Adapting to the Behavior of Environments with Bounded Memory

TL;DR

This paper investigates the complexity of synthesizing controllers in environments with limited memory, providing tighter bounds and proposing a safe learning approach that is computationally more feasible.

Contribution

It tightens the complexity bounds for environment-adaptive synthesis and introduces a safe learning framework with lower computational complexity.

Findings

Complexity bounds for environment synthesis are refined to PSPACE lower bound.

Safe environment behavior learning is shown to be co-NP-complete.

Environment behavior can be learned with polynomial space in the size of the transducer.

Abstract

We study the problem of synthesizing implementations from temporal logic specifications that need to work correctly in all environments that can be represented as transducers with a limited number of states. This problem was originally defined and studied by Kupferman, Lustig, Vardi, and Yannakakis. They provide NP and 2-EXPTIME lower and upper bounds (respectively) for the complexity of this problem, in the size of the transducer. We tighten the gap by providing a PSPACE lower bound, thereby showing that algorithms for solving this problem are unlikely to scale to large environment sizes. This result is somewhat unfortunate as solving this problem enables tackling some high-level control problems in which an agent has to infer the environment behavior from observations. To address this observation, we study a modified synthesis problem in which the synthesized controller must gather information about the environment's behavior safely. We show that the problem of determining whether the behavior of such an environment can be safely learned is only co-NP-complete. Furthermore, in such scenarios, the behavior of the environment can be learned using a Turing machine that requires at most polynomial space in the size of the environment's transducer.