CONCEPTS COVERED

MULTIVARIABLE CALCULUS
❑ Maxima and Minima

o Necessary Conditions

o Worked Problems

1
Local Maximum or Minimum A function 𝑧 = 𝑓(𝑥, 𝑦) has a maximum (or a
2 −4𝑦 2
𝑧 = 4𝑥 2 + 𝑦 2 𝑒 −𝑥 minimum) at the point (𝑥0 , 𝑦0 ) if at every point
in a neighborhood of (𝑥0 , 𝑦0 ), the function
assumes a smaller value (or a larger value) than
at the point itself. Such a maximum (or
minimum) is called relative or local maximum
(or local minimum).

Maximum and minimum

values together are called
extreme values.

2
Absolute or Global Maximum/Minimum
The smallest and the largest values
𝑓 𝑥, 𝑦 = 𝑥2 + 2𝑦 2 , 𝑥2 + 𝑦2 ≤1
attained by a function over entire
domain including the boundary of the
domain are called absolute (or global)
minimum and absolute (or global)
maximum, respectively.

3
Critical point & Saddle Points

The point (𝑥0 , 𝑦0 ) is called critical point

(or stationary point) of 𝑓(𝑥, 𝑦) if
𝑓𝑥 (𝑥0 , 𝑦0 ) = 0 and 𝑓𝑦 𝑥0 , 𝑦0 = 0.
OR 𝑓𝑥 (𝑥0 , 𝑦0 ) and/or 𝑓𝑦 𝑥0 , 𝑦0 do/does
not exist.

A critical point where the

function has no minimum
or maximum is called a
saddle point.

4
Necessary condition for a function to have extremum

Let 𝑓(𝑥, 𝑦) be continuous and have first order partial derivatives at a point 𝑃(𝑎, 𝑏).
Then necessary conditions for the existence of an extreme value of it at the point 𝑃 are

𝑓𝑥 𝑎, 𝑏 = 0 & 𝑓𝑦 𝑎, 𝑏 = 0 (The point 𝑃 is a critical point)

OR

If the point 𝑃(𝑎, 𝑏) is a relative extremum of the function 𝑓(𝑥, 𝑦) then

𝑃(𝑎, 𝑏) is also a critical point of 𝑓(𝑥, 𝑦).

5
Necessary Condition for a function to have extremum

Let (𝑎 + ℎ, 𝑏 + 𝑘) be a point in the neighborhood of the point 𝑃(𝑎, 𝑏).

Then 𝑃 will be point of maximum if

Δ𝑓 = 𝑓 𝑎 + ℎ, 𝑏 + 𝑘 − 𝑓 𝑎, 𝑏 ≤ 0 for all sufficiently small ℎ & 𝑘

and a point of minimum if

Δ𝑓 = 𝑓 𝑎 + ℎ, 𝑏 + 𝑘 − 𝑓 𝑎, 𝑏 ≥ 0 for all sufficiently small ℎ & 𝑘

6
Necessary Condition for a function to have extremum

Taylor series expansion about the point (𝑎, 𝑏)

1 2
𝑓 𝑎 + ℎ, 𝑏 + 𝑘 = 𝑓 𝑎, 𝑏 + ℎ 𝑓𝑥 + 𝑘 𝑓𝑦 + ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑦𝑦 +⋯
𝑎,𝑏 2 𝑎,𝑏

Noting Δ𝑓 = 𝑓 𝑎 + ℎ, 𝑏 + 𝑘 − 𝑓 𝑎, 𝑏

1 2
Δ𝑓 = ℎ 𝑓𝑥 𝑎, 𝑏 + 𝑘 𝑓𝑦 𝑎, 𝑏 + ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑦𝑦 +⋯
2 𝑎,𝑏

For sufficiently small ℎ & 𝑘, the sign of Δ𝑓 will depend on the sign of
ℎ 𝑓𝑥 𝑎, 𝑏 + 𝑘 𝑓𝑦 𝑎, 𝑏

7
 1 2
Δ𝑓 = ℎ 𝑓𝑥 𝑎, 𝑏 + 𝑘 𝑓𝑦 𝑎, 𝑏 + ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑦𝑦 +⋯
2 𝑎,𝑏

Letting ℎ → 0, we get
ℎ ℎ − 1000ℎ2 − 2000ℎ3
1 2 0.1 −11.9
Δ𝑓 = 𝑘 𝑓𝑦 𝑎, 𝑏 + 𝑘 𝑓𝑦𝑦 (𝑎, 𝑏) + ⋯ 0.01 −0.092
2
0.001 −0.000002
Note that the sign of Δ𝑓 depends on the sign of
0.0001 0.000089998
𝑘 𝑓𝑦 𝑎, 𝑏 . That is
Assuming 𝒇𝒚 > 𝟎: Assuming 𝒇𝒚 < 𝟎:
Δ𝑓 > 0 for 𝑘 > 0 Δ𝑓 < 0 for 𝑘 > 0
Δ𝑓 < 0 for 𝑘 < 0 Δ𝑓 > 0 for 𝑘 < 0
Therefore the function cannot have an extremum unless 𝑓𝑦 = 0

