Gradient Descent over Metagrammars for Syntax-Guided Synthesis

TL;DR

This paper introduces a gradient descent approach over metagrammars to automatically improve syntax-guided synthesis performance, leading to more benchmarks solved faster by CVC4.

Contribution

It proposes a novel method of learning metagrammars via gradient descent to enhance syntax-guided synthesis efficiency and effectiveness.

Findings

Metagrammars improved benchmark solving by 26% within 300 seconds.

Learned metagrammars generalize well across many benchmarks.

The approach outperforms default grammars in synthesis tasks.

Abstract

The performance of a syntax-guided synthesis algorithm is highly dependent on the provision of a good syntactic template, or grammar. Provision of such a template is often left to the user to do manually, though in the absence of such a grammar, state-of-the-art solvers will provide their own default grammar, which is dependent on the signature of the target program to be sythesized. In this work, we speculate this default grammar could be improved upon substantially. We build sets of rules, or metagrammars, for constructing grammars, and perform a gradient descent over these metagrammars aiming to find a metagrammar which solves more benchmarks and on average faster. We show the resulting metagrammar enables CVC4 to solve 26% more benchmarks than the default grammar within a 300s time-out, and that metagrammars learnt from tens of benchmarks generalize to performance on 100s of benchmarks.