partial derivatives exists at (0, 0), f
x
(0, 0) = f
y
(, 0) = 0. By taking h = cos , k =
sin ,
f df

=
h
2
k

3
=

3
cos
2
sin

3
= cos
2
sin .
The limit does not exist. Therefore f is not dierentiable at (0, 0).
Problem: Show that f(x, y) =

xsin
1
x
+ y sin
1
y
, xy = 0
0 xy = 0
is not dierentiable at
(0, 0).
Solution: |f(x, y)| |x| +|y| 2

x
2
+ y
2
implies that f is continuous at (0, 0).
Also
f
x
(0, 0) = lim
h0
f(h, 0) f(0, 0)
h
= 0.
f
y
(0, k) = lim
k0
f(0, k) f(0, 0)
k
0.
If f is dierentiable, then there exists
1
,
2
such that
f(h, k) f(0, 0) =
1
h +
2
k
5
where
1
,
2
0 as h, k 0. Now taking h = k, we get
f(h, h) = (
1
+
2
)h = 2hsin
1
h
= h(
1
+
2
).
So as h 0, we get sin
1
h
0, a contradiction.
Problem: Show that f(x, y) =

|xy| is not dierentiable at the origin.

Solution: Easy to check the continuity (take = ).
f
x
(0, 0) = lim
h0
0 0
h
= 0, andsimilarcalculationshows f
y
(0.0) = 0
So if f is dierentiable at (0, 0), then there exist,
1
,
2
such that
f(h, k) =
1
h +
2
k.
Taking h = k, we get
|h| = (
1
+
2
)h.
This implies that (
1
+
2
) 0.
The following theorem is on Sucient condition for dierentiability:
Theorem 0.0.1. Suppose f
x
(x, y) and f
y
(x, y) exist in an open neighborhood con-
taining (a, b) and both functions are continuous at (a, b). Then f is dierentiable
at (a, b).
Proof:
Since
f
y
is continuous at (a,b), there exists a neighborhood N(say) of (a,b) at
every point of which f
y
exists. We take (a+h, b+k), a point of this neighborhood
so that (a+h,b), (a,b+k) also belongs to N.
We write
f(a+h, b+k) f(a, b) = {f(a+h, b+k) f(a+h, b)} + {f(a+h, b) f(a, b)}.
Consider a function of one variable (y) = f(a + h, y).
Since f
y
exists in N, (y) is dierentiable with respect to y in the closed interval
[b,b+k] and as such we can apply Lagranges MVT for function of one variable y
in this interval and thus obtain
(b + k) (b) = k

(b + k), 0 < < 1

= kf
y
(a + h, b + k)
f(a + h, b + k) f(a + h, b) = kf
y
(a + h, b + k), 0 < < 1.
6
Now, if we write
f
y
(a + h, b + k) f
y
(a, b) =
2
(a function of h,k)
then from the fact that f
y
is continuous at (a,b). we may obtain

2
0 as (h, k) (0, 0).
Again because f
x
exists at (a, b) implies
f(a + h, b) f(a, b) = hf
x
(a, b) +
1
h,
where
1
0 as h 0. Combining all these we get
f(a + h, b + k) f(a, b) = k[f
y
(a, b) +
2
] + hf
x
(a, b) +
1
h
= hf
x
(a, b) + kf
y
(a, b) +
1
h +
2
k
where
1
,
2
are functions of (h, k) and they tend to zero as (h, k) (0, 0).
This proves that f(x, y) is dierentiable at (a,b). Remark: The above proof still
holds if f
y
is continuous and f
x
exists at (a, b).
(b) Dierentiable function with partial derivatives not continuous
Consider the function f(x, y) =

x
3
sin
1
x
2
+ y
3
sin
1
y
2
xy = 0
0 xy = 0
. Then f
x
(x, y) =

3x
2
sin
1
x
2
2 cos
1
x
2
xy = 0
0 xy = 0
Also f
x
(0, 0) = lim
h0
f(h,0)f(0,0)
h
= 0. So partial
derivatives are not continuous at (0, 0).
f(x, y) = (x)
3
sin
1
(x)
2
+ (y)
3
sin
1
(y)
2
= 0 + 0 +
1
x +
2
y
where
1
= (x)
2
sin
1
(x)
2
and
2
= (y)
2
sin
1
(y)
2
. It is easy to check that

