Abstract Fixpoint Computations with Numerical Acceleration Methods

📝 Original Info

• Title: Abstract Fixpoint Computations with Numerical Acceleration Methods
• ArXiv ID: 1006.3159
• Date: 2013-05-02
• Authors: Researchers from original ArXiv paper

📝 Abstract

Static analysis by abstract interpretation aims at automatically proving properties of computer programs. To do this, an over-approximation of program semantics, defined as the least fixpoint of a system of semantic equations, must be computed. To enforce the convergence of this computation, widening operator is used but it may lead to coarse results. We propose a new method to accelerate the computation of this fixpoint by using standard techniques of numerical analysis. Our goal is to automatically and dynamically adapt the widening operator in order to maintain precision.

💡 Deep Analysis

Deep Dive into Abstract Fixpoint Computations with Numerical Acceleration Methods.

Static analysis by abstract interpretation aims at automatically proving properties of computer programs. To do this, an over-approximation of program semantics, defined as the least fixpoint of a system of semantic equations, must be computed. To enforce the convergence of this computation, widening operator is used but it may lead to coarse results. We propose a new method to accelerate the computation of this fixpoint by using standard techniques of numerical analysis. Our goal is to automatically and dynamically adapt the widening operator in order to maintain precision.

📄 Full Content

Reference