Oliver kullmann theory and applications of satis

Oliver kullmann theory and applications of satis tài liệu, giáo án, bài giảng , luận văn, luận án, đồ án, bài tập lớn về...

Lecture Notes in Computer Science 5584 Commenced Publication in 1973 Founding and Former Series Editors: Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen Editorial Board David Hutchison Lancaster University, UK Takeo Kanade Carnegie Mellon University, Pittsburgh, PA, USA Josef Kittler University of Surrey, Guildford, UK Jon M. Kleinberg Cornell University, Ithaca, NY, USA Alfred Kobsa University of California, Irvine, CA, USA Friedemann Mattern ETH Zurich, Switzerland John C. Mitchell Stanford University, CA, USA Moni Naor Weizmann Institute of Science, Rehovot, Israel Oscar Nierstrasz University of Bern, Switzerland C. Pandu Rangan Indian Institute of Technology, Madras, India Bernhard Steffen University of Dortmund, Germany Madhu Sudan Massachusetts Institute of Technology, MA, USA Demetri Terzopoulos University of California, Los Angeles, CA, USA Doug Tygar University of California, Berkeley, CA, USA Gerhard Weikum Max-Planck Institute of Computer Science, Saarbruecken, Germany Oliver Kullmann (Ed.) Theory and Applications of Satisfiability Testing – SAT 2009 12th International Conference, SAT 2009 Swansea, UK, June 30 - July 3, 2009 Proceedings 13 Volume Editor Oliver Kullmann Computer Science Department Swansea University Faraday Building, Singleton Park Swansea, SA2 8PP, UK E-mail: o.kullmann@swansea.ac.uk Library of Congress Control Number: Applied for CR Subject Classification (1998): F.4.1, I.2.3, I.2.8, I.2, F.2.2, G.1.6 LNCS Sublibrary: SL 1 – Theoretical Computer Science and General Issues ISSN 0302-9743 ISBN-10 3-642-02776-8 Springer Berlin Heidelberg New York ISBN-13 978-3-642-02776-5 Springer Berlin Heidelberg New York This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. springer.com © Springer-Verlag Berlin Heidelberg 2009 Printed in Germany Typesetting: Camera-ready by author, data conversion by Scientific Publishing Services, Chennai, India Printed on acid-free paper SPIN: 12712779 06/3180 543210 Preface This volume contains the papers presented at SAT 2009: 12th International Conference on Theory and Applications of Satisfiability Testing, held from June 30 to July 3, 2009 in Swansea (UK). The International Conference on Theory and Applications of Satisfiability Testing (SAT) started in 1996 as a series of workshops, and, in parallel with the growth of SAT, developed into the main event for SAT research. This year’s con- ference testified to the strong interest in SAT, regarding theoretical research, re- search on algorithms, investigations into applications, and development of solvers and software systems. As a core problem of computer science, SAT is central for many research areas, and has deep interactions with many mathematical sub- jects. Major impulses for the development of SAT came from concrete practical applications as well as from fundamental theoretical research. This fruitful col- laboration can be seen in virtually all papers of this volume. There were 86 submissions (completed papers within the scope of the con- ference). Each submission was reviewed by at least three, and on average 4.0 Programme Committee members. The Committee decided to accept 45 papers, consisting of 34 regular and 11 short papers (restricted to 6 pages). A main nov- elty was a “shepherding process”, where 29% of the papers were accepted only conditionally, and requirements on necessary improvements were formulated by the Programme Committee and its installment monitored by the “shepherd” for that paper (using possibly several rounds of feedback). This process helped enor- mously to improve the quality of the papers, and it also enabled the Programme Committee to accept 13 papers, which have very interesting contributions, but which due to weaknesses normally wouldn’t have made it into the proceedings. 