Predicting the Sensitivity of Multiscale Coarse-Grained Models to their Underlying Fine-Grained Model Parameters

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Statistical Mechanics

Background:

• Coarse-grained (CG) models are essential for simulating large molecular systems.
• Accurate CG models require understanding their sensitivity to underlying fine-grained (FG) parameters.
• Existing methods for calculating this sensitivity are computationally expensive or inaccurate.

Purpose of the Study:

• To develop a computationally efficient and accurate method for calculating CG model sensitivity.
• To improve the transferability of CG models across interaction parameters and thermodynamic conditions.
• To provide insights into the limitations of current CG models.

Main Methods:

• Introduction of the self-consistent basis (SCB) single point formula for reweighting-free, single-simulation sensitivity calculation.
• Development of the self-consistent iterative (SCI) single point formula for identifying many-body error sources.
• Application of the SCB formula to methanol and sodium chloride CG models.

Main Results:

• The SCB formula provides practical, high signal-to-noise calculations of CG model sensitivity.
• Demonstrated substantially reduced noise compared to previous methods.
• Identified remaining challenges with bias in the SCB method.

Conclusions:

• The SCB formula offers a novel, computationally efficient method for assessing alchemical transferability.
• This work advances the understanding of CG model transferability limits.
• The developed methods enable high-fidelity calculations at low computational cost.