BASIC MATHEMATICS FOR

ECONOMICS AND BUSINESS

Optimization with Equality Constraints
After watching this video, students should be able to solve
optimization problem with equality constraints.

Prepared by: Ledi Trialdi, MPP.

  EFFECTS OF A CONSTRAINT
 L A G R A N G E M U LT I P L I E R M E T H O D
n-variable case
Multi-constraint case
 SECOND-ORDER CONDITIONS
The bordered Hessian
n-variable case
 E X A M P L E A N D A P P L I C AT I O N

LEDI TRIALDI
lecturer
 EFFECTS OF A CONSTRAINT

Objective:
𝑀𝑎𝑥.𝑈=𝑥 1 𝑥 2 +2𝑥 1
Wi t h o u t
CONSTRAINT
Solution : 𝑥1∗ = ∞ , 𝑥 2∗ =∞ U has no maximum point)

Wi t h C O N S T R A I N T
Objective:
𝑀𝑎𝑥.𝑈=𝑥 1 𝑥 2 +2𝑥 1
Constraint:

2
After substitution:
𝑈 =𝑥 1 ( 30 −2 𝑥1 ) + 2 𝑥 1=3 2 𝑥1 −2 𝑥1
𝑑𝑈
=3 2 − 4 𝑥1 =0 (Optimization with single variable)
𝑑 𝑥1
Sumber: Chiang & Wainwright (2005)

Solution : 𝑥1∗ =8 , 𝑥 2∗ =14 ,𝑈 ∗ =128

(Constrained maximum)
 L A G R A N G E M U LT I P L I E R M E T H O D

Constraint
𝑠.𝑡 . 4 𝑥 1+2𝑥 2=60
𝑍=𝑥1 𝑥2 +2𝑥 1+ 𝜆(60− 4 𝑥 1 − 2𝑥 2 ) LAGRANGIAN
function
Objectiv a Lagrange multiplier, treated as an additional
e variable:

(
𝑍1 ≡
𝜕𝑍
𝜕 𝑥1 )
=𝑥 2 +2 − 4 𝜆=0
𝑥 2 +2
= 2↔ 𝑥2 =2 𝑥 1 − 2
(
𝑍2 ≡
𝜕𝑍
𝜕 𝑥2 )
=𝑥1 −2 𝜆=0 𝑥 1

(
𝑍𝜆 ≡
𝜕𝑍
𝜕𝜆 )
=60 − 4 𝑥 1 − 2 𝑥 2=0 → 8 𝑥1 =64
60 − 4 𝑥 1 − 2(2 𝑥 1 − 2)=0

Solution :
∗ ∗ ∗
𝑥1 =8 , 𝑥 2 =14 ,𝑈 =128
(Constrained maximum)
 L A G R A N G E M U LT I P L I E R M E T H O D

In general : n − variable case

Maximize/minimize 𝑧= 𝑓 (𝑥 , 𝑦) 𝑧= 𝑓 (𝑥 1 , 𝑥 2 , … , 𝑥 𝑛 )
Subject to 𝑔 (𝑥 , 𝑦 )=𝑐 𝑔 (𝑥1 , 𝑥 2 ,… , 𝑥 𝑛)=𝑐

Lagrangian function 𝑍= 𝑓 (𝑥 , 𝑦)+ 𝜆[𝑐 − 𝑔(𝑥 , 𝑦)] 𝑍 = 𝑓 (𝑥 1 , 𝑥 2 , … , 𝑥 𝑛 )+𝜆 [𝑐 −𝑔 (𝑥1 , 𝑥 2 , … , 𝑥 𝑛 )]

Necessary condition
...... ...... ......