8
 1 2
Δ𝑓 = ℎ 𝑓𝑥 𝑎, 𝑏 + 𝑘 𝑓𝑦 𝑎, 𝑏 + ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑦𝑦 +⋯
2 𝑎,𝑏

Similarly, letting 𝑘 → 0 we find that Δ𝑓 changes sign ℎ:

Assuming 𝑓𝑥 > 0: Assuming 𝑓𝑥 < 0:

Δ𝑓 > 0 for ℎ > 0 Δ𝑓 < 0 for ℎ > 0

Δ𝑓 < 0 for ℎ < 0 Δ𝑓 > 0 for ℎ < 0

Therefore the function cannot have an extremum unless 𝑓𝑥 = 0

Thus, the necessary conditions for the existence of an extremum at
the point (𝑎, 𝑏) is that 𝑓𝑥 (𝑎, 𝑏) = 0 & 𝑓𝑦 (𝑎, 𝑏) = 0

9
Problem - 1
Find all critical points of the function
𝑓(𝑥, 𝑦) = 𝑥 3 + 𝑦 3 − 3𝑥 − 12𝑦 + 20.
Critical points are obtained by solving
𝑓𝑥 (𝑥, 𝑦) = 0 & 𝑓𝑦 (𝑥, 𝑦) = 0

𝑓𝑥 (𝑥, 𝑦) = 0 ⟹ 3𝑥 2 − 3 = 0

𝑓𝑦 (𝑥, 𝑦) = 0 ⟹ 3𝑦 2 − 12 = 0

Critical Points are:

±1, ±2

10
 Local maximum
Problem - 2
Local minimum Find all critical points of the function
−𝑥2 −𝑦2
𝑓 𝑥, 𝑦 = 𝑥 2 − 𝑦 2 𝑒 2 .

𝑓𝑥 = 0 ⟹ 2 − (𝑥 2 − 𝑦 2 ) 𝑥 = 0

𝑓𝑦 = 0 ⟹ −2 − (𝑥 2 − 𝑦 2 ) 𝑦 = 0

Critical Points:

0, 0 , ± 2, 0 , 0, ± 2
Saddle point
Local minimum

11
 CONCLUSIONS

Necessary condition for extrema

𝑓𝑥 𝑎, 𝑏 = 0
𝑓𝑦 𝑎, 𝑏 = 0

Critical points are candidates for

• Local extrema
• Saddle points

12
CONCEPTS COVERED

MULTIVARIABLE CALCULUS
❑ Maxima and Minima

o Sufficient Conditions

o Worked Problems

13
Local Extrema (Previous Lecture)
A point (𝑎, 𝑏) will be a point of local extrema if
Δ𝑓 = 𝑓 𝑎 + ℎ, 𝑏 + 𝑘 − 𝑓(𝑎, 𝑏)
does not change its sign for all sufficiently small ℎ & 𝑘

Taylor ′ s Series Δ𝑓 = ℎ 𝑓𝑥 𝑎, 𝑏 + 𝑘 𝑓𝑦 𝑎, 𝑏
1 2
+ ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑦𝑦 𝑎,𝑏 + ⋯
2

Necessary Condition

𝑓𝑥 𝑎, 𝑏 = 0 & 𝑓𝑦 𝑎, 𝑏 = 0

14
Sufficient condition for a function to have extremum

Notation: 𝑟 = 𝑓𝑥𝑥 𝑎, 𝑏 , 𝑠 = 𝑓𝑥𝑦 𝑎, 𝑏 , 𝑡 = 𝑓𝑦𝑦 (𝑎, 𝑏)

Let a function 𝑓(𝑥, 𝑦) be continuous and have continuous second order partial derivatives
at a point 𝑃(𝑎, 𝑏). If 𝑃(𝑎, 𝑏) is a critical point, then the point 𝑃 is a point of

i. local maximum if 𝑟𝑡 − 𝑠 2 > 0 and 𝑟 < 0

ii. local minimum if 𝑟𝑡 − 𝑠 2 > 0 and 𝑟 > 0
iii. saddle point if 𝑟𝑡 − 𝑠 2 < 0
iv. test fails if 𝑟𝑡 − 𝑠 2 = 0 (some other way to characterize)

15
Sufficient condition for a function to have extremum

Consider Δ𝑓 = 𝑓 𝑎 + ℎ, 𝑏 + 𝑘 − 𝑓(𝑎, 𝑏)

1 2
Δ𝑓 = ℎ 𝑓𝑥 𝑎, 𝑏 + 𝑘 𝑓𝑦 𝑎, 𝑏 + ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑦𝑦 +⋯
2 𝑎,𝑏

