Reproduce “Multiagent Stochastic Shortest Path Problem” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Martin Jonáš, Antonín Kučera, Vojtěch Kůr, Jan Mačák, Vojtěch Řehák

MSSP asks how autonomous agents can minimize expected time until the first reaches a target in a known Markov decision process. AUTOHIT uses gradient descent on a differentiable approximation to search randomized strategy profiles. On random city-like maps, LP-initialized runs averaged 96.0% of the single-agent-optimal baseline’s expected first-arrival time; this does not establish global optimality or generalize beyond tested instances.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.