The SCIP Optimization Suite 8.0

Abstract

The SCIP Optimization Suite provides a collection of software packages for mathematical optimization centered around the constraint integer programming framework SCIP. This paper discusses enhancements and extensions contained in version 8.0 of the SCIP Optimization Suite. Major updates in SCIP include improvements in symmetry handling and decomposition algorithms, new cutting planes, a new plugin type for cut selection, and a complete rework of the way nonlinear constraints are handled. Additionally, SCIP 8.0 now supports interfaces for Julia as well as Matlab. Further, UG now includes a unified framework to parallelize all solvers, a utility to analyze computational experiments has been added to GCG, dual solutions can be postsolved by PaPILO, new heuristics and presolving methods were added to SCIP-SDP, and additional problem classes and major performance improvements are available in SCIP-Jack. Keywords Constraint integer programming • linear programming • mixed-integer linear programming • mixed-integer nonlinear programming • optimization solver • branch-andcut • branch-and-price • column generation • parallelization • mixed-integer semidefinite programming Mathematics Subject Classification 90C05 • 90C10 • 90C11 • 90C30 • 90C90 • 65Y05 * Extended author information is available at the end of the paper.

Assessment of a two-step approach for global optimization of mixed-integer polynomial programs using quadratic reformulation

“…The main workflow in CANON entails: (i) reading the MIPOP model, for instance in GAMS format; (ii) preprocessing the MIPOP by applying local search and bound-tightening techniques; (iii) reformulating the MIPOP as an equivalent MIQCP, possibly with redundant constraints; and (iv) writing the reformulated MIQCP to file in GAMS format. In this way, the reformulated MIPOP can be solved with any dedicated MIQCP solver such as GUROBI (Gurobi Optimization LLC, 2022), or with any generic global MINLP solver such as BARON (Kılınç and Sahinidis, 2018) and SCIP (Bestuzheva et al, 2021).…”

“…State-of-the-art global solvers based on spatial branch-and-bound (sBB), such as BARON (Tawarmalani and Sahinidis, 2005;Kılınç and Sahinidis, 2018), ANTIGONE (Misener and Floudas, 2014) and SCIP (Bestuzheva et al, 2021), can handle MIPOPs as special cases. As well as relying on mixed-integer linear (MIP) relaxations constructed from factorable representations and term-wise relaxations (McCormick, 1976), these solvers implement a range of strategies tailored to bilinear, multilinear or polynomial subexpressions.…”

Long-Term Planning for a Mixed Urban Freight Fleet with EVs and ICEVs in the USA

“…The optimization problem is written in Python using the Google OR Tools library and is solved using a Google OR Tools solver, called Solving Constraint Integer Programs (SCIP) [43]. This solver was chosen because it is one of the fastest non-commercial solvers for linear programming (LP) and MIP, and provides a constraint programming framework.…”

The Binary Linearization Complexity of Pseudo-Boolean Functions

“…Hence, we compare the new IP model (9) to the standard formulation (8). We tried to solve the low auto-correlation binary sequences problem with both models using the SCIP solver framework [2] 1 . Note that both approaches are too slow to compete with the most recent stateof-the-art approach that is based on combinatorial branch-and-bound with which the problem could be solved to optimality up to N = 66 [18].…”