Heymann, M., Vanden-Eijnden, E. (2008). The geometric minimum action method: A least action principle on the space of curves. Communications on Pure and Applied Mathematics, 61(8), 1052-1117. doi:10.1002/cpa.20238.

Read in part on 2026-11-03 (NOTE-1504, Skimmed; THEORY-1309).

Key takeaways

  • V(x₁, x₂) = 2 inf ∫ |b|_a sin²(θ/2) ds over curves (eq. 1.18).
  • Gradient drift, identity noise: the escape path is the reversed relaxation path (p. 1058).
  • Maier–Stein: gradient only at β = 1, yet the path is the heteroclinic orbit up to β = 4 (§4.1).

Standing in the record

Computes the most likely transition paths of a noisy system. Bears on the note of 2026-11-02 on QUESTION-071, that a transition’s path, not its end state, can tell a noise-driven escape from a directed one.