Improving Branch-and-Price Algorithms and Applying Them to Stochastic Programs

Abstract

The first phase of this research demonstrates improvements in the performance of branch-and-price algorithms (B&P) for solving integer programs by (i) stabilizing dual variables during column generation, (ii) performing strong branching, (iii) inserting multiple near-optimal columns from each subproblem, (iv) heuristically improving feasible integer solutions, and by applying several other techniques. Computational testing on generalized-assignment problems shows that solution times decrease over na ve B&P by as much as 96%; and, some problems that could not be solved by standard branch and bound on the standard model formulation have become easy. In the second phase, this research shows how to solve a class of difficult, stochastic mixed-integer programs using B&P. A new, column-oriented formulation of a stochastic facility-location problem (SFLP), using a scenario representation of uncertainty, provides a vehicle for demonstrating this method s viability. Computational results show that B&P can be orders of magnitude faster than solving the original problem by branch and bound, and this can be true even for single-scenario problems; i.e., for deterministic problems. B&P also solves SFLP exactly when random parameters are modeled through certain continuous probability distributions. In practice, these problems solve more quickly than comparable scenario-based problems.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 2004
Accession Number
ADA427284

Entities

People

  • Eduardo F. Silva

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Computational Science
  • Computer Programming
  • Computer Programs
  • Computers
  • Electronic Mail
  • Integer Programming
  • Linear Programming
  • Mathematical Models
  • Mathematical Programming
  • Operating Systems
  • Operations Research
  • Optimization
  • Probability
  • Probability Distributions
  • Random Variables
  • Systems Engineering

Readers

  • Operations Research