∗
∗ 𝑑𝑍 measures the sensitivity of Z
𝜆 = to changes in the constraint
𝑑𝑐
 L A G R A N G E M U LT I P L I E R M E T H O D

Multiconstraint

Maximize/minimize 𝑧= 𝑓 (𝑥 1 , 𝑥 2 , … , 𝑥 𝑛 )

Subject to 𝑔 (𝑥1 , 𝑥 2 ,… , 𝑥 𝑛)=𝑐

h (𝑥1 , 𝑥 2 , … , 𝑥𝑛 )=𝑑

Lagrangian function 𝑍 = 𝑓 ( 𝑥 1, 𝑥 2 , … , 𝑥 𝑛 ) + 𝜆 [ 𝑐 − 𝑔 ( 𝑥 1 , 𝑥 2 , … , 𝑥 𝑛 ) ] +𝜇 [ 𝑑 − h ( 𝑥1 , 𝑥 2 ,… , 𝑥 𝑛 ) ]

Necessary condition
 SECOND-ORDER CONDITIONS
n − variable case

Objective function 𝑧= 𝑓 (𝑥 1 , 𝑥 2 , … , 𝑥 𝑛 ) The

HESSIAN
Constraint function 𝑔 (𝑥1 , 𝑥 2 ,… , 𝑥 𝑛)=𝑐 Bordered

| |
0 𝑔1 𝑔2 … 𝑔𝑛
𝑔1 𝑍 11 𝑍 12 … 𝑍1𝑛
Lagrangian function 𝑍 = 𝑓 (𝑥 1 , 𝑥 2 , … , 𝑥 𝑛 )+𝜆 [𝑐 −𝑔 (𝑥1 , 𝑥 2 , … , 𝑥 𝑛 )] |𝐻|= 𝑔 2 𝑍 21 𝑍 22 … 𝑍2𝑛
1st-order … … … … …
...... ...... ...... 𝑔𝑛 𝑍 𝑛1 𝑍 𝑛2 … 𝑍 𝑛𝑛
Necessary

| |
condition 0 𝑔1 𝑔2
|𝐻 2|= 𝑔 1 𝑍 11 𝑍 12
𝑔2 𝑍 21 𝑍 22

| |
(𝑑¿¿2 𝑧 <0)¿ 0 𝑔1 𝑔2 𝑔3
2 order
|𝐻 2|>0 ;| 𝐻 3|< 0 ; … ; ¿
nd-
Max. 𝑔 𝑍 11 𝑍 12 𝑍 13
Sufficient |𝐻 3|= 𝑔1 𝑍 21 𝑍 22 𝑍 23
condition Min. |𝐻 2| ,|𝐻 3|, … ,|𝐻 𝑛|<0 2
𝑔3 𝑍 31 𝑍 32 𝑍 33
(𝑑¿¿2 𝑧 >0)¿
 E X A M P L E A N D A P P L I C AT I O N
EXAMPLE 1

Objective function 𝑧= 𝑓 (𝑥 , 𝑦) → 𝑧=𝑥𝑦 The

HESSIAN
Constraint function 𝑔 (𝑥 , 𝑦 )=𝑐 → 𝑥 + 𝑦=6 Bordered

| |
0 𝑔𝑥 𝑔𝑦
Lagrangian function 𝑍 = 𝑥𝑦 + 𝜆 [ 6 − 𝑥 − 𝑦 ] |𝐻|=|𝐻 2|= 𝑔 𝑥 𝑍 𝑥𝑥 𝑍 𝑥𝑦
1st-order 𝑔𝑦 𝑍 𝑦𝑥 𝑍 𝑦𝑦

| |
Necessary 0 1 1
, ¿ 1 0 1 =2
condition
1 1 0

(𝑑¿¿2 𝑧 <0)¿ ∴ 𝑧∗=9 is a maximum

2 order
|𝐻 2|>0
nd-
Max.
Sufficient
condition
 E X A M P L E A N D A P P L I C AT I O N
EXAMPLE 2

