Automatic Reformulation Search (ARS)

This 36-months project, starting 2007, is funded by the ANR.

Official partners

Other scientific partners

Abstract

Optimization problems are usually defined in terms of their mathematical programming formulation. This consists of a set of objective functions to be optimized subject to a set of constraints, all expressed in terms of a set of decision variables which may be discrete or continuous. Formulations may be symbolically transformed so that some of their numerical properties (e.g. optimal solutions, feasible region,...) are invariant. Yet sometimes the reformulated problem is easier to solve or is useful within a given solution algorithm. The main aims of this project are a systematic study of reformulation theory, the production of software tools for automating mathematical programming reformulation, and the formalization and implementation of solution algorithms based on reformulation techniques.

Temporary personnel

Sonia Cafieri, 2-year contract as a postdoctoral fellow at LIX, Ecole Polytechnique

Outcomes

International journal papers

Book chapters

International conference papers

Software

Reformulation/Optimization Software Engine (ROSE). Maintainer: Dr. David Savourey