Program Synthesis via Bi-directional Reduced-product Abstract Interpretation - N40AI'24

Explore a 39-minute conference talk on program synthesis using bi-directional reduced-product abstract interpretation. Delve into Kwangkeun Yi's presentation of a novel approach to example-based program synthesis that combines forward and backward abstract interpretations to efficiently reduce the search space of programs. Learn how this method iteratively analyzes incomplete programs to determine if they can satisfy given input-output examples. Discover the significant performance improvements achieved in synthesizing bit-manipulation programs compared to state-of-the-art approaches. Gain insights into this sophisticated abstract interpretation technique and its applications in inductive program synthesis, as presented at PLDI 2023.

[N40AI'24] Program Synthesis via Bi-directional Reduced-product Abstract Interpretation