Kissat SAT Solver

栏目: IT技术 · 发布时间: 4年前

内容简介:The Kissat SAT solver is a condensed and improved reimplementation ofCaDiCaL in C.It has improved data structures, better scheduling of inprocessing, optimized algorithms and implementation.The solver is not incremental yet, but can be used as a library.

The Kissat SAT solver is a condensed and improved reimplementation ofCaDiCaL in C.

It has improved data structures, better scheduling of inprocessing, optimized algorithms and implementation.

The solver is not incremental yet, but can be used as a library.

News

Kissat won first place in the main track of the SAT Competition 2020 and first place on unsatisfiable instances.

SAT Competition Winners

In order to show that SAT solving improved considerably in recent years, particularly also with Kissat in 2020, we collectedwinners of the SAT Competition from 2002 to 2020 and fixed and ported them to be compatabile with modern compilers.

Then we were running those "all-time winners" on our cluster on the competition instances from 2011, 2019 and 2020, more precisely on the 300 instances from the SAT Competition 2011 application track , the 400 instances from the SAT Race 2019 and the 400 instances from SAT Competition 2020 main track .

Here are plots showing the number of instances solved within a certain time limit, thus in essence the runtime cumulative distribution function (higher is better).

Kissat SAT Solver Kissat SAT Solver Kissat SAT Solver

Note, the SAT Competition 2011 application track had only 300 instances, while the other two benchmarks sets from 2019 and 2020 consist of 400 apparently much harder instances each.

Our cluster has two Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz CPUs and 128 GB of main memory on each node, thus is slightly slower, but comparable to the StarExec cluster used in the competition. As time limit we used 5000 seconds as it is now common in the competition. Main memory was limited to 8 GB for 2011 and 2019 to utilize our cluster best. For 2020 we increased the limit to 32 GB, which allowed all solvers to stay below the memory limit (the organizers announced a 24 GB memory limit originally but actually used 128 GB).

Download

The latest version is available on GitHub at https://github.com/arminbiere/kissat

The following source code submitted to the SAT Competition 2020 won first place in the main track and first place on the unsatisfiable instances of the main track:

[ kissat-sc2020-039805f2.tar.xz ]

The code has an MIT license.

Resources

See the following SAT Competition 2020 system description for more details:

Armin Biere, Katalin Fazekas, Mathias Fleury, Maximillian Heisinger.

CaDiCaL, Kissat, Paracooba, Plingeling and Treengeling Entering the SAT Competition 2020 .

To appear in Proc. of SAT Competition 2020 - Solver and Benchmark Descriptions .

[paper |bibtex |kissat |cadical | paracooba |plingeling |treengeling ]

There was also a related talk at POS'20 :

Armin Biere and Mathias Fleury.

Chasing Target Phases

11th Workshop on Pragmatics of SAT (POS'20)

Cyberspace, July 3, 2020.

[slides [ video |experiments ]


以上所述就是小编给大家介绍的《Kissat SAT Solver》,希望对大家有所帮助,如果大家有任何疑问请给我留言,小编会及时回复大家的。在此也非常感谢大家对 码农网 的支持!

查看所有标签

猜你喜欢:

本站部分资源来源于网络,本站转载出于传递更多信息之目的,版权归原作者或者来源机构所有,如转载稿涉及版权问题,请联系我们

计算广告

计算广告

刘鹏、王超 / 人民邮电出版社 / 2015-9-1 / 69.00元

计算广告是一项新兴的研究课题,它涉及大规模搜索和文本分析、信息获取、统计模型、机器学习、分类、优化以及微观经济学等诸多领域的知识。本书从实践出发,系统地介绍计算广告的产品、问题、系统和算法,并且从工业界的视角对这一领域具体技术的深入剖析。 本书立足于广告市场的根本问题,从计算广告各个阶段所遇到的市场挑战出发,以广告系统业务形态的需求和变化为主线,依次介绍合约广告系统、竞价广告系统、程序化交易......一起来看看 《计算广告》 这本书的介绍吧!

JSON 在线解析
JSON 在线解析

在线 JSON 格式化工具

XML、JSON 在线转换
XML、JSON 在线转换

在线XML、JSON转换工具

Markdown 在线编辑器
Markdown 在线编辑器

Markdown 在线编辑器