Scribd
Upload
9 views

Lecture6

The document discusses the interior point method as an alternative to the simplex method for solving linear programming problems, highlighting its advantages in terms of computational complexity. It outlines key concepts such as the primal and dual affine scaling algorithms, optimality conditions, and methods for finding initial interior feasible solutions. Additionally, it addresses performance improvements and potential issues related to dual degeneracy and numerical errors.

Uploaded by

die6o.andres.cas
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
9 views

Lecture6

The document discusses the interior point method as an alternative to the simplex method for solving linear programming problems, highlighting its advantages in terms of computational complexity. It outlines key concepts such as the primal and dual affine scaling algorithms, optimality conditions, and methods for finding initial interior feasible solutions. Additionally, it addresses performance improvements and potential issues related to dual degeneracy and numerical errors.

Uploaded by

die6o.andres.cas
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 64

LECTURE 6: INTERIOR

POINT METHOD
1. Motivation
2. Basic concepts
3. Primal affine scaling algorithm
4. Dual affine scaling algorithm
Motivation
• Simplex method works well in general, but suffers from
exponential-time computational complexity.
• Klee-Minty example shows simplex method may have to
visit every vertex to reach the optimal one.
• Total complexity of an iterative algorithm
= # of iterations x # of operations in each iteration
• Simplex method
- Simple operations: Only check adjacent extreme points
- May take many iterations: Klee-Minty example

Question: any fix?


Complexity of the simplex method
Worst case performance of the simplex method
Klee-Minty Example:
• Victor Klee, George J. Minty, “How good is the simplex
algorithm?’’ in (O. Shisha edited) Inequalities, Vol. III
(1972), pp. 159-175.
Klee-Minty Example
Klee-Minty Example
Karmarkar’s (interior point) approach
• Basic idea: approach optimal solutions from the interior of
the feasible domain

• Take more complicated operations in each iteration to find


a better moving direction
• Require much fewer iterations
General scheme of an interior point method
• An iterative method that moves in
the interior of the feasible domain
Interior movement (iteration)
• Given a current interior feasible solution , we have

> 0

An interior movement has a general format


Key knowledge
• 1. Who is in the interior?
- Initial solution

• 2. How do we know a current solution is optimal?


- Optimality condition

• 3. How to move to a new solution?


- Which direction to move? (good feasible direction)
- How far to go? (step-length)
Q1 - Who is in the interior?
• Standard for LP

• Who is at the vertex?


• Who is on the edge?
• Who is on the boundary?
• Who is in the interior?
What have learned before
Who is in the interior?
• Two criteria for a point x to be an interior feasible solution:

1. Ax = b (every linear constraint is satisfied)


2. x > 0 (every component is positive)

• Comments:
1. On a hyperplane ,
every point is interior relative to H.
2. For the first orthant K = ,
only those x > 0 are interior relative to K.
Example
How to find an initial interior solution?
• Like the simplex method, we have

- Big M method
- Two-phase method

(to be discussed later!)


Key knowledge
• 1. Who is in the interior?
- Initial solution

• 2. How do we know a current solution is optimal?


- Optimality condition

• 3. How to move to a new solution?


- Which direction to move? (good feasible direction)
- How far to go? (step-length)
Q2 - How do we know a current solution is optimal?
• Basic concept of optimality:
A current feasible solution is optimal if and only if
“no feasible direction at this point is a good direction.”

• In other words, “every feasible direction is not a good


direction to move!”
Feasible direction
• In an interior-point method, a feasible direction at a
current solution is a direction that allows it to take a
small movement while staying to be interior feasible.
• Observations:

- There is no problem to stay interior if the step-length is


small enough.
- To maintain feasibility, we need
Good direction
• In an interior-point method, a good direction at a
current solution is a direction that leads it to
a new solution with a lower objective value.

• Observations:
Optimality check
• Principle:
“no feasible direction at this point is a good direction.”
Key knowledge
• 1. Who is in the interior?
- Initial solution

• 2. How do we know a current solution is optimal?


- Optimality condition

• 3. How to move to a new solution?


- Which direction to move? (good feasible direction)
- How far to go? (step-length)
Q3 – How to move to a new solution?
1. Which direction to move?
- a good, feasible direction
“Good” requires

“Feasible” requires

Question: any suggestion?


A good feasible direction
• Reduce the objective value

• Maintain feasibility
Projection mapping
• A projection mapping projects the negative gradient vector
–c into the null space of matrix A
Q3 – How to move to a new solution?
2. How far to go?
- To satisfy every linear constraint
Since

the step-length can be real number.

- To stay to be an interior solution, we need


How to choose step-length?
• One easy approach
- in order to keep
>0
we may use the “minimum ratio test” to determine the
step-length.

Observation:
- when is close to the boundary, the step-length
may be very small.
Question: then what?
Observations
• If a current solution is near the center of the feasible
domain (polyhedral set), in average we can make a
decently long move.

• If a current solution is not near the center, we need to


re-scale its coordinates to transform it to become “near
the center".

Question: but how?


Where is the center?
• We need to know where is the “center” of the
non-negative/first orthant .
-Concept of equal distance
to the boundary

Question: If not,
what to do?
Concept of scaling
• Scale
• Define a diagonal scaling matrix
Transformation – affine scaling
• Affine scaling transformation

• The transformation is
1. one-to-one
2. onto
3. Invertible 4. boundary to boundary
5. interior to interior
Transformed LP
Step-length in the transformed space
• Minimum ratio test in the y-space
Property 1
• Iteration in the x-space
Property 2
• Feasible direction in x-space
Property 3
• Good direction in x-space
Property 4
• Optimality check (Lemma 7.2)
Property 5
• Well-defined iteration sequence (Lemma 7.3)

From properties 3 and 4, if the standard form LP


is bounded below and is not a constant, then
the sequence
is well-defined and strictly decreasing.
Property 6
• Dual estimate, reduced cost and stopping rule
We may define
Property 7
• Moving direction and reduced cost
Primal affine scaling algorithm
Example
Example
How to find an initial interior feasible solution?
• Big-M method Idea: add an artificial
variable with a big penalty

(big-M)

• Objective
Properties of (big-M) problem
(1) It is a standard form LP with n+1 variables and m
constraints.

(2) e is an interior feasible solution of (big-M).

(3)
Two-phase method
Properties of (Phase-I) problem
Facts of the primal affine scaling algorithm
More facts
Improving performance – potential push
• Idea: (Potential push method)
- Stay away from the boundary by adding a
potential push. Consider

Use
Improving performance – logarithmic barrier
• Idea: (Logarithmic barrier function method)
Consider

Properties:

(2)
Dual affine scaling algorithm
• Affine scaling method applied to the dual LP

• Idea:
Key knowledge
• Dual scaling (centering)
• Dual feasible direction
• Dual good direction – increase the dual objective value
• Dual step-length
• Primal estimate for stopping rule
Observation 1
• Dual scaling (centering)
Observation 2
• Dual feasibility (feasible direction)
Observation 3
• Increase dual objective function (good direction)
Observation 4
• Dual step-length
Observation 5
• Primal estimate
We define
Dual affine scaling algorithm
Find an initial interior feasible solution
Properties of (big-M) problem
Performance of dual affine scaling
• No polynomial-time proof !
• Computational bottleneck

• Less sensitive to primal degeneracy and numerical errors,


but sensitive to dual degeneracy.
• Improves dual objective value very fast, but attains primal
feasibility slowly.
Improving performance
1. Logarithmic barrier function method

Polynomial-time proof
Improving performance
• Power series method
- Basic idea: following a higher order trajectory
Power series expansion

You might also like