Diese Arbeit befasst sich mit Worst-Case-Algorithmen für das Erfüllbarkeitsproblem boolescher Ausdrücke in konjunktiver Normalform. Im Wesentlichen betrachten wir Laufzeitschranken drei verschiedener Algorithmen, zwei für 3-SAT und einen für Unique-k-SAT. Wir entwickeln einen randomisierten Algorithmus, der eine Lösung eines erfüllbaren 3-KNF-Ausdrucks G mit n Variablen mit einer erwarteten Laufzeit von O(1.32793^n) findet. Der Algorithmus basiert auf der Analyse sogenannter Strings, welche Sequenzen von Klauseln der Länge drei sind. Dabei dürfen einerseits nicht aufeinanderfolgende Klauseln keine Variablen und andererseits aufeinanderfolgende Klauseln ein oder zwei Variablen gemeinsam haben. Gibt es wenige Strings, so treffen wir wahrscheinlich bereits während der String-Suche auf eine Lösung von G. 1999 entwarf Schöning einen Algorithmus mit einer Schranke von O(1.3334^n) für 3-SAT. Viele Strings erlauben es, die Laufzeit dieses Algorithmusses zu verbessern. Weiterhin werden wir den PPSZ-Algorithmus für Unique-k-SAT derandomisieren. Der 1998 von Paturi, Pudlak, Saks und Zane vorgestellte PPSZ-Algorithmus hat die besondere Eigenschaft, dass die Lösung eines eindeutig erfüllbaren 3-KNF-Ausdrucks in höchstens O(1.3071^n) erwarteter Laufzeit gefunden wird. Die derandomisierte Variante des Algorithmusses für Unique-k-SAT hat im Wesentlichen die gleiche Laufzeitschranke. Wir erreichen damit die momentan beste deterministische Worst-Case-Schranke für Unique-k-SAT. Zur Derandomisierung wenden wir die "Methode der kleinen Zufallsräume" an. Schließlich verbessern wir die Schranke für den Algorithmus von Iwama und Tamaki. 2003 kombinierten Iwama und Tamaki den PPSZ-Algorithmus mit Schönigs Algorithmus und konnten eine Schranke von O(1.3238^n) bewiesen. Um seinen Beitrag zum kombinierten Algorithmus zu steigern, justieren wir die Schranke des PPSZ-Algorithmusses. Damit erhalten wir die momentan beste randomisierte Worst-Case-Schranke für das 3-SAT-Problem von O(1.32216^n). / This work deals with worst-case algorithms for the satisfiability problem regarding boolean formulas in conjunctive normal form. The main part of this work consists of the analysis of the running time of three different algorithms, two for 3-SAT and one for Unique-k-SAT. We establish a randomized algorithm that finds a satisfying assignment for a satisfiable 3-CNF formula G on n variables in O(1.32793^n) expected running time. The algorithm is based on the analysis of so-called strings, which are sequences of clauses of size three, whereby non-succeeding clauses do not share a variable, and succeeding clauses share one or two variables. If there are not many strings, it is likely that we already encounter a solution of G while searching for strings. In 1999, Schöning proved a bound of O(1.3334^n) for 3-SAT. If there are many strings, we use them to improve the running time of Schöning''s algorithm. Furthermore, we derandomize the PPSZ algorithm for Unique-k-SAT. The PPSZ algorithm presented by Paturi, Pudlak, Saks, and Zane in 1998 has the feature that the solution of a uniquely satisfiable 3-CNF formula can be found in expected running time at most O(1.3071^n). In general, we will obtain a derandomized version of the algorithm for Unique-k-SAT that has essentially the same bound as the randomized version. This settles the currently best known deterministic worst-case bound for the Unique-k-SAT problem. We apply the `Method of Small Sample Spaces'' in order to derandomize the algorithm. Finally, we improve the bound for the algorithm of Iwama and Tamaki to get the currently best known randomized worst-case bound for the 3-SAT problem of O(1.32216^n). In 2003 Iwama and Tamaki combined Schöning''s and the PPSZ algorithm to yield an O(1.3238^n) bound. We tweak the bound for the PPSZ algorithm to get a slightly better contribution to the combined algorithm.

