Filters








1,290 Hits in 5.4 sec

Characterization of Glue Variables in CDCL SAT Solving [article]

Md Solimul Chowdhury and Martin Müller and Jia-Huai You
2019 arXiv   pre-print
We first show experimentally, by running the state-of-the-art CDCL SAT solver MapleLCMDist on benchmarks from SAT Competition-2017 and 2018, that branching decisions with glue variables are categorically  ...  The learned clauses with LBD score of 2, called glue clauses, are known to possess high pruning power which are never deleted from the clause databases of the modern CDCL SAT solvers.  ...  We also show that prioritizing glue variable selection improves the performance of CDCL SAT solvers MapleLCMDist and MapleL-CMDistChronoBT.  ... 
arXiv:1904.11106v1 fatcat:i6acswlvhjfrnli6txolbyaagi

Guiding CDCL SAT Search via Random Exploration amid Conflict Depression

Md Solimul Chowdhury, Martin Müller, Jia You
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
The efficiency of Conflict Driven Clause Learning (CDCL) SAT solving depends crucially on finding conflicts at a fast rate.  ...  Based on this analysis, we propose an exploration strategy, called expSAT, which randomly samples variable selection sequences in order to learn an updated heuristic from the generated conflicts.  ...  Acknowledgements We thank the anonymous reviewers for their valuable advice.  ... 
doi:10.1609/aaai.v34i02.5500 fatcat:sttjmffzzrfn3c2q37tlb3weyi

A branching heuristic for SAT solvers based on complete implication graphs

Fan Xiao, Chu-Min Li, Mao Luo, Felip Manyà, Zhipeng Lü, Yu Li
2019 Science China Information Sciences  
A branching heuristic for SAT solvers based on complete implication graphs. Sci China Inf Sci, 2019, 62(7): 072103, https://doi. The SAT problem is a classical NP-complete problem.  ...  State-of-the-art branching heuristics such as variable state independent decaying sum (VSIDS) [7] and learning rate branching (LRB) [8] maintain a score for each variable during the search process by aggregating  ...  Existing branching heuristics in CDCL SAT solvers Intensive efforts have been devoted to devising clever branching heuristics for SAT solvers [8, [13] [14] [15] [16] .  ... 
doi:10.1007/s11432-017-9467-7 fatcat:x2njcvo53rcclmde2lhhccw3tu

An Empirical Study of Branching Heuristics through the Lens of Global Learning Rate

Jia Liang, Hari Govind, Pascal Poupart, Krzysztof Czarnecki, Vijay Ganesh
2018 Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence  
as the global learning rate (GLR).  ...  We propose GLR as a metric for the branching heuristic to optimize. We test our hypothesis by developing a new branching heuristic that maximizes GLR greedily.  ...  Acknowledgement We thank Sharon Devasia Isac and Nisha Mariam Johnson from the College Of Engineering, Thiruvananthapuram, for their help in implementing the Berkmin and DLIS branching heuristics.  ... 
doi:10.24963/ijcai.2018/745 dblp:conf/ijcai/LiangKPCG18 fatcat:nklfvycndjcfbfxpnryxoxufhy

A Propagation Rate Based Splitting Heuristic for Divide-and-Conquer Solvers [chapter]

Saeed Nejati, Zack Newsham, Joseph Scott, Jia Hui Liang, Catherine Gebotys, Pascal Poupart, Vijay Ganesh
2017 Lecture Notes in Computer Science  
In this paper, we present a divide-and-conquer SAT solver, MapleAmpharos, that uses a novel propagation-rate (PR) based splitting heuristic.  ...  The key idea is that we rank variables based on the ratio of how many propagations they cause during the run of the worker conflictdriven clause-learning solvers to the number of times they are branched  ...  Conclusion We present a propagation rate-based splitting heuristic for divide-and-conquer solvers, implemented on top of the AMPHAROS solver, that is competitive with respect to 5 other parallel SAT solvers  ... 
doi:10.1007/978-3-319-66263-3_16 fatcat:ah3q5tnbqjatxpw2x7tvndfifi

Preliminary Results on Exploration-Driven Satisfiability Solving

Md Solimul Chowdhury, Martin Müller, Jia-Huai You
2018 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
Experiments with application benchmarks from recent SAT competitions demonstrate the potential of the expSAT approach for improving CDCL SAT solvers.  ...  Our proposed CDCL SAT solving algorithm expSAT uses a novel branching heuristic expVSIDS. It combines the standard VSIDS scores with heuristic scores derived from exploration.  ...  Complete SAT solvers based on the DPLL framework employ heuristics-guided state space search.  ... 
doi:10.1609/aaai.v32i1.12164 fatcat:r62vrbvom5hejit2nr2s4npsa4

A New Rewarding Mechanism for Branching Heuristic in SAT Solvers

Wenjing Chang, Yang Xu, Shuwei Chen
2019 International Journal of Computational Intelligence Systems  
features of state-of-the-art conflict-driven clause-learning SAT solvers.  ...  ID:p0075 Satisfiability problem Conflict-driven clause learning Branching heuristic Literal block distance A B S T R A C T ID:p0070 Decision heuristic strategy can be viewed as one of the most central  ...  ACKNOWLEDGMENTS ID:p0420 This work is supported by the National Natural Science Foundation of China (Grant No. 61673320), and by the Fundamental Research Funds for the Central Universities (Grant  ... 
doi:10.2991/ijcis.2019.125905649 fatcat:ffgfsb3pqndatgbs5f7eg5eoem