27 regular and 5 short papers were accepted unconditionally, and 7 long and 7 = 3 + 4 short papers were accepted conditionally (with 4 required conversions from regular to short papers). All these 7 long papers and 6 of the 7 short papers could then be accepted in the “second round”, involving in all cases substantial work for the authors (often a complete revision) and the shepherd (ranging from providing general advice to complete grammatical overhauls). As one author put it: “I would, however, like to congratulate the reviewers, as their review is the most useful and thorough I have ever received from any conference - indeed, if integrated correctly, it brings a new level of quality to the paper.” The organisation of the papers is by subjects (and within the categories alphabetically). The programme included two invited talks: – Robert Niewenhuis considered how SMT (“SAT modulo theories”) can en- hance SAT solving in a systematic way by special algorithms, as it is possible in constraint programming. – Moshe Vardi investigated how the strong inference power delivered by OB- DDs (“ordered binary decision diagrams”) can be harnessed by SAT solving. VI Preface One of the major topics of this conference was the MAXSAT problem (max- imising the number of satisfied clauses), and boolean optimisation problems in general. Besides these extensions, the papers of this conference show that “core SAT”, that is, boolean CNF-SAT solving, has still a huge potential (I expect that we just scratched the surface, and fascinating discoveries are waiting for us). One fundamental topic was the understanding of why and when SAT solvers are efficient, and interesting approaches were considered, towards a more precise intelligent control of the execution of SAT solvers. Another strong area of this year was the intelligent translation of problems into SAT. Regarding QBF, the extension of SAT by allowing quantification, the quest for a “good” problem representation becomes even more urgent, and we find theoretical and practical approaches. Several additional events were associated with the SAT conference, including the SAT competition, the PB competition (“pseudo-boolean”, allowing certain forms of arithmetic), the Max-SAT evaluation, and a special session on the var- ious aspects of the process of developing SAT software. Arnold Beckmann and Matthew Gwynne helped with the local organisation. We gladly acknowledge the following people in organising the satellite events: – the main organisers of the SAT competition Daniel Le Berre, Olivier Roussel, Laurent Simon, the judges Andreas Goerdt, Inˆes Lynce and Aaron Stump, and the special organisers Allen Van Gelder, Armin Biere, Edmund Clarke, John Franco and Sean Weaver – the organisers of the PB competition Vasco Manquinho and Olivier Roussel; – and the organisers of the Max-SAT evaluation Josep Argelich, Chu Min Li, Felip Many`a and Jordi Planes A special thanks goes to the Programme Committee and the additional external reviewers, who through their thorough and knowledgeable work enabled the assembly of this body of high-quality work. We also thank the authors for their enthusiastic collaboration in further improving their papers. The EasyChair conference management system helped us with handling of the paper submissions, paper reviewing, paper discussion and assembly of the proceedings. I would like to thank the Chairs of the previous years, Hans Kleine B¨uning, Xishun Zhao and Joao Marques-Silva, for their important advice on run- ning a conference. The Department of Computer Science of Swansea University provided logistic support. Finally I would like to thank the following sponsors for their support of SAT 2009: Intel Corporation, NEC Laboratories, and Invensys Rail Group. 1 April 2009 Oliver Kullmann 1 Due to the difficult economic circumstances a number of former sponsors expressed their regret for not being able to provide funding this year. Conference Organisation Conference and Programme Chair Oliver Kullmann Computer Science Department, Swansea University, UK Local Organisation Arnold Beckmann Computer Science Department, Swansea University, UK Matthew Gwynne Computer Science Department, Swansea University, UK Programme Committee Dimitris Achlioptas Armin Biere Stephen Cook Nadia Creignou Evgeny Dantsin Adnan Darwiche John Franco Nicola Galesi Enrico Giunchiglia Ziyad Hanna Marijn Heule Edward Hirsch Kazuo Iwama Hans Kleine B¨uning Daniel LeBerre Chu Min Li Ines Lynce Panagiotis Manolios Joao Marques-Silva David Mitchell Albert Oliveras Ramamohan Paturi Lakhdar Sais Karem Sakallah Uwe Sch¨oning Roberto Sebastiani Robert Sloan Carsten Sinz Niklas S¨orensson Ewald Speckenmeyer Stefan Szeider Armando Tacchella Miroslaw Truszczynski Alasdair Urquhart Allen Van Gelder Hans van Maaren Toby Walsh Sean Weaver Emo Welzl Lintao Zhang Xishun Zhao External Reviewers Anbulagan Anbulagan Carlos Ans´otegui Josep Argelich Regis Barbanchon Maria Luisa Bonet Simone Bova Roberto Bruttomesso Uwe Bubeck Lorenzo Carlucci HarshRajuChamarthi Benjamin Chambers Hubie Chen Gilles Dequen Laure Devendeville Juan Luis Esteban Paulo Flores Anders Franzen Heidi