Abstract:A shadow molecular dynamics scheme for flexible charge models is presented, where the shadow Born-Oppenheimer potential is derived from a coarse-grained approximation of range-separated density functional theory. The interatomic potential, including the atomic electronegativities and the charge-independent short-range part of the potential and force terms, are modeled by the linear atomic cluster expansion (ACE), which provides a computationally efficient alternative to many machine learning methods. The shadow molecular dynamics scheme is based on extended Lagrangian (XL) Born-Oppenheimer molecular dynamics (BOMD) [Eur. Phys. J. B 94, 164 (2021)]. XL-BOMD provides a stable dynamics, while avoiding the costly computational overhead associated with solving an all-to-all system of equations, which normally is required to determine the relaxed electronic ground state prior to each force evaluation. To demonstrate the proposed shadow molecular dynamics scheme for flexible charge models using the atomic cluster expansion, we emulate the dynamics generated from self-consistent charge density functional tight-binding (SCC-DFTB) theory using a second-order charge equilibration (QEq) model. The charge-independent potentials and electronegativities of the QEq model are trained for a supercell of uranium oxide (UO2) and a molecular system of liquid water. The combined ACE + XL-QEq dynamics are stable over a wide range of temperatures both for the oxide and the molecular systems, and provide a precise sampling of the Born-Oppenheimer potential energy surfaces. Accurate ground Coulomb energies are produced by the ACE-based electronegativity model during an NVE simulation of UO2, predicted to be within 1 meV of those from SCC-DFTB on average during comparable simulations.

What problem does this paper attempt to address?

This paper attempts to address the problem of how to effectively handle the long-range charge relaxation issue in flexible charge models within molecular dynamics simulations to improve computational efficiency and accuracy. Specifically, the authors propose a Shadow Molecular Dynamics scheme based on Extended Lagrangian (XL), combined with the Atomic Cluster Expansion (ACE) method, to construct a flexible charge model. This approach aims to avoid the high computational cost associated with solving long-range Coulomb interactions between all atoms before each force calculation, while maintaining the stability and accuracy of the simulation. ### Main Issues: 1. **Long-range charge relaxation problem**: In traditional molecular dynamics simulations, handling flexible charge models requires solving long-range Coulomb interactions between all atoms before each force calculation, leading to very high computational costs. 2. **Balance between computational efficiency and accuracy**: Existing machine learning methods can provide high-accuracy short-range interaction potentials but cannot effectively handle the long-range charge relaxation problem. Therefore, a new method is needed to balance computational efficiency and simulation accuracy. ### Solutions: 1. **Shadow Molecular Dynamics**: By introducing a Shadow Potential Surface, it is possible to avoid solving the long-range charge relaxation problem without significantly losing accuracy. 2. **Atomic Cluster Expansion (ACE)**: The ACE method is used to model short-range interaction potentials and atomic electronegativity, providing a computationally efficient alternative. 3. **Extended Lagrangian (XL) framework**: Combining the XL framework allows for efficient handling of the integration of electronic degrees of freedom while maintaining simulation stability. ### Specific Implementation: - **Shadow Potential Surface**: By linearizing the energy function, a shadow energy function \( E^{(2)}(R, \eta, n) \) dependent on the approximate ground state solution \( n \) is constructed, thus obtaining the shadow Born-Oppenheimer potential surface \( U_{BO}(R, n) \). - **Atomic Cluster Expansion (ACE)**: The ACE method is used to model short-range interaction potentials and atomic electronegativity, providing a tunable and systematically improvable method for describing local atomic properties. - **Extended Lagrangian (XL) framework**: By introducing extended Lagrangian variables, the electronic degrees of freedom are propagated as dynamic field variables, avoiding the nonlinear quantum mechanical Kohn-Sham eigenvalue problem. ### Application Examples: - **Uranium Dioxide (UO₂)**: Molecular dynamics simulations in solid uranium dioxide validate the stability and accuracy of this method at high temperatures. - **Liquid Water (H₂O)**: Molecular dynamics simulations in liquid water demonstrate the stability of this method at different temperatures and its accurate sampling of the Born-Oppenheimer potential surface. Through these methods, the authors successfully address the long-range charge relaxation problem in flexible charge models, improving the computational efficiency and accuracy of molecular dynamics simulations.

Shadow molecular dynamics and atomic cluster expansions for flexible charge models

Charge-constrained Atomic Cluster Expansion

Charge equilibration model with shielded long-range Coulomb for reactive molecular dynamics simulations

Stochastic Constrained Extended System Dynamics for Solving Charge Equilibration Models

Implementation and Optimization of the Embedded Cluster Reference Interaction Site Model with Atomic Charges

Embedded-Cluster Calculations in a Numeric Atomic Orbital Density-Functional Theory Framework

Advanced Charge Extended Hückel (CEH) Model and a Consistent Adaptive Minimal Basis Set for the Elements Z=1-103

Cartesian atomic cluster expansion for machine learning interatomic potentials

Expanding the Scope of Density Derived Electrostatic and Chemical Charge Partitioning to Thousands of Atoms

Atomic cluster expansion for quantum-accurate large-scale simulations of carbon

Systematic Improvement of Empirical Energy Functions in the Era of Machine Learning

Advanced Potential Energy Surfaces for Molecular Simulation

Van Der Waals Coefficients Beyond the Classical Shell Model

Extended Lagrangian Born-Oppenheimer Molecular Dynamics with DFT+U

Generalized Method for Charge Transfer Equilibration in Reactive Molecular Dynamics

Subsystem Density Functional Theory Augmented by a Delta Learning Approach to Achieve Kohn–Sham Accuracy

Semi-Empirical Shadow Molecular Dynamics: A PyTorch implementation

Self-consistent Coulomb interactions for machine learning interatomic potentials

Scalable molecular simulation of electrolyte solutions with quantum chemical accuracy