Exact Projection Functions for Floating Point Number Constraints.
Claude Michel · 2002
This paper introduces a new framework for filtering constraints over floating-point numbers. An important application area where such solvers are required is test case generation, i.e., finding input values for which a selected point in a procedure is executed. It has been shown that the latter problem can be handled efficiently by translating a non-trivial imperative program into a CSP over finite domains. However, when these programs contain arithmetic operations that involve floating-point numbers, the challenge is to compute test data that are valid even when the arithmetic operations performed by the program to be tested are unsafe. Filtering techniques used in solvers...