First hitting and strong metastability times
F.Manzo, E.Scoppola

TL;DR
This paper derives exact results for the distribution of first hitting times in non-reversible Markov chains, introducing a new concept of strong metastability time via a generalized strong stationary time, applicable to general initial distributions.
Contribution
It introduces the concept of strong metastability time through a generalized quasi-stationary time, providing a new framework for understanding metastability in non-reversible Markov chains.
Findings
Exact distribution results for first hitting times.
Introduction of the conditionally strong quasi-stationary time (CSQST).
Clarification of the role of initial distribution on exponential hitting time laws.
Abstract
In the setting of non-reversible Markov chains on finite or countable state space, exact results on the distribution of the first hitting time to a given set are obtained. A new notion of "strong metastability time" is introduced to describe the local relaxation time. This time is defined via a generalization of the strong stationary time to a "conditionally strong quasi-stationary time"(CSQST). Rarity of the target set is not required and the initial distribution can be completely general. The results clarify the the role played by the initial distribution on the exponential law; they are used to give a general notion of metastability and to discuss the relation between the exponential distribution of the first hitting time and metastability.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSoftware System Performance and Reliability