1
,
2
0. So f is dierentiable at (0, 0).
(c) (Chain rule): Partial derivatives of composite functions
Let z = F(u, v) and u = (x, y), v = (x, y). Then z = F((x, y), (x, y)) as a
function of x, y. Suppose F, , have continuous partial derivatives, then we can
nd the partial derivatives of z w.r.t x, y as follows: Let x be increased by x,
keeping y constant. Then the increment in u is
x
u = u(x + x, y) u(x, y) and
similarly for v. Then the increment in z is (as z is dierentiable as a function of
7
u, v )

x
z := z(x + x, y + y) z(x, y) =
F
u

x
u +
F
v

x
v +
1

x
u +
2

x
v
Now dividing by x

x
z
x
=
F
u

x
u
x
+
F
v

x
v
x
+
1

x
u
x
+
2

x
v
x
Taking x 0, we get
z
x
=
F
u
u
x
+
F
v
v
x
+ ( lim
x0

1
)
u
x
+ ( lim
x0

2
)
v
x
=
F
u
u
x
+
F
v
v
x
similarly, one can show
z
y
=
F
u
u
y
+
F
v
v
y
.
Example: Let z = ln(u
2
+ v
2
), u = e
x+y
2
, v = x
2
+ y.
Then z
u
=
2u
u
2
+v
, z
v
=
1
u
2
+v
, u
x
= e
x+y
2
, v
x
= 2x. Then
z
x
=
2u
u
2
+ v
e
x+y
2
+
2x
u
2
+ v
(d) Derivative of Implicitly dened function
Let y = y(x) be dened as
F(x, y) = 0, y = y(x)
where F, F
x
, F
y
are continuous at (x
0
, y
0
) and F
y
(x
0
, y
0
) = 0. Then
dy
dx
=
F
x
F
y
at
(x
0
, y
0
).
Increase x by x, then y receives y increment and F(x +x, y +y) = 0. Also
0 = F = F
x
x + F
y
y +
1
x +
2
y
where
1
,
2
0 as x 0. This is same as
y
x
=
F
x
+
1
F
y
+
2
Now taking limit x 0, we get
dy
dx
=
F
x
F
y
.
Example: Dene the function y = y(x) as e
y
e
x
+ xy = 0. Then F
x
= e
x
+
y, F
y
= e
y
+ x. Then
dy
dx
=
e
x
y
e
y
+x
.
8
(e) If the function is dierentiable, then the directional derivative in the direction p
at (a, b)is
D
p
f(a, b) = f(a, b) p.
Proof: Let p = (p
1
, p
2
). Then from the denition,
lim
s0
f(a + sp
1
, b + sp
2
) f(a, b)
s
= lim
s0
f(x(s), y(s)) f(x(0), y(0))
s
where x(s) = a + sp
1
, y(s) = b + sp
2
.
From the chain rule,
lim
s0
f(x(s), y(s)) f(x(0), y(0))
s
=
f
x
(a, b)
dx
ds
+
f
y
(a, b)
dy
ds
= f(a, b) (p
1
, p
2
).
(f) Direction of maximum rate of change
Properties of directional derivative for dierentiable function
D
p
f = f p = |f| cos . So the function f increases most rapidly when cos = 1
or when p is the direction of f. The derivative in the direction
f
|f|
is equal to
|f|.
Similarly, f decreases most rapidly in the direction of f. The derivative in this
direction is D
p
f = |f|. Finally, the direction of no change is when =

