Genetic Algorithm for Program Synthesis

TL;DR

This paper enhances a deductive program synthesis tool by integrating evolutionary computation to improve search efficiency, demonstrating generalization to new problems through cross-validation.

Contribution

It introduces a novel hybrid approach combining deductive synthesis with evolutionary algorithms to accelerate program derivation.

Findings

Improved search speed in program synthesis

Generalization of the method to unseen problems

Successful integration of evolutionary computation with deductive methods

Abstract

A deductive program synthesis tool takes a specification as input and derives a program that satisfies the specification. The drawback of this approach is that search spaces for such correct programs tend to be enormous, making it difficult to derive correct programs within a realistic timeout. To speed up such program derivation, we improve the search strategy of a deductive program synthesis tool, SuSLik, using evolutionary computation. Our cross-validation shows that the improvement brought by evolutionary computation generalises to unforeseen problems.