Tero Laitinen

Extending Clause Learning DPLL with Parity Reasoning

We consider a combined satisfiability problem where an instance is given in two parts: a set of traditional clauses extended with a set of parity (xor) constraints. To solve such problems without translation to CNF, we develop a parity constraint reasoning method that can be integrated to a clause learning solver. The idea is to devise a module that offers a decision procedure and implied literal detection for parity constraints and also provides clausal explanations for implied literals and conflicts. We have implemented the method and integrated it to a state-of-the-art clause learning solver. The resulting system is experimentally evaluated and compared to state-of-the-art solvers.

Conflict-driven XOR-clause learning

Modern conflict-driven clause learning (CDCL) SAT solvers are very good in solving conjunctive normal form (CNF) formulas. However, some application problems involve lots of parity (xor) constraints which are not necessarily efficiently handled if translated into CNF. This paper studies solving CNF formulas augmented with xor-clauses in the DPLL(XOR) framework where a CDCL SAT solver is coupled with a separate xor-reasoning module. New techniques for analyzing xor-reasoning derivations are developed, allowing one to obtain smaller CNF clausal explanations for xor-implied literals and also to derive and learn new xor-clauses. It is proven that these new techniques allow very short unsatisfiability proofs for some formulas whose CNF translations do not have polynomial size resolution proofs, even when a very simple xor-reasoning module capable only of unit propagation is applied. The efficiency of the proposed techniques is evaluated on a set of challenging logical cryptanalysis instances.

Classifying and propagating parity constraints

Parity constraints, common in application domains such as circuit verification, bounded model checking, and logical cryptanalysis, are not necessarily most efficiently solved if translated into conjunctive normal form. Thus, specialized parity reasoning techniques have been developed in the past for propagating parity constraints. This paper studies the questions of deciding whether unit propagation or equivalence reasoning is enough to achieve full propagation in a given parity constraint set. Efficient approximating tests for answering these questions are developed. It is also shown that equivalence reasoning can be simulated by unit propagation by adding a polynomial amount of redundant parity constraints to the problem. It is proven that without using additional variables, an exponential number of new parity constraints would be needed in the worst case. The presented classification and propagation methods are evaluated experimentally.

Equivalence Class Based Parity Reasoning with DPLL(XOR)

The recently introduced DPLL (XOR) framework for deciding satisfiability of propositional formulas with parity constraints is studied. A new parity reasoning module, based on equivalence class manipulation, is developed and implementation techniques for it described. It is shown that the deduction power of the new module is equivalent to another one proposed earlier. Additional reasoning module independent techniques are presented. Different design choices and module integration strategies are experimentally evaluated on three stream ciphers Trivium, Grain, and Hitag2. The new approach achieves major runtime speedups on the Trivium cipher and significant reduction in the number of decisions on Grain and Hitag2 ciphers.

Extending Clause Learning SAT Solvers with Complete Parity Reasoning

Instances of logical cryptanalysis, circuit verification, and bounded model checking can often be succinctly represented as a combined satisfiability (SAT) problem where an instance is a combination of traditional clauses and parity constraints. This paper studies how such combined problems can be efficiently solved by augmenting a modern SAT solver with an xor-reasoning module in the DPLL(XOR) framework. A new xor-reasoning module that deduces all possible implied literals using incremental Gauss-Jordan elimination is presented. A decomposition technique that can greatly reduce the size of parity constraint matrices while still allowing to deduce all implied literals is presented. It is shown how to eliminate variables occuring only in parity constraints while preserving the decomposition. The proposed techniques are evaluated experimentally.

Simulating Parity Reasoning

Propositional satisfiability (SAT) solvers, which typically operate using conjunctive normal form (CNF), have been successfully applied in many domains. However, in some application areas such as circuit verification, bounded model checking, and logical cryptanalysis, instances can have many parity (xor) constraints which may not be handled efficiently if translated to CNF. Thus, extensions to the CNF-driven search with various parity reasoning engines ranging from equivalence reasoning to incremental Gaussian elimination have been proposed. This paper studies how stronger parity reasoning techniques in the DPLL(XOR) framework can be simulated by simpler systems: resolution, unit propagation, and parity explanations. Such simulations are interesting, for example, for developing the next generation SAT solvers capable of handling parity constraints efficiently.