Hybrid Algorithm for Here-and-Now Stochastic Network Design Problem with Pricing

Hybrid Algorithm for Here-and-Now Stochastic Network Design Problem with Pricing

Abstrakt

The purpose of the paper is to discuss an advanced hybrid algorithm for the solution of a hereand-now scenario-based expected objective reformulation of an underlying stochastic program dealing with transportation problem involving random demand parameters and 0-1 network design variables. At the beginning, the deterministic transportation model with network design variables is reviewed. Then, uncertain demand parameters are introduced and modeled by random variables. The following deterministic reformulation is based on the here-and-now (HN) approach. The model is enhanced with implemetation of pricing. A finite discrete probability distribution is assumed for a random vector involving all stochastic parameters. Then, the obtained nonlinear binary program can be solved by GAMS solvers. However, the large problems represent true challenge, so the authors combine a classical optimization algorithm and a suitable genetic algorithm to obtain a hybrid algorithm that is modified for the HN case. The implementation and test results illustrated by figures are included.

The purpose of the paper is to discuss an advanced hybrid algorithm for the solution of a hereand-now scenario-based expected objective reformulation of an underlying stochastic program dealing with transportation problem involving random demand parameters and 0-1 network design variables. At the beginning, the deterministic transportation model with network design variables is reviewed. Then, uncertain demand parameters are introduced and modeled by random variables. The following deterministic reformulation is based on the here-and-now (HN) approach. The model is enhanced with implemetation of pricing. A finite discrete probability distribution is assumed for a random vector involving all stochastic parameters. Then, the obtained nonlinear binary program can be solved by GAMS solvers. However, the large problems represent true challenge, so the authors combine a classical optimization algorithm and a suitable genetic algorithm to obtain a hybrid algorithm that is modified for the HN case. The implementation and test results illustrated by figures are included.

Hybrid algorithm, stochastic network design, pricing

Hybrid algorithm, stochastic network design, pricing

HRABEC, D.; POPELA, P.; ROUPEC, J.; HAUGEN, K.

01.09.2014

NEUVEDENO