Yu Joins Recipients of Best Paper Award at ASPLOS

Professor Cunxi Yu is the co-author of “SmoothE: Differentiable E-Graph Extraction”, which received the Best Paper Award at the 2025 Association for Computing Machinery (ACM) International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS). The paper is a collaborative effort between Cornell University and the University of Maryland with support by an NSF research program.

The paper focuses on the benefits of e-graphs and methods for the extraction of an expression from large areas of equivalent expressions in order to best meet optimization objectives or cost models. The challenge with current e-graphs is that extraction methods can result in unfavorable scalability-quality trade-offs and only support simple linear cost functions, which limits their applicability to more realistic optimization problems.

In order to address these issues, SmoothE introduces a novel GPU-accelerated differentiable optimization technique for e-graph extraction. The work demonstrates significant potential across a broad range of domains, including design automation, compiler optimization, program synthesis and theorem proving.

Yu, an affiliate faculty member at the Department of Computer Science, joined the ECE Department in 2023, and received Best Paper Award at Design Automation Conference (DAC) in 2023 and NSF CAREER Award in 2021.

—By UMD Electrical and Computer Engineering