[madPL] Synthesis / SemGuS Reading Group


Date: Wed, 2 Jun 2021 00:18:58 +0000
From: Keith Jens Carl Johnson <keith.johnson@xxxxxxxx>
Subject: [madPL] Synthesis / SemGuS Reading Group

Hi Everyone,

 

You are welcome to join the Synthesis / SemGuS Reading Group, which meets Thursdays from 1 to 2 on Zoom:

https://uwmadison.zoom.us/j/96450343227?pwd=emtGcHROSVZoOWsyWEFhN2wyazA2Zz09

 

 

This week, I will be presenting on Combining the top-down propagation and bottom-up enumeration for inductive program synthesis (Lee, 2021).

 

Abstract:

We present an effective method for scalable and general-purpose inductive program synthesis. There have been two main approaches for inductive synthesis: enumerative search, which repeatedly enumerates possible candidate programs, and the top-down propagation (TDP), which recursively decomposes a given large synthesis problem into smaller subproblems. Enumerative search is generally applicable but limited in scalability, and the TDP is efficient but only works for special grammars or applications. In this paper, we synergistically combine the two approaches. We generate small program subexpressions via enumerative search and put them together into the desired program by using the TDP. Enumerative search enables to bring the power of TDP into arbitrary grammars, and the TDP helps to overcome the limited scalability of enumerative search. We apply our approach to a standard formulation, syntax-guided synthesis (SyGuS), thereby supporting a broad class of inductive synthesis problems. We have implemented our approach in a tool called Duet and evaluate it on SyGuS benchmark problems from various domains. We show that Duet achieves significant performance gains over existing general-purpose as well as domain-specific synthesizers.

 

âHomework Questionâ:

Inverse semantics operators (a.k.a. witness functions) are used to deduce specifications on sub-terms of an operator, given an outer specification on that operator. For example, consider the term Substring(str, start, len). Given a spec that this term must satisfy, the jobs of the witness functions are to deduce specs on str, start, and len.

 

For Substring(str, start, len), think about how you can constrain str from a spec on the entire term. These specs are typically given as input-output examples, so given that Substring(str, start, len) must return a string s when evaluated on an input state i, what might a witness function for str look like?

 

 

-Keith

[← Prev in Thread] Current Thread [Next in Thread→]