DiffGLE: Differentiable Coarse-Grained Dynamics using Generalized Langevin Equation

Jinu Jeong,Ishan Nadkarni,Narayana. R. Aluru

2024-10-11

Abstract:Capturing the correct dynamics at the Coarse-Grained (CG) scale remains a central challenge in the advancement of systematic CG models for soft matter simulations. The Generalized Langevin Equation (GLE), rooted in the Mori-Zwanzig formalism, provides a robust framework for incorporating friction and stochastic forces into CG models, that are lost due to the reduction in degrees of freedom. Leveraging recent advancements in Automatic Differentiation (AD) and reformulating the non-Markovian GLE using a colored noise ansatz, we present a top-down approach for accurately parameterizing the non-Markovian GLE for different coarse-grained fluids that accurately reproduces the velocity-autocorrelation function of the original All-Atom (AA) model. We demonstrate our approach on two different fluids namely, SPC/E water and carbon dioxide which have distinct structure and dynamical characteristics. Importantly, by being end-to-end differentiable, this approach offers a simplified and efficient solution to achieving accurate CG dynamics, effectively bypassing the complexities inherent in most bottom-up methods.

Soft Condensed Matter

What problem does this paper attempt to address?

The paper attempts to address the issue of accurately capturing system dynamics in Coarse-Grained (CG) models. Specifically, the authors focus on how to construct dynamically accurate CG models through the Generalized Langevin Equation (GLE) to overcome the problem of inaccurate dynamics predictions in existing CG models when simulating soft matter. This inaccuracy is mainly reflected in the fact that, compared to All-Atom (AA) models, the dynamics processes in CG models are too fast, making it impossible to accurately study the dynamic effects and transport properties of the system on relevant time scales. To achieve this goal, the authors propose an end-to-end differentiable method based on Automatic Differentiation (AD) technology for accurately parameterizing non-Markovian GLE, thereby accurately reproducing the Velocity-Autocorrelation Function (VACF) of the original AA model. This method not only simplifies the parameterization process but also improves computational efficiency, avoiding the complexity of traditional bottom-up approaches. In the paper, the authors demonstrate the effectiveness of this method in two different fluids—SPC/E water and carbon dioxide—proving its effectiveness in maintaining dynamic accuracy and structural properties.

DiffGLE: Differentiable Coarse-Grained Dynamics using Generalized Langevin Equation

Understanding Dynamics in Coarse-Grained Models: IV. Connection of Fine-Grained and Coarse-Grained Dynamics with the Stokes-Einstein and Stokes-Einstein-Debye Relations

Understanding Dynamics in Coarse-Grained Models: III. Roles of Rotational Motion and Translation-Rotation Coupling in Coarse-Grained Dynamics

Machine Learning of coarse-grained Molecular Dynamics Force Fields

Coarse-graining conformational dynamics with multi-dimensional generalized Langevin equation: how, when, and why

ML-GLE: A machine learning enhanced Generalized Langevin equation framework for transient anomalous diffusion in polymer dynamics

Data-driven coarse-grained modeling of polymers in solution with structural and dynamic properties conserved

Structural Coarse-Graining via Multi-Objective Optimization with Differentiable Simulation

Implicit-solvent coarse-grained modeling for polymer solutions via Mori-Zwanzig formalism

A First Principle Approach to Rescale the Dynamics of Simulated Coarse-Grained Macromolecular Liquids

Chemically Specific Systematic Coarse-Grained Polymer Model with Both Consistently Structural and Dynamical Properties

Two for One: Diffusion Models and Force Fields for Coarse-Grained Molecular Dynamics

Data-driven dynamical coarse-graining for condensed matter systems

On the generalization ability of coarse-grained molecular dynamics models for non-equilibrium processes

Internal noise-driven generalized Langevin equation from a nonlocal continuum model

Generalized Langevin Equation Formulation for Anomalous Polymer Dynamics

Coarse-Graining with Equivariant Neural Networks: A Path Towards Accurate and Data-Efficient Models

Internal noise driven generalized Langevin equation from a nonlocal continuum model

Simulate Time-integrated Coarse-grained Molecular Dynamics with Geometric Machine Learning

ADD: Analytically Differentiable Dynamics for Multi-Body Systems with Frictional Contact