Este trabalho estuda o problema da Satisfazibilidade Probabilística (PSAT), revendo a sua solução via programação linear, além de propor novos algoritmos para resolvê-lo através da redução ao SAT. Construímos uma redução polinomial do PSAT para o SAT, chamada de Redução Canônica, codificando operações da aritmética racional em bits, como variáveis lógicas. Analisamos a complexidade computacional dessa redução e propomos uma Redução Canônica de Precisão Limitada para contornar tal complexidade. Apresentamos uma Redução de Turing do PSAT ao SAT, baseada no algoritmo Simplex e na Forma Normal Atômica que introduzimos. Sugerimos modificações em tal redução em busca de eficiência computacional. Por fim, implementamos essas reduções a m de investigar o perl de complexidade do PSAT, observamos o fenômeno de transição de fase e discutimos as condições para sua detecção. / This work studies the Probabilistic Satisfiability problem (PSAT), reviewing its solution through linear programming, and proposing new algorithms to solve it. We construct a polynomial many-to-one reduction from PSAT to SAT, called Canonical Reduction, codifying rational arithmetic operations into bits, as logical variables. We analyze the computational complexity of this reduction and we propose a Limited Precision Canonical Reduction to reduce such complexity. We present a Turing Reduction from PSAT to SAT, based on the Simplex algorithm and the Atomic Normal Form we introduced. We suggest modifications in such reduction looking for computational eficiency. Finally, we implement these reductions in order to investigate the complexity profile of PSAT, the phase transition phenomenom is observed and the conditions for its detection are discussed.

Este trabalho estuda o problema da Satisfazibilidade Probabilística (PSAT), revendo a sua solução via programação linear, além de propor novos algoritmos para resolvê-lo através da redução ao SAT. Construímos uma redução polinomial do PSAT para o SAT, chamada de Redução Canônica, codificando operações da aritmética racional em bits, como variáveis lógicas. Analisamos a complexidade computacional dessa redução e propomos uma Redução Canônica de Precisão Limitada para contornar tal complexidade. Apresentamos uma Redução de Turing do PSAT ao SAT, baseada no algoritmo Simplex e na Forma Normal Atômica que introduzimos. Sugerimos modificações em tal redução em busca de eficiência computacional. Por fim, implementamos essas reduções a m de investigar o perl de complexidade do PSAT, observamos o fenômeno de transição de fase e discutimos as condições para sua detecção. / This work studies the Probabilistic Satisfiability problem (PSAT), reviewing its solution through linear programming, and proposing new algorithms to solve it. We construct a polynomial many-to-one reduction from PSAT to SAT, called Canonical Reduction, codifying rational arithmetic operations into bits, as logical variables. We analyze the computational complexity of this reduction and we propose a Limited Precision Canonical Reduction to reduce such complexity. We present a Turing Reduction from PSAT to SAT, based on the Simplex algorithm and the Atomic Normal Form we introduced. We suggest modifications in such reduction looking for computational eficiency. Finally, we implement these reductions in order to investigate the complexity profile of PSAT, the phase transition phenomenom is observed and the conditions for its detection are discussed.

Boolean satisfiability (SAT) is the problem of determining whether there exists an assignment of the Boolean variables to the truth values such that a given Boolean formula evaluates to true. SAT was the first example of an NP-complete problem. Only two decades ago SAT was mainly considered as of a theoretical interest. Nowadays, the picture is very different. SAT solving becomes mature and is a successful approach for tackling a large number of applications, ranging from artificial intelligence to industrial hardware design and verification. SAT solving consists of encodings and solvers. In order to benefit from the tremendous advances in the development of solvers, one must first encode the original problems into SAT instances. These encodings should not only be easily generated, but should also be efficiently processed by SAT solvers. Furthermore, an increasing number of practical applications in computer science can be expressed as constraint satisfaction problems (CSPs). However, encoding a CSP to SAT is currently regarded as more of an art than a science, and choosing an appropriate encoding is considered as important as choosing an algorithm. Moreover, it is much easier and more efficient to benefit from highly optimized state-of-the-art SAT solvers than to develop specialized tools from scratch. Hence, finding appropriate SAT encodings of CSPs is one of the most fascinating challenges for solving problems by SAT. This thesis studies SAT encodings of CSPs and aims at: 1) conducting a comprehensively profound study of SAT encodings of CSPs by separately investigating encodings of CSP domains and constraints; 2) proposing new SAT encodings of CSP domains; 3) proposing new SAT encoding of the at-most-one constraint, which is essential for encoding CSP variables; 4) introducing the redundant encoding and the hybrid encoding that aim to benefit from both two efficient and common SAT encodings (i.e., the sparse and order encodings) by using the channeling constraint (a term used in Constraint Programming) for SAT; and 5) revealing interesting guidelines on how to choose an appropriate SAT encoding in the way that one can exploit the availability of many efficient SAT solvers to solve CSPs efficiently and effectively. Experiments show that the proposed encodings and guidelines improve the state-of-the-art SAT encodings of CSPs.