2
. i.,e.,
p f.
Problem: Find the direction in which f = x
2
/2 + y
2
/2 increases and decreases
most rapidly at the point (1, 1). Also nd the direction of zero change at (1, 1).
(g) Example of functions for which all directional derivatives exist but not dieren-
tiable
f(x, y) =

y
|y|

x
2
+ y
2
, y = 0
0 y = 0
D
p
f(a, b) = lim
s0
f(sp
1
, sp
2
) f(0, 0)
s
=
p
2
|p
2
|
.
f
x
(0, 0) = lim
h0
0 0
h
= 0, f
y
(0, 0) = lim
k0
k
|k|

k
2
k
= 1.
In this case, both D
p
f(a, b) and f(a, b) p exists but not equal.
(h) Tangents and Normal to Level curves
Let f(x, y) be dierentiable and consider the level curve f(x, y) = c. Let

r (t) =
g(t)

i + h(t)

j be its parametrization. for example f(x, y) = x

2
+ y
2
has x(t) =
a cos t, y(t) = a sin t as level curve x
2
+ y
2
= a
2
, which is a circle of radius a. Now
9
dierentiating the equation f(x(t), y(t)) = a
2
with respect to t, we get
f
x
dx
dt
+ f
y
dy
dt
= 0.
Now since

r

(t) = x

(t)

i +y

(t)

j is the tangent to the curve, we can infer from the

above equation that f is the direction of Normal. Hence we have
Equation of Normal at (a, b) is x = a + f
x
(a, b)t, y = b + f
y
(a, b)t, t IR
Equation of Tangent is (x a)f
x
(a, b) + (y b)f
y
(a, b) = 0.
Example Find the normal and tangent to
x
2
4
+ y
2
= 2 at (2, 1).
Solution: f =
x
2

i + 2y

j
(2,1)
=

i + 2

j Tangent line through (2, 1) is (x +

2) + 2(y 1) = 0.
(i) Tangent Plane and Normal lines
Let

r (t) = g(t)

i + h(t)

j + k(t)

k is a smooth level curve(space curve) of the level

surface f(x, y, z) = c. Then dierentiating f(x(t), y(t), z(t)) = c with respect to t
and applying chain rule, we get
f(a, b, c) (x

(t), y

(t), z

(t)) = 0
Now as in the above, we infer the following:
Normal line at (a, b, c) is x = a + f
x
t, y = b + f
y
t, z = c + f
z
t.
Tangent plane: (x a)f
x
+ (y b)f
y
+ (z c)f
z
= 0.
Example: Find the tangent plane and normal line of f(x, y, z) = x
2
+y
2
+z9 =
0 at (1, 2, 4).
f +2x

i +2y

j +

k at (1,2,4) is = 2

i +4

j +

k and tangent plane is 2(x1) +4(y

2)(z 4) = 0. Then normal line is x = 1 + 2t, y = 2 + 4t, z = 4 + t.
Problem: Find the tangent line to the curve of intersection of two surfaces
f(x, y, z) = x
2
+ y
2
2 = 0, z R, g(x, y, z) = x + z 4 = 0.
The intersection of these two surfaces is an an ellipse on the plane g = 0. The
direction of normal to g(x, y, z) = 0 at (1, 1, 3) is

i +

k and normal to f(x, y, z) = 0

is 2

i + 2

j. The required tangent line is orthogonal to both these normals. So the

direction of tangent is
v = f g = 2

i 2

j 2

k.
Tangent through (1, 1, 3) is x = 1 + 2t, y = 1 2t, z = 3 2t.
(j) Linearization: Let f be a dierentiable function in a rectangle containing (a, b).
The linearization of a function f(x, y) at a point (a, b) where f is dierentiable is
10
the function
L(x, y) = f(a, b) + f
x
(a, b)(x a) + f
y
(a, b)(y b).
The approximation f(x, y) L(x, y) is the standard linear approximation of f(x, y)
at (a, b). Since the function is dierentiable,
E(x, y) = f(x, y) L(x, y) =
1
(x a) +
2
(y b)
where
1
,
2
0 as x a, y b. The error of this approximation is
|E(x, y)|
M
2
(|x a| +|y b|)
2
.
where M = max{|f
xx
|, |f
xy
|, |f
yy
|}.
Example: Find the linearization and error in the approximation of f(x, y, z) =
x
2
xy +
1
2
y
2
+ 3 at (3, 2).
f(3, 2) = 8, f
x
(3, 2) = 4, f
y
(3, 2) 1. So
L(x, y) = 8 + 4(x 3) (y 2) = 4x y 2
Also max{|f
xx
|, |f
xy
|, |f
yy
|} = 2 and
|E(x, y)| (|x 3| +|y 2|)
2
If we take the rectangle R : |x 3| 0.1, |y 2| 0.1.
5. Higher order mixed partial derivatives, Mean Value theorem, Taylors theorem and error
estimation
It is not always true that the second order mixed derivatives f
xy
=

