#satisfiability — Public Fediverse posts

“#NeuralNetworks and the #Satisfiability Problem” is the 2019 Stanford PhD dissertation by Salsam. It describes NeuroSAT, a #GNN, that learns to solve propositional satisfiability #SAT, that simple, yet quintessentially NP-complete, problem.

Salsam is an active member of the #Lean theorem prover community, who had worked closely with de Moura.

https://stacks.stanford.edu/file/druid:jt562cf4590/dselsam_dissertation_final-augmented.pdf

“#NeuralNetworks and the #Satisfiability Problem” is the 2019 Stanford PhD dissertation by Salsam. It describes NeuroSAT, a #GNN, that learns to solve propositional satisfiability #SAT, that simple, yet quintessentially NP-complete, problem.

Salsam is an active member of the #Lean theorem prover community, who had worked closely with de Moura.

https://stacks.stanford.edu/file/druid:jt562cf4590/dselsam_dissertation_final-augmented.pdf

'Critically Assessing the State of the Art in Neural Network Verification', by Matthias König, Annelot W. Bosman, Holger H. Hoos, Jan N. van Rijn.

http://jmlr.org/papers/v25/23-0119.html

#robustness #benchmarks #satisfiability

'Critically Assessing the State of the Art in Neural Network Verification', by Matthias König, Annelot W. Bosman, Holger H. Hoos, Jan N. van Rijn.

http://jmlr.org/papers/v25/23-0119.html

#robustness #benchmarks #satisfiability

Weekend project: try to solve some #combinatorics #enumeration problems by reduction to #SharpSAT. (Which, to be clear, I thought was unlikely to succeed!)

I picked c2d http://reasoning.cs.ucla.edu/c2d/ because it scored highly in the 2020 Model Counting Competition https://arxiv.org/abs/2012.01323 but I am not sure this is the same version. The one I got is dated 2005 and was 32-bit only. It ran out of memory on this 364-variable 942-clause instance (corresponding to 6 playing cards chosen from a standard 52-card deck.)

Looking at the 2023 competition instead, I think I should try SharpSAT-TD https://github.com/Laakeri/sharpsat-td but it is not as well documented. For example, I don't know if it supports the "eclauses" (exactly-one clauses) extension of the Dimacs CNF format.

#Satisfiability

Weekend project: try to solve some #combinatorics #enumeration problems by reduction to #SharpSAT. (Which, to be clear, I thought was unlikely to succeed!)

I picked c2d http://reasoning.cs.ucla.edu/c2d/ because it scored highly in the 2020 Model Counting Competition https://arxiv.org/abs/2012.01323 but I am not sure this is the same version. The one I got is dated 2005 and was 32-bit only. It ran out of memory on this 364-variable 942-clause instance (corresponding to 6 playing cards chosen from a standard 52-card deck.)

Looking at the 2023 competition instead, I think I should try SharpSAT-TD https://github.com/Laakeri/sharpsat-td but it is not as well documented. For example, I don't know if it supports the "eclauses" (exactly-one clauses) extension of the Dimacs CNF format.

#Satisfiability

Breakfast? check ✔️
Coffee? check ✔️
Submission to SAT fast track? DO IT NOW! (or forever wish you had)

https://cp2023.a4cp.org/cfp.html

#ConstraintProgramming #CallForPapers #ArtificialIntelligence #CP2023 #AcademicMastodon #AcademicChatter #Satisfiability

There is still time to submit the abstract of your paper in the SAT fast track. Check out the Call for Papers for more details: https://cp2023.a4cp.org/cfp.html

#ConstraintProgramming #CP2023 #CallForPapers #deadline #Satisfiability #AcademicMastodon #AcademicChatter #ArtificialIntelligence

Papers that were sadly not accepted to SAT 2023, can be revised and submitted to CP 2023 in a special fast track:

- SAT fast track abstract registration: May 1, 2023
- SAT fast track paper submission: May 17, 2023

#Satisfiability #AI #CP2023 #CfP #ConstraintProgramming #AcademicMastodon #CFP #CallForPapers