9 Lagrangian Relaxation (1)

9 Lagrangian Relaxation (1) - IEE 598 - 9 (II.3.6)...

Info iconThis preview shows pages 1–7. Sign up to view the full content.

View Full Document Right Arrow Icon
IEE 598 - 9 (II.3.6) Lagrangian Relaxation Muhong Zhang DEPARTMENT OF INDUSTRIAL ENGINEERING Mar. 04, 2011
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Lagrangian Relaxation Recall that for LP relaxation, we dropped the integrality constraint. For an optimization problem z IP = max { cx : Ax b , Dx d , x Z n + } , by dropping some “complicating” constraints Dx d , we get a relaxation of the original IP. However, such relaxation might be weak. We might try to penalizing the violation of dropping constraints. Lagrangian relaxation: LR ( λ ) : z LR ( λ ) = max { cx - λ ( Dx - d ) : Ax b , x Z n + } Proposition: Problem LR ( λ ) is a relaxation of the IP problem for all λ 0. Corollary: z IP z LR ( λ ) for all λ 0. Zhang IEE 598 OPT II 2 / 19
Background image of page 2
Lagrangian Relaxation Lagrangian relaxation gives an upper bound of the IP problem. To determine the best λ , we solve the following Lagrangian Dual problem: z LD = min λ 0 z LR ( λ ) . Corollary: z IP z LD . Proposition: Let x ( λ ) be an optimal solution for LR ( λ ) . If (1) Dx ( λ ) d and (ii) λ ( d - Dx ) = 0, then x ( λ ) is optimal for the IP problem. Zhang IEE 598 OPT II 3 / 19
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Uncapacited facility location problem Variables: x j : indicator of build facility j , j N ; y ij : proportion of demand i to be meet by facility j , i I , j N . Formulation: max i I j N c ij y ij - j N f j x j s.t. j N y ij = 1 i I y ij x j i I , j N y R mn + , x B n Let Q = { ( x , y ) B n × R mn + : y ij x j , i I , j N } . LR ( λ ) : Zhang IEE 598 OPT II 4 / 19
Background image of page 4
Lagrangian Dual 1 How to solve the Lagrangian Dual problem? 2 How strong is the z LD ? 3 When is z LD = z IP ? Zhang IEE 598 OPT II 5 / 19
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
The Strength of the Lagrangian Dual LR ( λ ) : z LR ( λ ) = max { cx + λ ( d - Dx ) : Ax b , x Z n + } Let Q = { x Z n + : Ax b } and S = { x Q
Background image of page 6
Image of page 7
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 05/01/2011 for the course IEE 598 taught by Professor Hillary during the Spring '10 term at USC.

Page1 / 19

9 Lagrangian Relaxation (1) - IEE 598 - 9 (II.3.6)...

This preview shows document pages 1 - 7. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online