Challenges in Decomposing Encodings of Verification Problems
نویسنده
چکیده
Modern program verifiers use logic-based encodings of the verification problem that are discharged by a back end reasoning engine. However, instances of such encodings for large programs can quickly overwhelm these back end solvers. Hence, we need techniques to make the solving process scale to large systems, such as partitioning (divide-and-conquer) and abstraction. In recent work, we showed how decomposing the formula encoding of a termination analysis can significantly increase efficiency. The analysis generates a sequence of logical formulas with existentially quantified predicates that are solved by a synthesis-based program analysis engine. However, decomposition introduces abstractions in addition to those required for finding the unknown predicates in the formula, and can hence deteriorate precision. We discuss the challenges associated with such decompositions and their interdependencies with the solving process.
منابع مشابه
Modular Schemes for Constructing Equivalent Boolean Encodings of Cardinality Constraints and Application to Error Diagnosis in Formal Verification of Pipelined Microprocessors
We present a novel method for generating a wide range of equivalent Boolean encodings of cardinality, while in contrast all previous Boolean encodings of cardinality have only one form. Experiments for applying this method to automated error diagnosis in formal verification of buggy variants of a complex reconfigurable VLIW processor indicate speedup of up to two orders of magnitude, relative t...
متن کاملChallenging SMT solvers to verify neural networks
In this paper we evaluate state-of-the-art SMT solvers on encodings of verification problems involving Multi-Layer Perceptrons (MLPs), a widely used type of neural network. Verification is a key technology to foster adoption of MLPs in safety-related applications, where stringent requirements about performance and robustness must be ensured and demonstrated. In previous contributions, we have s...
متن کاملEncodings of Non-Binary Constraint Satisfaction Problems
We perform a detailed theoretical and empirical comparison of the dual and hidden variable encodings of non-binary constraint satisfaction problems. We identify a simple relationship between the two encodings by showing how we can translate between the two by composing or decomposing relations. This translation suggests that we will tend to achieve more pruning in the dual than in the hidden va...
متن کاملChecking Safety of Neural Networks with SMT Solvers: A Comparative Evaluation
In this paper we evaluate state-of-the-art SMT solvers on encodings of verification problems involving Multi-Layer Perceptrons (MLPs), a widely used type of neural network. Verification is a key technology to foster adoption of MLPs in safety-related applications, where stringent requirements about performance and robustness must be ensured and demonstrated. While safety problems for MLPs can b...
متن کاملQBF-Based Formal Verification: Experience and Perspectives
The language of Quantified Boolean Formulas (QBF) has a lot of potential applications to Formal Verification (FV) tasks, as it captures many of these tasks in a natural and compact way. Practical experience has been disappointing though. When compared with contending approaches such as SAT, QBF-based FV has invariably yielded unfavorable experimental results. This paper makes two contributions....
متن کامل