Papers
Topics
Authors
Recent
Search
2000 character limit reached

Gamma Expansions of Level Two Large Deviation Rate Functionals for Metastable Diffusions

Published 15 Oct 2026 in math.PR | (2610.22117v1)

Abstract: Consider the diffusions dXt<sup>ε=b(Xt<sup>ε)dt+2εσ(Xt<sup>ε)dBtdX_t<sup>\varepsilon=b(X_t<sup>\varepsilon)dt+\sqrt{2\varepsilon}σ(X_t<sup>\varepsilon)dB_t on R<sup>d\mathbb{R}<sup>d, where bb and σσ are smooth coefficients and BtB_t is an rr-dimensional Brownian motion, with r≤dr\le d. We allow nonreversible dynamics and a degenerate diffusion matrix. Let I<em>ε\mathcal{I}<em>\varepsilon be the level-two large deviation rate functional of the empirical measure. Under our assumptions, including a complete metastable hierarchy described by resolvent convergence, we establish the full ΓΓ-expansion I</em>ε=1εJ<sup>(−1)+J<sup>(0)+∑p=1<sup>q1θε<sup>(p)J<sup>(p)\mathcal{I}</em>\varepsilon=\frac{1}{\varepsilon}\mathcal{J}<sup>{(-1)}+\mathcal{J}<sup>{(0)}+\sum_{p=1}<sup>{\mathfrak{q}}\frac{1}{θ_\varepsilon<sup>{(p)}}\mathcal{J}<sup>{(p)} in the sense of successive ΓΓ-limits in the weak topology on probability measures. Here θε<sup>(p)θ_\varepsilon<sup>{(p)} are the metastable time scales, and each J<sup>(p)\mathcal{J}<sup>{(p)}, p≥1p\ge1, is the lift of the level-two rate functional of the corresponding finite-state Markov chain. We establish the pre-metastable ΓΓ-limits for generalized overdamped, underdamped, and third-order Langevin dynamics. For the generalized overdamped model, existing metastability results yield the full expansion under additional conditions. For the underdamped and third-order models, the full expansion remains conditional on verification of the complete metastable hierarchy.

Authors (2)

Summary

No one has generated a summary of this paper yet.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Continue Learning

We haven't generated follow-up questions for this paper yet.

Tweets

Sign up for free to view the 1 tweet with 0 likes about this paper.