Gebauer Eugene Goldberg Alexandra Goultiaeva Alberto Griggio Djamal Habet Shai Haim Miki Hermann VIII Organization Dmitry Itsykson George Katsirelos George Katsirelose Arist Kojevnikov Stephan Kottler Alexander Kulikov Javier Larrosa Silvio Lattanzi Massimo Lauria Jimmy Lee Theodor Lettmann Florian Lonsing Toni Mancini Vasco Manquinho Felip Many`a Marco Maratea Paolo Marin John Moondanos Robin Moser Massimo Narizzano Nina Naroditskaya Sergey Nikolenko Sergey Nurk Richard Ostrowski C´edric Piette Knot Pipatsrisawat Jordi Planes Stefan Porschen Luca Pulina Silvio Ranise Andreas Razen Alyson Reeves Olivier Roussel Emanuele Di Rosa Jabbour Said Dominik Scheder Thomas Schiex Tatjana Schmidt Henning Schnoor Yuping Shen Michael Soltys Stefano Tonetta Patrick Traxler Enrico Tronci Gyorgy Turan Olga Tveretina Alexander Wolpert Stefan Woltran Grigory Yaroslavtsev Weiya Yue Bruno Zanuttini Michele Zito Philipp Zumstein Sponsoring Institutions Computer Science Department, Swansea University Invensys Rail Group Intel Corporation NEC Laboratories Table of Contents 1. Invited Talks SAT Modulo Theories: Enhancing SAT with Special-Purpose Algorithms 1 Robert Nieuwenhuis Symbolic Techniques in Propositional Satisfiability Solving 2 Moshe Y. Vardi 2. Applications of SAT Efficiently Calculating Evolutionary Tree Measures Using SAT 4 Mar´ıa Luisa Bonet and Katherine St. John Finding Lean Induced Cycles in Binary Hypercubes 18 Yury Chebiryak, Thomas Wahl, Daniel Kroening, and Leopold Haller Finding Efficient Circuits Using SAT-Solvers 32 Arist Kojevnikov, Alexander S. Kulikov, and Grigory Yaroslavtsev Encoding Treewidth into SAT 45 Marko Samer and Helmut Veith 3. Complexity Theory The Complexity of Reasoning for Fragments of Default Logic 51 Olaf Beyersdorff, Arne Meier, Michael Thomas, and Heribert Vollmer Does Advice Help to Prove Propositional Tautologies? 65 Olaf Beyersdorff and Sebastian M¨uller 4. Structures for SAT Backdoors in the Context of Learning 73 Bistra Dilkina, Carla P. Gomes, and Ashish Sabharwal Solving SAT for CNF Formulas with a One-Sided Restriction on Variable Occurrences 80 Daniel Johannsen, Igor Razgon, and Magnus Wahlstr¨om On Some Aspects of Mixed Horn Formulas 86 Stefan Porschen, Tatjana Schmidt, and Ewald Speckenmeyer X Table of Contents Variable Influences in Conjunctive Normal Forms 101 Patrick Traxler 5. Resolution and SAT Clause-Learning Algorithms with Many Restarts and Bounded-Width Resolution 114 Albert Atserias, Johannes Klaus Fichte, and Marc Thurley An Exponential Lower Bound for Width-Restricted Clause Learning 128 Jan Johannsen Improved Conflict-Clause Minimization Leads to Improved Propositional Proof Traces 141 Allen Van Gelder Boundary Points and Resolution 147 Eugene Goldberg 6. Translations to CNF Sequential Encodings from Max-CSP into Partial Max-SAT 161 Josep Argelich, Alba Cabiscol, Inˆes Lynce, and Felip Many`a Cardinality Networks and Their Applications 167 Roberto As´ın, Robert Nieuwenhuis, Albert Oliveras, and Enric Rodr´ıguez-Carbonell New Encodings of Pseudo-Boolean Constraints into CNF 181 Olivier Bailleux, Yacine Boufkhad, and Olivier Roussel Efficient Term-ITE Conversion for Satisfiability Modulo Theories 195 Hyondeuk Kim, Fabio Somenzi, and HoonSang Jin 7. Techniques for Conflict-Driven SAT Solvers On-the-Fly Clause Improvement 209 Hyojung Han and Fabio Somenzi Dynamic Symmetry Breaking by Simulating Zykov Contraction 223 Bas Schaafsma, Marijn J.H. Heule, and Hans van Maaren Minimizing Learned Clauses 237 Niklas S¨orensson and Armin Biere Extending SAT Solvers to Cryptographic Problems 244 Mate Soos, Karsten Nohl, and Claude Castelluccia Table of Contents XI 8. Solving SAT by Local Search Improving Variable Selection Process in Stochastic Local Search for Propositional Satisfiability 258 Anton Belov and Zbigniew Stachniak A Theoretical Analysis of Search in GSAT 265 Evgeny S. Skvortsov The Parameterized Complexity of k-Flip Local Search for SAT and MAX SAT 276 Stefan Szeider 9. Hybrid SAT Solvers A Novel Approach to Combine a SLS- and a DPLL-Solver for the Satisfiability Problem 284 Adrian Balint, Michael Henn, and Oliver Gableske Building a Hybrid SAT Solver via Conflict-Driven, Look-Ahead and XOR Reasoning Techniques 298 Jingchao Chen 10. Automatic Adaption of SAT Solvers Restart Strategy Selection Using Machine Learning Techniques 312 Shai Haim and Toby Walsh Instance-Based Selection of Policies for SAT Solvers 326 Mladen Nikoli´c, Filip Mari´c, and Predrag Janiˇci´c Width-Based Restart Policies for Clause-Learning Satisfiability Solvers 341 Knot Pipatsrisawat and Adnan Darwiche Problem-Sensitive Restart Heuristics for the DPLL Procedure 356 Carsten Sinz and Markus Iser 11. Stochastic Approaches to SAT Solving (1,2)-QSAT: A Good Candidate for Understanding Phase Transitions Mechanisms 363 Nadia Creignou, Herv´eDaud´e, Uwe Egly, and Rapha¨el Rossignol VARSAT: Integrating Novel Probabilistic Inference Techniques with DPLL Search 377 Eric I. Hsu and Sheila A. McIlraith [...]