David R. MacIver • 6/17/2026

Generating provably solvable instances for testing solvers: Do you wanna test a SAT solver? (It doesn't have to be a SAT solver)

This article presents a novel property-based testing technique for generating provably solvable instances to test solvers like SAT solvers. It explains the limitations of standard tests that only verify when a solver claims a problem is solvable, and introduces a method to also test when it claims unsolvability. The technique uses Hypothesis and Hegel frameworks, with code examples in Python, and is applicable to any solver or optimizer, such as scheduling problems. It emphasizes the importance of testing both directions of solver correctness.

0 comments

#property-based testing #Hypothesis #Sat