Optimal LTLf Synthesis Maximizes Realizable Objectives
This paper introduces optimal LTLf synthesis, a new paradigm for strategy synthesis that maximizes the number of realizable objectives when a specification cannot be fully guaranteed in an uncertain environment. The authors propose three variants: max-guarantee synthesis, which commits to a maximal set of a priori guaranteed objectives; max-observation synthesis, which maximizes a posteriori realized objectives that may vary across executions; and incremental max-observation synthesis, which dynamically improves strategies by exploiting opportunities for stronger guarantees during execution. Experimental results indicate that these optimal synthesis variations scale comparably to standard approaches.
Key facts
- Optimal LTLf synthesis aims to realize as many objectives as possible from a multi-objective specification.
- Max-guarantee synthesis commits to a maximal set of objectives that can be guaranteed a priori.
- Max-observation synthesis maximizes a posteriori realized objectives that may be incomparable on different executions.
- Incremental max-observation synthesis improves strategies by exploiting opportunities for stronger guarantees during execution.
- Experimental results show that different variations of optimal synthesis scale broadly equally.
- The paper addresses the all-or-nothing paradigm of traditional strategy synthesis.
- The approach is designed for cases where not all objectives are jointly realizable.
- The work is published on arXiv with ID 2605.11544.
Entities
Institutions
- arXiv