JAIST Repository >
School of Information Science >
Grants-in-aid for Scientific Research Papers >
FY 2014 >

Please use this identifier to cite or link to this item: http://hdl.handle.net/10119/12815

Title: 近似手法と数式処理の融合による実数多項式制約の効率化
Other Titles: Optimization of polynomial constraint solving based on fusion of approximation and algebraix methods
Authors: 小川, 瑞史
Authors(alternative): Mizuhito, Ogawa
Keywords: ソフトウェア
Issue Date: 10-Jun-2015
Abstract: 本研究では、多項式制約解消アルゴリズムである区間制約伝播の拡張としてraSAT ループを提案し、SMTソルバ raSATとして実装を行った。制約解消の解は SAT(充足可能)かUNSAT(充足不能)だが、SAT検出はエラー検出、UNSAT検出はループ不変式生成に用いられる。数百変数程度の大規模な問題に対し、実質的に後者が解けるのはUNSATコア(小さな部分制約で充足不能となるもの)を発見した場合に限られ、前者はすべての制約の充足性の確認が必要なため、実用上より困難である。本研究ではSAT検出に焦点をあて、SMTlibベンチマーク上の実験で答えが未知であった複数の問題のSATを検出した。 : ):This research extends the interval constraint propagation to the raSAT loop, which is implemented as an SMT solver raSAT. A polynomial constraint solving concludes either SAT (satisfiable) or UNSAT (unsatisfiable). raSAT is designed to intend the former, which is often used for error detection, where the latter is used for loop invariant generation. In practice, when tackling a larger problem with several hundred variables, UNSAT is detected only with the discovery of a small UNSAT core, whereas SAT is detected by finding an instance that satisfies all constraints, and often becomes a harder problem. In experiments, we newly found that several unsolved problems in SMTlib benchmark are SAT.
Description: 研究種目:基盤研究(B)
Language: jpn
URI: http://hdl.handle.net/10119/12815
Appears in Collections:2014年度 (FY 2014)

Files in This Item:

File Description SizeFormat
23300005seika.pdf100KbAdobe PDFView/Open

All items in DSpace are protected by copyright, with all rights reserved.


Contact : Library Information Section, Japan Advanced Institute of Science and Technology