OPTIMIZATION

Nguyễn Thị Thảo

Faculty of Economic Mathematics

Email: [email protected]

27/9/2022

1 / 17
Contents

1 Introduction

2 Mathematics Preliminaries

3 Unconstrained Optimization

4 Constrained Optimization

5 Linear programming problems

2 / 17
Materials

Edwin K. P. Chong, Stanislaw H. Zak, 2013, An Introduction to

Optimization, 4th edition, Wiley.

Daniel Leonard, Ngo Van Long, 1992, Optimal control theory and
static optimization in economics, Cambridge University Press.

Chiang Alpha.C, 2006, Fundamental Methods of Mathematical

Economics, 4th edition, McGraw-Hill.

Stephen P.Bradly, Arnoldo C. Hax, Thomas L.Magnanti, 1997

Applied Mathematical Programming, Addison-Wesley.

3 / 17
What is Optimization?

An approach used to find the combination of inputs to achieve the best

possible output satisfying certain pre-specified constraints and conditions.

4 / 17
Example 1 : Manufacturing Problem

m: number of types of resources.

bi : value of ith resource, i = 1, . . . , m
n: number of products
aij : units of ith resource to produce 1 unit of jth product (i = 1, . . . , m,
j = 1, . . . , n)
pj : price of 1 unit of jth product (j = 1, . . . , n)
Question: Construct the problem to find the number of each product to
maximize revenue.

5 / 17
Example 1 : Manufacturing Problem

m: number of types of resources.

bi : value of ith resource, i = 1, . . . , m
n: number of products
aij : units of ith resource to produce 1 unit of jth product (i = 1, . . . , m,
j = 1, . . . , n)
pj : price of 1 unit of jth product (j = 1, . . . , n)
Question: Construct the problem to find the number of each product to
maximize revenue.
Pn
maximize j =1 pj xj ,
n
X
subject to: aij xj ≤ bi (i = 1, . . . , m),
j =1

xj ≥ 0 (j = 1, . . . , n).
6 / 17
Example 2: Portfolio optimization

n assets
wi : allocation of ith asset, for i = 1, . . . , n
ri : return of ith asset. Assume we know E (ri ), covariance matrix.
Question: Find w =(w1 , . . . , wn ) to minimize the risk of portfolio subjected
to the constraints on expected value of portfolio return.

7 / 17
Example 2: Portfolio optimization

n assets
wi : allocation of ith asset, for i = 1, . . . , n
ri : return of ith asset. Assume we know E (ri ), covariance matrix.
Question: Find w =(w1 , . . . , wn ) to minimize the risk of portfolio subjected
to the constraints on expected value of portfolio return.

minimize Var (w1 r1 + w2 r2 + · · · + wn rn ),

subject to: w1 + w2 + · · · + wn = 1,
E (w1 r1 + · · · + wn rn ) ≥ R .

8 / 17
Example 3: Shortest Path

n: number of places
dij : distance from place i to place j (i = 1, . . . , n, j = 1, . . . , n)
Question: A man can start at any place. Find the shortest path for the
man to walk through all other n − 1 places once and then return to the
starting place.

9 / 17
Example 3: Shortest Path

n: number of places
dij : distance from place i to place j (i = 1, . . . , n, j = 1, . . . , n)
Question: A man can start at any place. Find the shortest path for the
man to walk through all other n − 1 places once and then return to the
starting place.
Pn Pn
minimize i =1 j =1 dij xij ,
n
X
subject to: xij = 1, j = 1, . . . , n,
i =1
X n
xij = 1, i = 1, . . . , n,
j =1

xij ∈ {0, 1}, i = 1, . . . , n, j = 1, . . . , n,

ui − uj + nxij ≤ n − 1, i = 1, . . . , n, j = 1, . . . , n.
10 / 17
Optimization Problem

Optimization Problem:

minimize f (x),
s.t: x ∈ D.

x = (x1 , . . . , xn )T : decision variables.

f (x): objective function or cost function.
The set D ⊂ Rn : constraint set or feasible set.
D = Rn : unconstrained problem

11 / 17
Optimization Problem

Optimization Programming:

minimize f (x),

s.t: hi (x) = 0 (i = 1, . . . , m),

g j (x) ≥ 0 (j = 1, . . . p)
x ∈ Rn

hi (x), g j (x): constraint function

hi (x), g j (x): continuous differentiable functions

12 / 17
Definition

Feasible point: x satisfying all the constraints

Feasible set: All feasible points.
Globally optimal solution (global minimizer): A point x∗ ∈ D is a
globally optimal solution of f over D if f (x) ≥ f (x∗ ) for all x ∈ D.

x∗ ∈ argminx∈D f (x)

Locally optimal solution/ local minimizer: A point x∗ ∈ D is a

locally optimal solution of f over D if exist ϵ > 0 such that
f (x) ≥ f (x∗ ) for all x ∈ D and ∥ x − x∗ ∥< ϵ.
We call a problem having a globally optimal solution a solvable
problem.

Question: When is a problem unsolvable?

13 / 17