| 本文已被：浏览 6581次 下载 7837次 |
码上扫一扫！ |
|
|
| 求解QBF 问题的启发式调查传播算法 |
|
殷明浩1,2, 周俊萍1,3, 孙吉贵3,2, 谷文祥1
|
|
1.东北师范大学 计算机学院,吉林 长春 130117;2.吉林大学 教育部符号计算与知识工程重点实验室,吉林 长春 130012;3.吉林大学 计算机科学与技术学院,吉林 长春 130012
|
|
| 摘要: |
| 提出了一种启发式调查传播算法,并基于该算法设计了一种QBF(quantified Boolean formulae)求解器——HSPQBF(heuristic survey propagation algorithm for solving QBF)系统.它将Survey Propagation 信息传递方法应QBF 求解问题中.利用Survey Propagation 作为启发式引导DPLL(Davis,Putnam,Logemann and Loveland)算法,合适的变量进行分支,从而可以减小搜索空间,并减少算法回退的次数.在分支处理过程中,HSPQBF 系统结合元传播、冲突学习和满足蕴涵学习等一些优秀的QBF 求解技术,从而能够提高QBF 问题的求解效率.实验结明,HSPQBF 无论在随机问题上还是在QBF 标准测试问题上都有很好的表现,验证了调查传播技术在QBF 问解中的实际价值. |
| 关键词: 人工智能 QBF 问题 QBF 问题求解器 因子图 调查传播 冲突学习 满足蕴涵学习 |
| DOI：10.3724/SP.J.1001.2011.03859 |
| 分类号: |
| 基金项目:国家自然科学基金(60773097, 60803102) |
|
| Heuristic Survey Propagation Algorithm for Solving QBF Problem |
|
YIN Ming-Hao1,2, ZHOU Jun-Ping1,3, SUN Ji-Gui3,2, GU Wen-Xiang1
|
|
1.College of Computer Science, Northeast Normal University, Changchun 130117, China;2.Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Changchun 130012, China;3.College of Computer Science and Technology, Jilin University, Changchun 130012, China
|
| Abstract: |
| This paper presents a heuristic survey propagation algorithm for solving Quantified Boolean Formulae
(QBF) problem. A QBF solver based on the algorithm is designed, namely HSPQBF (heuristic survey propagation
algorithm for solving QBF). This solver is a QBF reasoning engine that incorporates Survey Propagation method for
problem solving. Using the information obtained from the survey propagation procedure, HSPQBF can select a
branch accurately. Furthermore, when handling the branches, HSPQBF uses efficient technology to solve QBF
problems, such as unit propagation, conflict driven learning, and satisfiability directed at implication and learning.
The experimental results also show that HSPQBF can solve both random and QBF benchmark problems efficiently,
which validates the effect of using survey propagation in a QBF solving process. |
| Key words: artificial intelligence quantified Boolean formulae problem QBF(quantified Boolean formulae)
solver factor graph survey propagation conflict driven learning satisfiability directed implication
and learning |