This dissertation shows that satisfiability procedures are abstract interpreters. This insight provides a unified view of program analysis and satisfiability solving and enables technology transfer between the two fields. The framework underlying these developments provides systematic recipes that show how intuition from satisfiability solvers can be lifted to program analyzers, how approximation techniques from program analyzers can be integrated into satisfiability procedures and how program analyzers and satisfiability solvers can be combined. Based on this work, we have developed new tools for checking program correctness and for solving satisfiability of quantifier-free first-order formulas. These tools outperform existing approaches. We introduce abstract satisfaction, an algebraic framework for applying abstract interpre- tation to obtain sound, but potentially incomplete satisfiability procedures. The framework allows the operation of satisfiability procedures to be understood in terms of fixed point computations involving deduction and abduction transformers on lattices. It also enables satisfiability solving and program correctness to be viewed as the same algebraic problem. Using abstract satisfaction, we show that a number of satisfiability procedures can be understood as abstract interpreters, including Boolean constraint propagation, the dpll and cdcl algorithms, St ̊almarck’s procedure, the dpll(t) framework and solvers based on congruence closure and the Bellman-Ford algorithm. Our work leads to a novel understand- ing of satisfiability architectures as refinement procedures for abstract analyses and allows us to relate these procedures to independent developments in program analysis. We use this perspective to develop Abstract Conflict-Driven Clause Learning (acdcl), a rigorous, lattice-based generalization of cdcl, the central algorithm of modern satisfiability research. The acdcl framework provides a solution to the open problem of lifting cdcl to new prob- lem domains and can be instantiated over many lattices that occur in practice. We provide soundness and completeness arguments for acdcl that apply to all such instantiations. We evaluate the effectiveness of acdcl by investigating two practical instantiations: fp-acdcl, a satisfiability procedure for the first-order theory of floating point arithmetic, and cdfpl, an interval-based program analyzer that uses cdcl-style learning to improve the precision of a program analysis. fp-acdcl is faster than competing approaches in 80% of our benchmarks and it is faster by more than an order of magnitude in 60% of the benchmarks. Out of 33 safe programs, cdfpl proves 16 more programs correct than a mature interval analysis tool and can conclusively determine the presence of errors in 24 unsafe benchmarks. Compared to bounded model checking, cdfpl is on average at least 260 times faster on our benchmark set.

Behandelt werden zufällige lineare Gleichungssysteme modulo 3, wobei in jeder Gleichung genau k Variablen vorkommen. Es wird gezeigt, dass der Schwellwert der Lösbarkeit solcher Gleichungssysteme bei der 2-Kern-Dichte von 1 liegt. Das Resultat ist eine Verallgemeinerung bereits bekannter Resultate für den modulo 2 Fall. Dabei entsteht der 2-Kern dadurch, dass wir alle Variablen mit nur einem Vorkommen löschen. Die Dichte ist definiert als der Quotient der Anzahl der Gleichungen durch die Anzahl der Variablen. Im Rückblick ist dieses Resultat ein natürlicher Schwellwert und die Vermutung liegt nahe, dass er bei analogen Situationen über anderen Strukturen als Z3 auch gelten sollte. Allerdings sind schon im modulo 2 Fall die analytischen Probleme nicht gering, und der hier behandelte Fall braucht weitere analytische Einsichten.

