Computing rates in the over-biasing regime within EATR

Establish a method to compute unbiased barrier-crossing rates from Exponential Average Time-Dependent Rate (EATR) simulations in the over-biasing regime where the scaled bias γ times the applied metadynamics bias exceeds the true free-energy barrier, so as to prevent rate overestimation and enable systematic application to large systems with many slow degrees of freedom.

Background

The paper introduces EATR, a unified framework for extracting unbiased transition rates from time-dependent biased simulations that bridges iMetaD and KTR. While EATR performs well across a range of systems and biasing schedules, the authors note a failure mode when the applied bias becomes very large relative to the actual barrier height, particularly at fast deposition times.

In this over-biasing regime, their empirical results (e.g., for the LD1 coordinate at fast deposition times) suggest potential rate overestimation and small γ values. They explicitly state that a general solution for computing rates under these conditions has not yet been achieved, and stress its importance for applications to complex biomolecular systems with many slow degrees of freedom.