Objective Feasibility Pump 2.0.
The fundamental idea of the Feasibility Pump is to construct two sequences of points which hopefully converge to a feasible solution. One sequence consists of LP-feasiblepoints, the other one of integer feasible points. They are produced by alternately rounding an LP-feasible point and solvng an LP that finds a point on the LP polyhedron which is closest to the rounded, integral point (w.r.t. Manhattan distance).
The version implemented in SCIP supports using an Objective Feasibility Pump that uses a convex combination of the Manhattan distance and the original LP objective for reoptimization. It further features Feasibility Pump 2.0 capabilities, hence propagating the fixings for a faster convergence.
Definition in file heur_feaspump.h.
Go to the source code of this file.
|SCIP_RETCODE||SCIPincludeHeurFeaspump (SCIP *scip)|