Enhancing SAT solvers with glue variable predictions [article]

Jesse Michael Han
2020 arXiv   pre-print
NeuroCore, proposed by Selsam and Bjorner, offered a proof-of-concept that neural networks can still accelerate SAT solvers by only periodically refocusing a score-based branching heuristic.  ...  Modern SAT solvers routinely operate at scales that make it impractical to query a neural network for every branching decision.  ...  We also thank Volodymyr Skladanivskyy for assistance with CGEN.  ... 
arXiv:2007.02559v1 fatcat:etqv3jkqwbcyvdc4lywybj5tyu

NeuroComb: Improving SAT Solving with Graph Neural Networks [article]

Wenxi Wang, Yang Hu, Mohit Tiwari, Sarfraz Khurshid, Kenneth McMillan, Risto Miikkulainen
2022 arXiv   pre-print
Mainstream modern SAT solvers are based on the Conflict-Driven Clause Learning (CDCL) algorithm.  ...  Recent work aimed to enhance CDCL SAT solvers by improving their variable branching heuristics through predictions generated by Graph Neural Networks(GNNs).  ...  [22] proposed a branching heuristic called GQSAT trained with value-based reinforcement learning.  ... 
arXiv:2110.14053v3 fatcat:mn6id5fg2ra7zgdqvayejbovfm

Learning Branching Heuristics for Propositional Model Counting [article]

Pashootan Vaezipoor, Gil Lederman, Yuhuai Wu, Chris J. Maddison, Roger Grosse, Edward Lee, Sanjit A. Seshia, Fahiem Bacchus
2020 arXiv   pre-print
In this paper, we present Neuro#, an approach for learning branching heuristics for exact #SAT solvers via evolution strategies (ES) to reduce the number of branching steps the solver takes to solve an  ...  counting problem to be solved by #SAT solvers.  ...  More concretely, our work is similar to [45] , which used Reinforcement Learning (RL) and graph neural networks to learn branching heuristics of a local search-based SAT solver WalkSAT [34] .  ... 
arXiv:2007.03204v1 fatcat:uf562j3kozakrf3oxgvjsbtgqm

Better Decision Heuristics in CDCL through Local Search and Target Phases

Shaowei Cai, Xindi Zhang, Mathias Fleury, Armin Biere
2022 The Journal of Artificial Intelligence Research  
On practical applications, state-of-the-art SAT solvers dominantly use the conflict-driven clause learning (CDCL) paradigm.  ...  Finally, the conflict frequency of variables in local search can be exploited during variable selection in branching heuristics of CDCL.  ...  Learning-Rate Branching (LRB) (Liang, Ganesh, Poupart, & Czarnecki, 2016 ) frames branching as an optimization problem that picks a variable to maximize a metric called learning rate.  ... 
doi:10.1613/jair.1.13666 fatcat:e6amd57tf5g2jdhcygo6unz3sy

Can Q-Learning with Graph Networks Learn a Generalizable Branching Heuristic for a SAT Solver? [article]

Vitaly Kurin, Saad Godil, Shimon Whiteson, Bryan Catanzaro
2020 arXiv   pre-print
We present Graph-Q-SAT, a branching heuristic for a Boolean SAT solver trained with value-based reinforcement learning (RL) using Graph Neural Networks for function approximation.  ...  a generalizable branching heuristic for SAT search.  ...  We present Graph-Q-SAT, a branching heuristic in a Conflict Driven Clause Learning [40, 21, CDCL] SAT solver trained with value-based reinforcement learning (RL), based on deep Q-networks [30, DQN]  ... 
arXiv:1909.11830v2 fatcat:fo6ze5umwrew7g2y2ay3bw2gwa

Improving Implementation of SAT Competitions 2017–2019 Winners [chapter]

Stepan Kochemazov
2020 Lecture Notes in Computer Science  
These modifications mostly consist in employing a deterministic strategy for switching between branching heuristics and in augmentations of the treatment of Tier2 and Core clauses.  ...  In this study we propose small modifications of implementations of existing heuristics in several related SAT solvers.  ...  The author thanks anonymous reviewers for their helpful comments and Oleg Zaikin and Alexander Semenov for many fruitful discussions.  ... 
doi:10.1007/978-3-030-51825-7_11 fatcat:freyygac75hl7c24utrh34ldka

Perceptron Learning of SAT

Alex Flint, Matthew B. Blaschko
2012 Neural Information Processing Systems  
for that SAT subset based on the Davis-Putnam-Logemann-Loveland algorithm.  ...  Such SAT instances are likely to have shared characteristics and substructures. This work approaches the exploration of a family of SAT solvers as a learning problem.  ...  A number of authors have considered learning branching rules for SAT solvers. Ruml applied reinforcement learning to find valuations of satisfiable sentences [25] .  ... 
dblp:conf/nips/FlintB12 fatcat:lkyvjx2jjjc5dcrr3kxtosynfq

Guiding High-Performance SAT Solvers with Unsat-Core Predictions [article]

Daniel Selsam, Nikolaj Bjørner
2019 arXiv   pre-print
Our results demonstrate that NeuroSAT can provide effective guidance to high-performance SAT solvers on real problems.  ...  We modify several high-performance SAT solvers to periodically replace their variable activity scores with NeuroSAT's prediction of how likely the variables are to appear in an unsatisfiable core.  ...  Heule for helpful discussions.  ... 
arXiv:1903.04671v7 fatcat:a7znzck2mne5bposnxt3hgtiim
« Previous Showing results 1 — 15 out of 1,290 results