Extending top-down enumerative optimizations to SemGuS problems

Identify methods to extend domain-specific optimizations used in top-down enumerative program synthesis to synthesis problems expressed in the SemGuS framework, enabling effective pruning of partial programs with holes under semantics defined via Constrained Horn Clauses.

Background

The paper implements a top-down enumerative synthesizer that expands partial programs with holes and verifies complete candidates against examples. In traditional settings, top-down enumerators leverage domain-specific pruning optimizations to eliminate partial programs early.

Because SemGuS allows users to define arbitrary languages and semantics via CHCs, existing domain-specific optimizations do not directly carry over. The authors explicitly note that discovering how to generalize or adapt these optimizations to SemGuS constitutes an open question.