Paula Amaral

Citation:

Amaral, P., and P. Barahona. "After infeasibility in linear programming." Proceedings of CP-AI-OR99 workshop on integration of AI and OR techniques in Constraint Programming for Combinatorial Optimization problems. Vol. 1. Universit, 1999.

Abstract:

This work is focused on the correction of Infeasible Linear problems. Its motivation is not difficult to understand if one thinks on the complexity of model building in large real problems. The inconsistency can arise in the definition of the model due, for instance, to structural or data type errors. The identification of conflict sets of constraints is very useful but might not be enough to overcome the problem, since the implementation of a solution may require the definition of a new feasible model. We present a short review on known procedures for the diagnosis of these problems. The approach we propose is based on the correction of (potentially) all the parameters of the model restrictions. We present a pure algebraic methodology based on the Singular Value Decomposition of a matrix. This method is quite rigid in the changes of the matrix coefficients changes, so we give insights on a heuristic based approach in order to attain more flexibility.

Notes:

n/a