x
(
f
y
) and f
yx
=

y
(
f
x
) are equal. The following is the example
Example: f(x, y) =

xy(x
2
y
2
)
x
2
+y
2
, x = 0, y = 0
0 x = y = 0
.
f
xy
(0, 0) = lim
h0
f
y
(h, 0) f
y
(0, 0)
h
f
y
(h, 0) = lim
k0
f(h, k) f(h, 0)
k
= lim
k0
1
k
hk(h
2
k
2
)
h
2
+ k
2
= h
Also f
y
(0, 0) = 0. Hence f
xy
(0, 0) = lim
h0
h0
h
= 1.
Now
f
x
(0, k) = lim
h0
f(h, k) f(0, k)
h
= lim
h0
1
h
hk(h
2
k
2
)
h
2
+ k
2
= k
11
and
f
yx
(0, 0) = lim
h0
f
x
(0, k) f
x
(0, 0)
k
= 1.
The following is the sucient condition for equality of mixed derivatives.
Theorem: If f, f
x
, f
y
, f
xy
, f
yx
are continuous in a neighbourhood of (a, b). Then
f
xy
(a, b) = f
yx
(a, b). But this is not a necessary condition as can be seen from the
following example
Example: f
xy
, f
yx
not continuous but mixed derivatives are equal. f(x, y) =

x
2
y
2
x
2
+y
2
x = 0, y = 0
0 x = y = 0
Calculate the mixed derivatives as in the previous example to see that f
xy
= f
yx
at
(0, 0).
Theorem: Suppose f(x, y) and its partial derivatives through order n+1 are continuous
throughout an open rectangular region R centered at a point (a, b). Then, throughout
R,
f(a + h, b + k) =f(a, b) + (hf
x
+ kf
y
)
(a,b)
+
1
2!
(h
2
f
xx
+ 2hkf
xy
+ k
2
f
yy
)
(a,b)
+
1
3!
(h
3
f
xxx
+ 3h
2
kf
xxy
+ 3hk
2
f
xyy
+ k
3
f
yyy
)
(a,b)
+ ... +
1
n!
(h

x
+ k

y
)
n
f
(a,b)
+
1
(n + 1)!
(h

x
+ k

y
)
n+1
f
(a+ch,b+ck)
where (a + ch, b + ck) is a point on the line segment joining (a, b) and (a + h, b + k).
Proof: Proof following by applying the Taylors theorem, Chain rule on the one dimen-
sional function (t) = f(x + ht, y + kt) at t = 0.
6. Maxima, Minima and Lagrange multiplies
(a) Maxima, minima, saddle points and derivative test, examples to nd local and
global extremum, boundary points as critical points
Derivative test:
From our discussion on one variable calculus, it is enough to determine the sign of
f = f(a + x, b + y) f(a, b). Now by Taylors theorem,
f =
1
2
(
f
xx
(a, b)(x)
2
+ 2f
xy
(a, b)xy + f
yy
(a, b)(y)
2
)
+ ()
3
where =

(x)
2
+ (y)
2
. We use the notation
A = f
xx
, B = f
xy
, C = f
yy
In polar form,
x = cos , y = sin
12
Then we have
f =
1
2
()
2
(
Acos
2
+ 2B cos sin + C sin
2
+ 2
)
.
Suppose A = 0, then
f =
1
2
()
2
(
(Acos + B sin )
2
+ (AC B
2
) sin
2

A
+ 2
)
Now we consider the 4 possible cases:
Case 1: Let AC B
2
> 0, A < 0. Then (Acos +B sin )
2
0, sin
2
0 implies
f =
1
2
()
2
(
m
2
+ 2
)
where m is independent of , 0 as 0. Hence for small, f < 0.
Hence (a, b) is a point of local maximum
Case 2: Let AC B
2
> 0, A > 0.
In this case
f =
1
2
()
2
(m
2
+ )
So f > 0. (a, b) is a point of local minimum.
Case 3:
i. Let AC B
2
< 0, A > 0. When we move along = 0, we have
f =
1
2
()
2
(A + 2) > 0.
for small. When we move along tan
0
= A/B, then
f =
1
2
()
2
(
AC B
2
A
sin
2