Answer set programming (ASP) is a declarative programming paradigm oriented towards difficult combinatorial search problems. Syntactically, ASP programs look like Prolog programs, but solutions are represented in ASP by sets of atoms, and not by substitutions, as in Prolog. Answer set systems, such as Smodels, Smodelscc, and DLV, compute answer sets of a given program in the sense of the answer set (stable model) semantics. This is different from the functionality of Prolog systems, which determine when a given query is true relative to a given logic program. ASP has been applied to many areas of science and technology, from the design of a decision support system for the Space Shuttle to graph-theoretic problems arising in zoology and linguistics. The "native" answer set systems mentioned above are based on specialized search procedures. Usually these procedures are described fairly informally with the use of pseudocode. We propose an alternative approach to describing algorithms of answer set solvers. In this approach we specify what "states of computation" are, and which transitions between states are allowed. In this way, we define a directed graph such that every execution of a procedure corresponds to a path in this graph. This allows us to model algorithms of answer set solvers by a mathematically simple and elegant object, graph, rather than a collection of pseudocode statements. We use this abstract framework to describe and prove the correctness of the answer set solver Smodels, and also of Smodelscc, which enhances the former using learning and backjumping techniques. Answer sets of a tight program can be found by running a SAT solver on the program's completion, because for such a program answer sets are in a one-to-one correspondence with models of completion. SAT is one of the most widely studied problems in computational logic, and many efficient SAT procedures were developed over the last decade. Using SAT solvers for computing answer sets allows us to take advantage of the advances in the SAT area. For a nontight program it is still the case that each answer set corresponds to a model of program's completion but not vice versa. We show how to modify the search method typically used in SAT solvers to allow testing models of completion and employ learning to utilize testing information to guide the search. We develop a new SAT-based answer set solver, called Cmodels, based on this idea. We develop an abstract graph based framework for describing SAT-based answer set solvers and use it to represent the Cmodels algorithm and to demonstrate its correctness. Such representations allow us to better understand similarities and differences between native and SAT-based answer set solvers. We formally compare the Smodels algorithm with a variant of the Cmodels algorithm without learning. Abstract frameworks for describing native and SAT-based answer set solvers facilitate the development of new systems. We propose and implement the answer set solver called SUP that can be seen as a combination of computational ideas behind Cmodels and Smodels. Like Cmodels, solver SUP operates by computing a sequence of models of completion of the given program, but it does not form the completion. Instead, SUP runs the Atleast algorithm, one of the main building blocks of the Smodels procedure. Both systems Cmodels and SUP, developed in this dissertation, proved to be competitive answer set programming systems. / text

This dissertation explores approaches to the satisfiability problem, focusing on local search methods. The research endeavours to better understand how and why some local search methods are effective. At the root of this understanding are a set of metrics that characterize the behaviour of local search methods. Based on this understanding, two new local search methods are proposed and tested, the first, SDF, demonstrating the value of the insights drawn from the metrics, and the second, ESG, achieving state-of-the-art performance and generalizing the approach to arbitrary 0-1 integer linear programming problems. This generality is demonstrated by applying ESG to combinatorial auction winner determination. Further augmentations to local search are proposed and examined, exploring hybrids that incorporate aspects of backtrack search methods.

Since its introduction in 1999, bounded model checking (BMC) has quickly become a serious and indispensable tool for the formal verification of hardware designs and, more recently, software. By leveraging propositional satisfiability (SAT) solvers, BMC overcomes some of the shortcomings of more conventional model checking methods. In model checking we automatically verify whether a state transition system (STS) describing a design has some property, commonly expressed in linear temporal logic (LTL). BMC is the restriction to only checking the looping and non-looping runs of the system that have bounded descriptions. The conventional BMC approach is to translate the STS runs and LTL formulae into propositional logic and then conjunctive normal form (CNF). This CNF expression is then checked by a SAT solver. In this thesis we study the effect on the performance of BMC of changing the translation to propositional logic. One novelty is to use a normal form for LTL which originates in resolution theorem provers. We introduce the normal form conversion early on in the encoding process and examine the simplifications that it brings to the generation of propositional logic. We further enhance the encoding by specialising the normal form to take advantage of the types of runs peculiar to BMC. We also improve the conversion from propositional logic to CNF. We investigate the behaviour of the new encodings by a series of detailed experimental comparisons using both hand-crafted and industrial benchmarks from a variety of sources. These reveal that the new normal form based encodings can reduce the solving time by a half in most cases, and up to an order of magnitude in some cases, the size of the improvement corresponding to the complexity of the LTL expression. We also compare our method to the popular automata-based methods for model checking and BMC.

This thesis describes a decision and minimization procedure for modal logic. The decision procedure answers the question of whether there exists a satisfying pointed model for a formula which obeys user-specified first-order conditions on the underlying frame. Then the minimization procedure produces a minimal model with respect to the number of worlds that satisfies the desired formula while obeying the requisite conditions on the underlying frame. A proof of correctness for the decision and minimization procedures is supplied, as well as a description of an implementation built upon the Enfragmo model expansion solver. / Graduate / 0984 / 0318 / wbkboyer@gmail.com