The Multi-Point Separation Strategy as a Column GenerationStabilization Technique
VANDERBECK, François
Reformulations based algorithms for Combinatorial Optimization [Realopt]
Institut de Mathématiques de Bordeaux [IMB]
Reformulations based algorithms for Combinatorial Optimization [Realopt]
Institut de Mathématiques de Bordeaux [IMB]
SADYKOV, Ruslan
Institut de Mathématiques de Bordeaux [IMB]
Reformulations based algorithms for Combinatorial Optimization [Realopt]
See more >
Institut de Mathématiques de Bordeaux [IMB]
Reformulations based algorithms for Combinatorial Optimization [Realopt]
VANDERBECK, François
Reformulations based algorithms for Combinatorial Optimization [Realopt]
Institut de Mathématiques de Bordeaux [IMB]
Reformulations based algorithms for Combinatorial Optimization [Realopt]
Institut de Mathématiques de Bordeaux [IMB]
SADYKOV, Ruslan
Institut de Mathématiques de Bordeaux [IMB]
Reformulations based algorithms for Combinatorial Optimization [Realopt]
< Reduce
Institut de Mathématiques de Bordeaux [IMB]
Reformulations based algorithms for Combinatorial Optimization [Realopt]
Language
en
Communication dans un congrès
This item was published in
INFORMS Annual Meeting 2014, 2014-11-09, San Francisco.
English Abstract
Column generation suffers from slow convergence. Algorithmic strategies have been designed to accelerate convergence of cutting plane algorithms in mixed integer programming. They have their counterpart in column generation. ...Read more >
Column generation suffers from slow convergence. Algorithmic strategies have been designed to accelerate convergence of cutting plane algorithms in mixed integer programming. They have their counterpart in column generation. We show that the "Multi-Point Separation" strategy translates into a column generation stabilization technique that consists in restricting the dual solution to be in the convex hull of the selected multi-point set.Read less <
Origin
Hal imported