A Pre-processing Aware RINS Based MIP Heuristic

摘要

This paper proposes an adaptation of the RINS MIP heuristic which explicitly explores pre-processing techniques. The method systematically searches for the ideal number of fixations to produce sub-problems of controlled size. These problems are explored in a Variable Neighborhood Descent fashion until a stopping criterion is met. Preliminary experiments implemented upon the open source MIP solver COIN-OR CBC are presented.