A microcanonical approach to temperature-transferable coarse-grained models using the relative entropy.

Area of Science:

• Computational chemistry
• Molecular modeling
• Statistical mechanics

Background:

• Coarse-graining (CG) methods simplify molecular systems but often struggle with thermodynamic property accuracy and transferability to different conditions.
• Existing CG models face limitations in representing state-dependent systems accurately.
• The representability and transferability issues hinder the broad application of CG models.

Purpose of the Study:

• To develop a novel strategy for creating temperature-transferable coarse-grained (CG) models.
• To overcome the limitations of existing CG models in thermodynamic property prediction and transferability.
• To enable the use of CG models at various thermodynamic state points.

Main Methods:

• Formulating CG models in a microcanonical basis using effective entropy functions instead of energy functions.
• Incorporating information about atomistic energy fluctuations into CG models.
• Employing relative entropy minimization for systematic parameterization of temperature-transferable CG models.

Main Results:

• The developed CG models demonstrate accurate reproduction of reference atomistic energy distributions.
• The approach successfully predicts the temperature dependence of CG interactions, achieving temperature transferability.
• The method was validated using both idealized systems and complex molecular liquids.

Conclusions:

• The proposed strategy effectively creates temperature-transferable CG models from a single reference system and temperature.
• Focusing on effective entropy functions and utilizing relative entropy minimization enhances model accuracy and transferability.
• This work provides a robust framework for building reliable CG models for diverse molecular systems.