Year: 2023
Type: conference-paper
Abstract: Despite great progress in algorithms for synthesizing recursive programs, state of the art approaches continue to have several limitations. Principal among these is their inability to “invent” auxiliary functions. This makes them sensitive to the available set of primitive components. In this paper, we present an alternative approach to recover recursive programs. We develop a system of constraints that characterizes patterns of data flow in the unrollings of a recursive program. Combined with a generator of seed nonrecursive circuits and a constraint solver, these constraints naturally form the basis of a general algorithm to synthesize recursive circuits.
Author: Marta Davila Mateu
Institution: University of Southern California
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4387793214