0
+ 2
)
< 0
for small. so we dont have constant sign along all directions. Hence (a, b)
is neither a point of maximum nor a point of minimum. Such point (a, b) is
called Saddle point.
ii. Let AC B
2
< 0, A < 0.
Similar as above the sign along path = 0 is < 0 and tan
0
= A/B is > 0.
iii. Let AC B
2
< 0, A = 0.
In this case B = 0 and
f =
1
2
()
2
(sin (2B cos + C sin ) + 2)
for small , 2B cos + C sin is close to 2B, but sin changes sign for > 0
13
or < 0. Again here (a, b) is a saddle point.
Case 4: Let AC B
2
= 0.
Again in this case it is dicult to decide the sign of f. For instance, if A = 0,
f =
1
2
()
2
(
(Acos + B sin )
2
A
+ 2
)
When = arctan(A/B), the sign of f is determined by the sign of . So
Additional investigation is required. No conclusion can be made with ACB
2
= 0.
We summarize the derivative test in two variables in the following table:
S.No. Condition Nature
1 AC B
2
> 0, A > 0 local minimum
2 AC B
2
> 0, A < 0 local maximum
3 AC B
2
< 0 Saddle point
4 AC B
2
= 0 No conclusion
Example: Find critical points and their nature of f(x, y) = xyx
2
y
2
2x2y+4
f
x
= y 2x 2 = 0, f
y
= x 2y 2 = 0
Therefore, the point (2, 2) is the only critical point. Also
f
xx
= 2, f
yy
= 2, f
xy
= 1.
Therefore, AC B
2
= 3 > 0 and A = 2 < 0. Therefore, (2, 2) is a point of
local maximum.
(b) The following is an example where the derivative test fails in contrast to one vari-
able case, for example
Example: Consider the function f(x, y) = (x y)
2
, then f
x
= 0, f
y
= 0 implies
x = y. Also, AC B
2
= 0. Moreover, all third order partial derivatives are zero.
so no further information can be expected from Taylors theorem.
(c) Global/Absolute maxima and Minima on closed and bounded domains:
i. Find all critical points of f(x, y). These are the interior points where partial
derivatives can be dened.
ii. Restrict the function to the each piece of the boundary. This will be one
variable function dened on closed interval I(say) and use the derivative test
of one variable calculus to nd the critical points that lie in the open interval
and their nature.
iii. Find the end points of these intervals I and evaluate f(x, y) at these points.
14
iv. The global/Absolute maximum will be the maximum of f among all these
points.
v. Similarly for global minimum.
Example: Find the absolute maxima and minima of f(x, y) = 2+2x+2yx
2
y
2
on the triangular region in the rst quadrant bounded by the lines x = 0, y = 0, y =
9 x.
Solution: f
x
= 2 2x = 0, f
y
= 2 2y = 0 implies that x = 1, y = 1 is
the only critical point and f(1, 1) = 4. f
xx
= 2, f
yy
= 2, f
xy
= 0. Therefore,
AC B
2
= 4 > 0 and A < 0. So this is local maximum.
case 1: On the segment y = 0, f(x, y) = f(x, 0) = 2+2xx
2
dened on I = [0, 9].
f(0, 0) = 2, f(9, 0) = 61 and the at the interior points where f