Since (𝑎, 𝑏) is a critical point, 𝑓𝑥 𝑎, 𝑏 = 0 & 𝑓𝑦 𝑎, 𝑏 = 0, we have

1 2
Δ𝑓 = ℎ 𝑓𝑥𝑥 + 2ℎ𝑘 𝑓𝑥𝑦 + 𝑘 2 𝑓𝑘𝑘 +⋯
2 𝑎,𝑏

OR
1 2
Δ𝑓 = ℎ 𝑟 + 2ℎ𝑘 𝑠 + 𝑘 2 𝑡 + ⋯
2

16
 1 2
Δ𝑓 = ℎ 𝑟 + 2ℎ𝑘 𝑠 + 𝑘 2 𝑡 + ⋯
2

Assuming 𝑟 ≠ 0
1 2 2
Δ𝑓 = ℎ 𝑟 + 2ℎ𝑘 𝑟𝑠 + 𝑘 2 𝑟𝑡 + ⋯
2𝑟
1 2 2
Δ𝑓 = ℎ 𝑟 + 2ℎ𝑘 𝑟𝑠 + 𝑘 2 𝑠 2 − 𝑘 2 𝑠 2 + 𝑘 2 𝑟𝑡 + ⋯
2𝑟
1 2
Δ𝑓 = ℎ𝑟 + 𝑘𝑠 − 𝑘 2 𝑠 2 + 𝑘 2 𝑟𝑡 + ⋯
2𝑟
1 2
Δ𝑓 = ℎ𝑟 + 𝑘𝑠 + 𝑘 2 (𝑟𝑡 − 𝑠 2 ) + ⋯
2𝑟

17
 1 2
Δ𝑓 = ℎ𝑟 + 𝑘𝑠 + 𝑘 2 (𝑟𝑡 − 𝑠 2 ) + ⋯
2𝑟

Case – I: 𝑟𝑡 − 𝑠 2 > 0

Δ𝑓 > 0 if 𝑟 > 0

Δ𝑓 < 0 if 𝑟 < 0

The point (𝑎, 𝑏) is a point of minimum if 𝑟𝑡 − 𝑠 2 > 0, 𝑟>0

The point (𝑎, 𝑏) is a point of maximum if 𝑟𝑡 − 𝑠 2 > 0, 𝑟<0

18
 1 2
Δ𝑓 = ℎ𝑟 + 𝑘𝑠 + 𝑘 2 (𝑟𝑡 − 𝑠 2 ) + ⋯
2𝑟

Case – II: 𝑟𝑡 − 𝑠 2 < 0

Let 𝑘 → 0 & ℎ ≠ 0 ⟹ Δ𝑓 > 0 if 𝑟 > 0

Let 𝑘 ≠ 0 & choose ℎ such that ℎ𝑟 + 𝑘𝑠 = 0 ⟹ Δ𝑓 < 0 if 𝑟 > 0

⟹ The sign of Δ𝑓 depends on ℎ & k

Hence no maximum/minimum of 𝑓 can occur at 𝑃 𝑎, 𝑏 .

⟹ The point 𝑃 𝑎, 𝑏 is a saddle point

19
 1 2
Δ𝑓 = ℎ𝑟 + 𝑘𝑠 + 𝑘 2 (𝑟𝑡 − 𝑠 2 ) + ⋯
2𝑟

Case – III: 𝑟𝑡 − 𝑠 2 = 0

1 2
Δ𝑓 = ℎ𝑟 + 𝑘𝑠 +⋯
2𝑟

If we take ℎ & 𝑘 such that ℎ𝑟 = −𝑘𝑠, then the whole second order terms of the
right hand side will vanish.

Therefore, the conclusion will depend on the higher order terms.

One has to find some other way to investigate such points.

20
Working rules for investigating local extrema
• Find all critical points 𝑓𝑥 = 0 & 𝑓𝑦 = 0

• For each critical point, evaluate

𝑟 = 𝑓𝑥𝑥 , 𝑠 = 𝑓𝑥𝑦 , 𝑡 = 𝑓𝑦𝑦
• Identification
➢ If 𝑟𝑡 − 𝑠 2 > 0 & 𝑟 < 0 maximum
➢ If 𝑟𝑡 − 𝑠 2 > 0 & 𝑟 > 0 minimum
➢ If 𝑟𝑡 − 𝑠 2 < 0 Saddle point
➢ If 𝑟𝑡 − 𝑠 2 = 0 Test Fails

21
Example: Find all critical points of 𝑓 𝑥, 𝑦 = 𝑥 3 − 6𝑥 2 − 8𝑦 2 and investigate their nature
for local maximum/minimum and saddle point.

𝑓𝑥 = 0
Critical points: ⟹ 0,0 & (4,0)
𝑓𝑦 = 0

0,0 (4,0)
𝑟 = 𝑓𝑥𝑥 −12 12
𝑠 = 𝑓𝑥𝑦 0 0
𝑡 = 𝑓𝑦𝑦 −16 −16
𝑟𝑡 − 𝑠 2 192 −192

(0,0) is a point of local maximum & (4, 0) is a saddle point.

22
CONCLUSIONS • Necessary Conditions
𝑓𝑥 = 0 & 𝑓𝑦 = 0

• Sufficient Conditions

➢ If 𝑟𝑡 − 𝑠 2 > 0 & 𝑟 < 0 maximum

➢ If 𝑟𝑡 − 𝑠 2 > 0 & 𝑟 > 0 minimum
➢ If 𝑟𝑡 − 𝑠 2 < 0 saddle point
➢ If 𝑟𝑡 − 𝑠 2 = 0 needs further investigation

23