HKUST CSE Paper Won SIGPLAN Distinguished Paper Award at OOPSLA

Dr. Amir Goharshady, along with his group members, CSE PhD student S. Hitarth and postdoctoral researcher Dr. Harshit Motwani, have been honored with an ACM SIGPLAN Distinguished Paper Award at the 33rd International Conference on Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA 2023). Their award-winning paper, "Algebro-geometric Algorithms for Template-Based Synthesis of Polynomial Programs," provided new solutions based on theorems from real algebraic geometry in the realm of template-based synthesis, also known as sketching.

About OOPSLA

The ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA) is widely regarded as one of the most influential forums for programming language research. OOPSLA is the premier venue to present innovative work in programming languages design, implementation, theory, applications, and more. Each year, OOPSLA recognizes selected papers that represent extraordinary inventive research likely to advance the field. These Best Paper Awards highlight theoretical and applied contributions poised to shape the future of programming languages and verification.

About the Award-Winning Paper

Their paper introduces a novel algorithm, using ideas from polyhedral and real algebraic geometry, for the automated synthesis of polynomial imperative programs from high-level sketches. It significantly extends previous works that could only handle linear programs.

In sketching-based program synthesis, programmers provide a partial sketch of the desired program structure, as well as a specification that the program must satisfy, and the synthesizer fills in the details. Their method allows efficient synthesis of real-valued polynomial programs without compromising completeness. This work has the potential to greatly expand the range and complexity of programs that can be automatically generated from sketches. The synthesis of tricky polynomial imperative code has been out of reach, but their algebro-geometric techniques open new doors.

The ability to generate programs automatically has major implications for fields like machine learning and could free programmers to work at higher levels of abstraction. This pioneering research from their ALPACAS research group takes an important step towards more powerful program synthesis capabilities.

We extend our hearty congratulations to Dr. Amir Goharshady, S. Hitarth, and Dr. Harshit Motwani for their remarkable achievements.

Award Announcement