Objective function 𝑧= 𝑓 (𝑥 1 , 𝑥 2→
) 𝑧=𝑥 12 + 𝑥22 The
HESSIAN
Constraint function 𝑔 (𝑥1 , 𝑥 2 )=𝑐→ 𝑥 1 + 4 𝑥2 =2 Bordered

| |
0 𝑔1 𝑔2
Lagrangian function 𝑍 =𝑥12 + 𝑥 22 + 𝜆 [2 − 𝑥 1 − 4 𝑥 2 ] |𝐻|=|𝐻 2|= 𝑔1 𝑍 11 𝑍 12
1 order
st- 𝑔2 𝑍 21 𝑍 22

| |
Necessary , 0 1 4
condition ¿ 1 2 0 =− 34
4 0 2

4
∴ 𝑧∗= is a minimum
17
2nd-order
Sufficient
condition Min. |𝐻 2|<0
(𝑑¿¿2 𝑧 >0)¿
 E X A M P L E A N D A P P L I C AT I O N
BUDGET
M I N I M I Z AT I O N
Objective function 𝐵= 𝑓 ( 𝑥 , 𝑦 )

Constraint function 𝑈 (𝑥 , 𝑦 )=𝑈

1 /2 1 /2
→𝑥 𝑦 =2

Lagrangian function 𝑍 = 40 𝑥+ 10 𝑦 + 𝜆[ 2− 𝑥1 /2 𝑦 1 /2 ]

1st-order
1 −1/2 1/2
Necessary
𝑍 𝑥=40− 𝜆𝑥 𝑦 =0 𝑦
=4 → 𝑦 =4 𝑥
condition
2 𝑥

Z 𝜆 =2 − 𝑥
1/ 2
𝑦
1/ 2
=0 2 − 𝑥 1/ 2 ( 4 𝑥 )1 / 2=0
2 𝑥=2
∗
𝑥 =1
∗
𝑦 =4 𝑍 ∗ =𝐵∗ =80
∗
𝜆 =40
 E X A M P L E A N D A P P L I C AT I O N
BUDGET
M I N I M I Z AT I O N

| |
Objective function 𝐵= 𝑓 ( 𝑥 , 𝑦 ) 0 𝑔𝑥 𝑔𝑦
Constraint function 𝑈 (𝑥 , 𝑦 )=𝑈 1 /2 1 /2 |𝐻|=|𝐻 2|= 𝑔 𝑥 𝑍 𝑥𝑥 𝑍 𝑥𝑦
→ 𝑥 𝑦 =2
𝑔𝑦 𝑍 𝑦𝑥 𝑍 𝑦𝑦
at
Lagrangian function 𝑍 = 40 𝑥+ 10 𝑦 + 𝜆[ 2− 𝑥 1 /2 1 /2
𝑦 ] 𝑔 𝑥 =𝑈 𝑥 = 1 𝑥 −1 /2 𝑦 1 /2=1
2
1 order
st-
1 −1/2 1/2 1 1 /2 − 1/ 2 1
Necessary
𝑍 𝑥=40− 𝜆𝑥 𝑦 =0 𝑔 𝑦 =𝑈 𝑦 = 𝑥 𝑦
2
=
4
condition
2 1
𝑍 𝑥𝑥 = 𝜆 𝑥
− 3 /2 1 /2
𝑦 =20
Z 𝜆 =2 − 𝑥
1/ 2
𝑦
1/ 2
=0 4
1 1 /2 − 3/ 2 5
𝑍 𝑦𝑦 = 𝜆 𝑥 𝑦 =
4 4
1 1
1 − −
𝑍 𝑥𝑦 =𝑍 𝑦𝑥 =− 𝜆 𝑥 𝑦 2 =− 5
2

2nd-order 4

| |
Sufficient 1
0 1
condition Min. |𝐻 2|<0 |𝐻|=|𝐻 2|= 1 20
4
− 5 =− 5
(𝑑¿¿2 𝑧 >0)¿ ∗
∴ 𝐵 =80 is a minimum budget 1
−5
5
4 4
Thank You