ParaSCIP: A Parallel Extension of SCIP

摘要: Mixed integer programming (MIP)has become one of the most important techniques in Operations Research and Discrete Optimization. SCIP (Solving Constraint Integer Programs) is currently fastest non-commercial MIP solvers. It based on branchandboundprocedure which problem recursively split into smaller subproblems, thereby creating a so-called branching tree. We present ParaSCIP, an extension SCIP, realizes parallelization distributed memory computing environment. ParaSCIP uses solvers as independently running processes to solve subproblems (nodes tree) locally. This makes development independent development. Thus, directly profits from any algorithmic progress future versions SCIP. Using first implementation we were able two previously unsolved instances MIPLIB2003, standard test set library for For these computations, used up 2048 cores HLRN II supercomputer.