Year: 2023
Type: conference-paper
Abstract: Numerical errors are errors caused by instructions in software code that violate the numerical properties of variables. In this paper, we propose a multi-property bounded model checking approach to detect multiple types of numerical type errors present in a program simultaneously. The method is an optimization of bounded model checking on multiple properties. By establishing a dependency between constraints through the relationship between variable ranges and utilizing this dependency in the validation phase, the number of SAT solutions is reduced and the overall detection efficiency is improved. Experiments show that the method proposed in this paper is able to shrink the detection time of multiple types of numerical errors in the program to a certain extent, and is more applicable to the detection of actual code.
Authors: Qilin Deng, Yu Wu, Fengli Sun, Wei Zhang
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4399254532