MultiGain 2.0 is a major extension to the previous MultiGain tool, built on top of the probabilistic model checker PRISM. The new version extends MultiGain's multi-objective capabilities by allowing for the formal verification and synthesis of controllers for probabilistic systems with multi-dimensional long-run average reward structures, steady-state constraints, and linear temporal logic properties.
The tool supports various types of queries, including computing the maximum achievable long-run average reward while satisfying LTL and steady-state constraints, approximating Pareto curves for multi-dimensional rewards, and synthesizing deterministic or unichain policies. MultiGain 2.0 can also modify the underlying linear program to prevent unbounded-memory and other unintuitive solutions, and visualizes Pareto curves in two and three dimensions to facilitate trade-off analysis.
The experimental evaluation demonstrates the scalability of the tool, showing its efficiency in handling grid world models of increasing size. The results also compare the performance of different LP solvers, highlighting the trade-offs between runtime and memory usage. Overall, MultiGain 2.0 provides a powerful and flexible tool for solving complex MDP control problems with heterogeneous specifications.
Naar een andere taal
vanuit de broninhoud
arxiv.org
Belangrijkste Inzichten Gedestilleerd Uit
by Seve... om arxiv.org 05-03-2024
https://arxiv.org/pdf/2305.16752.pdfDiepere vragen