Background

Program synthesis — the problem of automatically discovering a program that fits a given specification — is a classic problem in computer science. In the traditional formulation of the problem, the specification is a formal constraint, and the synthesizer's objective is to search for a program that satisfies this constraint. More recent research has extended this problem statement, sometimes requiring the synthesizer to optimize quantitative objective functions over programs and generalize to unseen specifications.

Program synthesis is an AI-hard problem, but there has been significant progress on the problem in the recent past. Contemporary program synthesis algorithms use symbolic techniques for pruning the search space of programs and discovering formally verified program parameters, and neural techniques for interpreting ambiguous specifications, representing modules that operate on perceptual inputs, and learning to search for programs. Recent work on this topic has had applications in a wide range of areas, including software engineering, automation of end user tasks, robotics and control, systems biology, and computer science education.

CS 395T is a course on the theory and practice of program synthesis at UT Austin. This Zoom workshop presents work done as part of course projects in the Spring 2020 edition of CS 395T.

Program

• Jon Stephens. Synthesizing Smart Contracts.   (Zoom link)   (poster)
• Chenxi Yang. Directing Program Synthesis by Extracting Image Features for Refinement Types.   (Zoom link)   (poster)
• Greg Anderson. Program Synthesis in Continuous Domains.   (Zoom link)   (poster)
• Benjamin Ghaemmaghami. Sorghum: Program Synthesis for Coarse Grained Spatial Architectures.   (Zoom link)   (poster)
• William Macke. Syntax Evolution through Augmenting Topologies.   (Zoom link)   (poster)
• Yang Hu. Learning Program Transformation for Sketching.   (Zoom link)   (poster)
• Wenxi Wang. Generating Program Sketches from Natural Language Descriptions .   (Zoom link)   (poster)
• Lucas Kabela. Neural Program Synthesis with Soft Actor Critic.   (Zoom link)   (poster)
• Siyu Zhang. Regression through program synthesis.   (Zoom link)   (poster)
• Rahul Krishnan. Comparing Conflict-Driven Learning on SyGuS.   (Zoom link)   (poster)
• Shankara Pailoor. Program Refinement from Data structure integrity constraints.   (Zoom link)   (poster)
• Kai-Chi Huang. Neuro-symbolic environment model learning.   (Zoom link)   (poster)
• Ben Mariano. Program Synthesis with Quantitative Objectives and Interpretable Neural Program Induction.   (Zoom link)   (poster)
• Sai Kiran Narayanaswami. Synthesizing Safe Multi-Agent Environments.   (Zoom link)   (poster)

Program Chair + Instructor

Swarat Chaudhuri