Lectures on Maths for Economics & Managements

Hồ Ngọc Kỳ

Institute of Applied Mathematics, UEH

Email: kyhn@ueh.edu.vn
Webpage: https://orcid.org/0000-0002-1362-9321

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 1 / 12
Lecture 7: Extrema of Multivariable Functions

CONTENT
(Chapter 17 of Textbook; Chương 6 trong sách Bài tập)

Free Extrema of Multivariable Functions

+ Definition
+ The Necessary Condition and Sufficient Conditions

Constrained Extrema of Multivariable Functions

+ Method of Reducing to a Function of One Variable
+ Method of Lagrange Multipliers

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 2 / 12
Extrema of Functions of Two Variables

Definition: Consider a function of two variables f : D ⊂ R2 → R.

If f (x, y ) ≤ f (a, b) in a neighborhood of (a, b), we say f has relative
maximum/ local maximum at (a, b).
If f (x, y ) ≥ f (a, b) in a neighborhood of (a, b), we say f has relative
minimum/ local minimum at (a, b).
If f (x, y ) ≤ f (a, b), ∀(x, y ) ∈ D we say f has absolute maximum/ global
maximum on D at (a, b). If f (x, y ) ≥ f (a, b), ∀(x, y ) ∈ D we say f has
absolute minimum/ global minimum on D at (a, b)

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 3 / 12
Extrema of Functions of Two Variables
Theorem (The necessary condition): If f has a relative extremum at (x0 , y0 ), and
fx (x0 , y0 ), fy (x0 , y0 ) exist, then

(*) fx (x0 , y0 ) = fy (x0 , y0 ) = 0.

In the following, we only consider the functions f such that fx and fy exist.

Definition: When (*) holds, we call (x0 , y0 ) a critical point/ stationary point of f .

Note: Condition (*) is only a necessary condition for a extreme point. It is

possible that f has no relative extremum at a critical point, and such a critical
point is called a saddle point of f .

Example: The function z = y 2 − x 2 has the

only stationary point O(0, 0), which is a
saddle point of z.

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 4 / 12
Sufficient Condition for Extrema
Consider a function of two variables f : D ⊂ Oxy → R. Let

fxx fxy
D1 := fxx and D2 := = fxx fyy − fxy2 .
fyx fyy

Theorem (The sufficient condition): Let P(x0 , y0 ) be a critical point of f (x, y )

and f has continuous second-order partial derivatives in a disk with center P.

If D2 (P) < 0, then P is a saddle point of f .

If D2 (P) > 0, then P is a relative extremum point of f . Precisely,

+ If D1 (P) > 0, then P is a relative minimum point of f .
+ If D1 (P) < 0, then P is a relative maximum point of f .

If D2 (P) = 0, then no conclusion about an extremum at P can be drawn.

If D1 > 0 and D2 > 0 on D, then f has a global minimum on D at P.

If D1 < 0 and D2 > 0 on D, then f has a global maximum on D at P.

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 5 / 12
Examples

Example: Find all relative extrema and saddle points of the function

f (x, y ) = x 4 + y 4 − 4xy + 1.

Example: Find all relative extrema and saddle points of the function

8 x
f (x, y ) = + + y.
x y

Example: Find all relative extrema and saddle points of the function

f (x, y ) = x 3 + y 3 − 3x − 12y .

Example: Let f (x, y ) = −x 2 − xy − y 2 + 4x − my + 2m, where m ∈ R is a

parameter. Find m such that f attains a global maximum at (2, 0).

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 6 / 12
Extrema of z = f (x1 , x2 , ..., xn )
Suppose that z = f (x1 , x2 , ..., xn ) has continuous second-order partial derivatives.
Definition: Hessian matrix of z = f (x1 , x2 , ..., xn ) is the matrix
 
fx1 x1 fx1 x2 · · · fx1 xn
fx2 x1 fx2 x2 · · · fx2 xn 
H= . .. .. 
 
 .. . . 
fxn x1 fxn x2 · · · fxn xn

fx1 x1 fx1 x2 ··· fx1 xi

fx2 x1 fx2 x2 ··· fx2 xi
Denote Di = . .. .. (i = 1, · · · , n).
.. . .
fxi x1 fxi x2 ··· fxi xi
Theorem: f only attains a local extremum at a stationary point P(x1 , · · · , xn ),
namely, fx1 (P) = · · · = fxn (P) = 0. Furthermore, at a stationary point P we have
if Di > 0 for all i = 1, · · · , n, then P is a local minimum of f ;
if (−1)i Di > 0 for all i = 1, · · · , n, then P is a local maximum of f .
Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 7 / 12
Constrained Extrema

To find the maximum and minimum values of the function z = f (x, y )

subject to the constraint g (x, y ) = C .

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 8 / 12
Method of Reducing to a Function of One Variable

Reducing to a function of one variable:

From g (x, y ) = C , we express y in terms of x or x in terms of y .

Substituting y = y (x) into f (x, y ), we obtain z = h(x) = f (x, y (x)).

To find the extrema of the function h(x), note that if h attains an extremum
at x0 , then the two-variable function f attains an extremum at (x0 , y (x0 )).

Example: Find the extrema of the function

z = f (x, y ) = x 2 − 3xy + 12x

under the constraint 2x + 3y = 6.

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 9 / 12
Method of Lagrange Multipliers

Let us consider the Lagrangian function

L (x, y , λ) = f (x, y ) + λ(g (x, y ) − C ),

where λ is called Lagrange multiplier. The maximum (resp. minimum) values of L

are also the maximum (resp. minimum) values of f under the constraint:

(*) g (x, y ) = C .

Step 1 (Finding stationary points): Find stationary points of L satisfying (*):

 
Lx = fx (x, y ) + λgx (x, y ) = 0
 λ = λ0

Ly = fy (x, y ) + λgy (x, y ) = 0 ⇐⇒ x = x0
 
g (x, y ) − C =0 y = y0 .
 

We say that (x0 , y0 ) is a stationary point corresponding to Lagrange multiplier λ0 .

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 10 / 12
Method of Lagrange Multipliers
Step 2 (Check the sufficient condition): Suppose that P(x0 , y0 ) is a stationary
point corresponding to Lagrange multiplier λ0 . Put
 
0 gx gy
H̄ =  gx Lxx Lyx 
gy Lxy Lyy

If det(H̄(P)) > 0, then f has a relative maximum at P.

If det(H̄(P)) < 0, then f has a relative minimum at P

Example:

(1) Find the maximum and minimum values of the function z = x + y with
constraint x 2 + y 2 = 1.

(2) Find the maximum and minimum values of the function z = x + y with
constraint x 2 + y 2 + xy = 1.
Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 11 / 12
 Example: To produce a certain type of product, two types of raw materials that
can substitute for each other are used. The production output is determined by:
2 1
Q = 3x 3 y 3 ,

where x is the quantity of the first raw material and y is the quantity of the
second raw material. The prices of these two raw materials are P1 = 2 and
P2 = 1, respectively. Find x and y such that the cost of producing 100 units of
output is minimized.

Hồ Ngọc Kỳ - Institute of Applied Mathematics Lectures on Maths for Economics & Managements Lecture 7: Extrema of Multivariable Funct. 12 / 12