BandMaxSAT: A Local Search MaxSAT Solver with Multi-armed Bandit

Authors: Jiongzhi Zheng, Kun He, Jianrong Zhou, Yan Jin, Chu-Min Li, Felip Manyà

IJCAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments demonstrate that Band Max SAT significantly outperforms the stateof-the-art (W)PMS local search algorithm SATLike3.0. Specifically, the number of instances in which Band Max SAT obtains better results is about twice that obtained by SATLike3.0. Moreover, we combine Band Max SAT with the complete solver TT-Open-WBO-Inc. The resulting solver Band Max SAT-c also outperforms some of the best state-of-the-art complete (W)PMS solvers, including SATLike-c, Loandra and TT-Open-WBO-Inc.
Researcher Affiliation Academia Jiongzhi Zheng1,2 , Kun He 1 , Jianrong Zhou1 , Yan Jin1 , Chu-Min Li3 , Felip Manya4 1School of Computer Science and Technology, Huazhong University of Science and Technology 2Institute of Artificial Intelligence, Huazhong University of Science and Technology, China 3MIS, University of Picardie Jules Verne, France 4Artificial Intelligence Research Institute (IIIA), CSIC, Bellaterra, Spain
Pseudocode Yes Algorithm 1: Hy Deci(F) ... Algorithm 2: Pick Arm(Arm Num, N, λ) ... Algorithm 3: Band Max SAT
Open Source Code Yes The code of Band Max SAT is available at https://github.com/JHLHUST/Band Max SAT/.
Open Datasets Yes We tested the algorithms on all the (W)PMS instances from the incomplete track of last four Max SAT Evaluations (MSE), i.e., MSE2018-MSE2021. ... We use all the (W)PMS instances from the incomplete track of MSE2017 as the training set to tune these parameters.
Dataset Splits Yes We use all the (W)PMS instances from the incomplete track of MSE2017 as the training set to tune these parameters.
Hardware Specification Yes Our experiments were performed on a server using an Intel Xeon E5-2650 v3 2.30 GHz 10-core CPU and 256 GB RAM, running Ubuntu 16.04 Linux operation system.
Software Dependencies No Band Max SAT is implemented in C++ and compiled by g++. No specific version numbers for g++ or any other software dependencies are provided.
Experiment Setup Yes The parameters in Band Max SAT include the BMS parameter k, the reward delay steps d, the reward discount factor γ, the number of sampled arms Arm Num, and the exploration bias parameter λ. ... Finally, the default settings of these parameters are as follows: k = 15, d = 20, γ = 0.9, Arm Num = 20, λ = 1.