(x, 0) = 22x = 0.
So x = 1 is the only critical point and f(1, 0) = 3.
case 2: On the segment x = 0, f(0, y) = 2 + 2y y
2
and f
y
= 2 2y = 0 implies
y = 1 and f(0, 1) = 3.
Case 3: On the segment y = 9 x, we have
f(x, 9 x) = 61 + 18x 2x
2
and the critical point is x = 9/2. At this point f(9/2, 9/2) = 41/2.
nally, f(0, 0) = 2, f(9, 0) = f(0, 9) = 61. so the global maximum is 4 at (1, 1)
and minimum is 61 at (9, 0) and (0, 9).
7. Constrained minimization, substitution method and Method of Lagrange multipliers
Example: Find the shortest distance from origin to the plane z = 2x + y 5.
Here we minimize the function f(x, y, z) = x
2
+ y
2
+ z
2
subject to the constraint
2x + y z 5 = 0. Substituting the constraint in the function, we get
h(x, y) = f(x, y, 2x + y 5) = x
2
+ y
2
+ (2x + y 5)
2
.
The critical points of this function are
h
x
= 2x + 2(2x + y 5)(2) = 0, h
y
= 2y + 2(2x + y 5) = 0.
This leads to x = 5/3, y = 5/6. Then z = 2x + y 5 implies z = 5/6. We can check
that AC B
2
> 0 and A > 0. So the point (5/3, 5/6, 5/6) is a point of minimum.
Does this substitution method always work? The answer is NO. The following
example explains
Example: The shortest distance from origin to x
2
z
2
= 1.
This is minimizing f(x, y, z) = x
2
+y
2
+z
2
subject to the constraint x
2
z
2
= 1. Then
15
substituting z
2
= x
2
1 in f, we get
h(x, y) = f(x, y, x
2
1) = 2x
2
+ y
2
1
The critical points of the function are h
x
= 4x = 0, h
y
= 2y = 0. That is, x = 0, y =
0, z
2
= 1. But this point is not on the hyperbolic cylinder. To overcome this diculty,
we can substitute x
2
= z
2
+ 1 in f and nd that z = y = 0 and x = 1. These points
are on the hyperbolic cylinder and we can check that AC B
2
> 0, A > 0. This implies
the points are of local minimum nature.
8. Another way to solve this problem is by Lagrange Multiplier Method:
Imagine a small sphere centered at the origin. Keep increasing the radius of the sphere
untill the sphere touches the hyperbolic cylinder. The required smallest distance is
the radius of that sphere which touches the cylinder. When the sphere touches the
cylinder, both these surfaces has common tangent plane. So at the point of touching,
both surfaces has normal proportional.
That is f = g for some . Now solving this equations along with g = 0 gives the
points of extrema. In the above example, taking f = x
2
+y
2
+z
2
and g = x
2
z
2
1 = 0,
we get
2x

i + 2y

j + 2z

k = (2x

i 2z

k)
This implies, 2x = 2x, 2y = 0 2z = 2z. x = 0 does not satisfy g = 0. So from
rst equation we get = 1. Then 2z = 2z. That is z = 0, and y = 0. Therefore, the
critical points are (x, 0, 0). Substituting this in the constraint equation, we get x = 1.
Hence the points of extrema are (, 0, 0).
(a) Lagrange multipliers
Figure 1:
Caution: The Method of Lagrange multiplies gives only the points of extremum. To
nd the maxima or minima one has to compare the function values at these extremum
16
points.
9. Method of Lagrange multipliers with many constraints in n variables
(a) Number of constraints should be less than the number of independent variables
say g
1
= 0, g
2
= 0, ....g
m
= 0.
(b) Write the Lagrange multiplier equation: f =
m

i=1

i
g
i
.
(c) Solve the set of equations to nd the extremal points
f =
m

i=1

i
g
i
, g
i
= 0, i = 1, 2, ...m
(d) Once we have extremum points, compare the values of f at these points to deter-
mine the maxima and minima.
10. Some boundary value problems of Mathematical physics:
(a) Temperature distribution in a medium is governed by the equation:
u
t
= k
(

2
u
x
2
+

2
u
y
2
)
where t is the time and x, y represents the space variables that describe the medium.
This equation with the so called initial and boundary conditions which involves in
knowing initial temperature and the temperature on the boundary of the doamin
determines the temperature distribution in a midium.
(b) Wave equation:

2
u
t
2
= k
(

2
u
x
2
+

2
u
y
2
)
Here again t represent time and x, y the space variables.
Refereces:
1. Thomas Calculus Chapter 14
2. N. Piskunov, Dierential and Integral calculus
17