Synthesizing probabilistic composers

  • Authors:
  • Sumit Nain;Moshe Y. Vardi

  • Affiliations:
  • Department of Computer Science, Rice University, Houston, TX;Department of Computer Science, Rice University, Houston, TX

  • Venue:
  • FOSSACS'12 Proceedings of the 15th international conference on Foundations of Software Science and Computational Structures
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Synthesis from components is the automated construction of a composite system from a library of reusable components such that the system satisfies the given specification. This is in contrast to classical synthesis, where systems are always "constructed from scratch". In the control-flow model of composition, exactly one component is in control at a given time and control is switched to another when the component reaches an exit state. The composition can then be described implicitly by a transducer, called a composer, which statically determines how the system transitions between components. Recently, Lustig, Nain and Vardi have shown that control-flow synthesis of deterministic composers from libraries of probabilistic components is decidable. In this work, we consider the more general case of probabilistic composers. We show that probabilistic composers are more expressive than deterministic composers, and that the synthesis problem still remains decidable.