Enabling Research through the SCIP Optimization Suite 8.0

Bestuzheva, Ksenia; Besançon, Mathieu; Chen, Weikun; Chmiela, Antonia; Donkiewicz, Tim; Doornmalen, Jasper van; Eifler, Leon; Gaul, Oliver; Gamrath, Gerald; Gleixner, Ambros; Gottwald, Leona; Graczyk, Christoph; Halbig, Katrin; Hoen, Alexander; Hojny, Christopher; Hulst, Rolf van der; Koch, Thorsten; Lübbecke, Marco E.; Maher, Stephen J.; Matter, Frederic; Mühmer, Erik; Müller, Benjamin; Pfetsch, Marc E.; Rehfeldt, Daniel; Schlein, Steffan; SchlÃŰsser, Franziska; Serrano, Felipe; Shinano, Yuji; Sofranac, Boro; Turner, M.; Vigerske, Stefan; Wegscheider, Fabian; Wellner, Philipp; Weninger, Dieter; Witzig, Jakob

doi:10.1145/3585516

Cited by 53 publications

(127 citation statements)

References 60 publications

Supporting

Mentioning

126

Contrasting

Order By: Relevance

“…The evaluation machine has two Intel(R) Xeon(R) Gold 5117 CPUs @ 2.00GHz, 256GB ram and two Nvidia V100 GPUs. SCIP 8.0.1 Bestuzheva et al (2021), Gurobi 9.5.2 Gurobi Optimization, LLC (2022 and PyTorch 1.10.2 Paszke et al (2019) are utilized in our experiments. The emphasis for Gurobi and SCIP is set to focus on finding better primal solutions.…”

Section: Evaluation Metricsmentioning

confidence: 99%

“…Hence, Neural Network Verification (denoted as NNV) is chosen as the benchmark dataset for the comparison study. It is noteworthy that, empirically, turning on the presolve option in SCIP (Bestuzheva et al, 2021) causes false assertion of feasibility on many NNV instances. Hence, in our experiments on the NNV dataset, the presolve option is turned off, which potentially hurts the performances of both SCIP itself and frameworks implemented with SCIP.…”

Section: Comparing Against Neural Divingmentioning

confidence: 99%

See 1 more Smart Citation

A GNN-Guided Predict-and-Search Framework for Mixed-Integer Linear Programming

Han¹,

Yang²,

Chen³

et al. 2023

Preprint

View full text Add to dashboard Cite

Mixed-integer linear programming (MILP) is widely employed for modeling combinatorial optimization problems. In practice, similar MILP instances with only coefficient variations are routinely solved, and machine learning (ML) algorithms are capable of capturing common patterns across these MILP instances. In this work, we combine ML with optimization and propose a novel predict-and-search framework for efficiently identifying high-quality feasible solutions. Specifically, we first utilize graph neural networks to predict the marginal probability of each variable, and then search for the best feasible solution within a properly defined ball around the predicted solution. We conduct extensive experiments on public datasets, and computational results demonstrate that our proposed framework achieves 51.1% and 9.9% performance improvements to MILP solvers SCIP and Gurobi on primal gaps, respectively.

show abstract

Section: Evaluation Metricsmentioning

confidence: 99%

Section: Comparing Against Neural Divingmentioning

confidence: 99%

A GNN-Guided Predict-and-Search Framework for Mixed-Integer Linear Programming

Han¹,

Yang²,

Chen³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…LNS proceeds to iteration t + 1 with x t+1 until no improving solution x t+1 could be found by the LB ILP within a runtime limit. In experiments, the LB ILP is solved with SCIP 8.0.1 (Bestuzheva et al, 2021) with an hour runtime limit and k t is fine-tuned for each type of instances. After each solve of the LB ILP, in addition to the best solution found, SCIP records all intermediate solutions found during the solve.…”

Section: Data Collectionmentioning

confidence: 99%

“…All experiments use the hyperparameters described below unless stated otherwise. We use SCIP (v8.0.1) (Bestuzheva et al, 2021) to solve the sub-ILP in every iteration of LNS. To run LNS, we find an initial solution by running SCIP for 10 seconds.…”

Section: Setupmentioning

confidence: 99%

“…ILPs can be solved by Branch and Bound (BnB) (Land & Doig, 2010), an optimal tree search algorithm that can achieve state-of-the-art for ILPs. Over the past decades, BnB has been improved tremendously to become the core of many popular ILP solvers such as SCIP (Bestuzheva et al, 2021), CPLEX (Cplex, 2009) and Gurobi (Gurobi Optimization, LLC, 2022). However, due to its exhaustive search nature, it is hard for BnB to scale to large instances (Khalil et al, 2016;Gasse et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Searching Large Neighborhoods for Integer Linear Programs with Contrastive Learning

Huang¹,

Ferber²,

Tian³

et al. 2023

Preprint

View full text Add to dashboard Cite

Integer Linear Programs (ILPs) are powerful tools for modeling and solving a large number of combinatorial optimization problems. Recently, it has been shown that Large Neighborhood Search (LNS), as a heuristic algorithm, can find high quality solutions to ILPs faster than Branch and Bound. However, how to find the right heuristics to maximize the performance of LNS remains an open problem. In this paper, we propose a novel approach, CL-LNS, that delivers state-of-the-art anytime performance on several ILP benchmarks measured by metrics including the primal gap, the primal integral, survival rates and the best performing rate. Specifically, CL-LNS collects positive and negative solution samples from an expert heuristic that is slow to compute and learns a more efficient one with contrastive learning. We use graph attention networks and a richer set of features to further improve its performance.

show abstract