... of a cycle Lemma 1 Let C1 and C2 be two equivalent cycles Then C1 and C2 have the same length and shun the same number of cube nodes Proof (sketch) Since C1 and C2 are equivalent, there is a sequence Π of permutations, of the type mentioned in definition 5, such that Π(C1 ) = C2 Reflections and rotations of the coordinate sequence of C1 translate to reversals of C1 ’s orientation, and to rotations of. .. the hybrid number is 3 Hybrid Number and Acyclicity of the Forest We define the graph, GF of a MAF F of two trees T1 and T2 as follows: the nodes are the trees of F , and there is an edge from one node (F1 ) to (F2 ) corresponding to two trees of F if the root of (F1 ) is a descendant of the root of (F2 ) in either T1 or T2 Adding the simple condition that the graph of the forest is acyclic yields a MAF... time and accuracy of both RIATA-HGT and SAT upper bound The standard deviation for the time for RIATA-HGT remains below 2% for all values For all other algorithms, the standard deviations rise for both time and accuracy to almost 20%, illustrating the variability of difficulty of problems even for small and medium values 7 Discussion and Conclusion Encoding problems as SAT instances has positive and negative... variables to reduce the number of clauses needed: ti,j and ti,j for i = 0, , r and j = 1, , n − 1 (these are similar to the s variables used for the leaves of the trees) The clauses for the internal nodes of the trees state: 1 Every internal vertex of T1 (and of T2 ) is in at most one subtree This follows the same idea as in the previous step with v and t for T1 and with v and t for T2 This is done... obtained from H by first deleting a subset of edges of H and any resulting isolated vertices, and then contracting edges Then given two trees T1 and T2 , h(T1 , T2 ) = min{h(H) : H is a hybridization network that displays T1 and T2 } We define the hybridization number of two trees T1 and T2 as the minimal hybridization number of all hybridization network H that display T1 and T2 Agreement Forest Originally... the size of the trees to 39 taxa from the initial 50 taxa and the starting upper bound is 13 The number of literals and clauses depend on the size of the reduced tree pairs and the starting upper bound They are 3,416 literals and 370,571 clauses, a huge reduction from the worst case bound for the full trees and half of the bound calculated for the reduced trees 5 Data We analyze both biological and simulated... problems vs random or handcrafted ones, and lemma learning Technical Univ of Catalonia (UPC), Barcelona, Spain Partially supported by Spanish Min of Science &Innovation, LogicTools-2 project (TIN2007-68093-C02-01) For more details and further references, see Robert Nieuwenhuis, Albert Oliveras and Cesare Tinelli: Solving SAT and SAT Modulo Theories: From an Abstract DavisPutnam-Logemann-Loveland Procedure... binary hypercube The orientation of the edges of the hypercube is determined by the choice of focal points of the PLDE The orientation of the edge shows the direction of the phase flow A part of this work was presented at the 7th Australia – New Zealand Mathematics Convention, Christchurch, New Zealand, December 11, 2008 The work was supported by ETH Research Grant TH-19 06-3 O Kullmann (Ed.): SAT 2009, LNCS... number of T1 and T2 is r iff there is a maximum acyclic agreement forest for T1 and T2 of size r Thus, most of the encoding focuses on saying that a agreement forest exists: Literals For each subtree i in the forest and leaf j from the original leaf set, we have a literal lij which is true iff leaf j is part of subtree i in the agreement forest We have similar sets of literals for internal vertices of T1 and. .. substructure of the subtrees are equal It is possible that all the O(n3 ) triplets of taxa will differ in structure in T1 and T2 , resulting in O(rn3 ) clauses In practice, most trees compared have are similar and as such most of triplets agree, and few are needed For example, the theoretical upper bound for unreduced trees with 50 taxa and with a starting upper bound of 13 is 1,625,000 For a pair chosen at random