Abstract:We propose a data-driven framework for identifying coarse-grained (CG) Lennard-Jones (LJ) potential parameters in confined systems for simple liquids. Our approach involves the use of a Deep Neural Network (DNN) that is trained to approximate the solution of the Inverse Liquid State (ILST) problem for confined systems. The DNN model inherently incorporates essential physical characteristics specific to confined fluids, enabling accurate prediction of inhomogeneity effects. By utilizing transfer learning, we predict single-site LJ potentials of simple multiatomic liquids confined in a slit-like channel, which effectively replicate both the fluid structure and molecular force of the target All-Atom (AA) system when the electrostatic interactions are not dominant. In addition, we showcase the synergy between the data-driven approach and the well-known Bottom-Up coarse-graining method utilizing Relative-Entropy (RE) Minimization. Through sequential utilization of these two methods, the robustness of the iterative RE method is significantly augmented, leading to a remarkable enhancement in convergence.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to perform coarse - grained modeling of simple liquids in constrained systems (such as nano - channels), especially how to determine the Lennard - Jones (LJ) potential parameters of these systems. Specifically, the authors propose a data - driven method, using deep neural networks (DNN) to approximately solve the inverse liquid - state problem (ILST) of the constrained system, thereby accurately predicting the inhomogeneity effect. ### Main problems and solutions 1. **Problem background** - **Coarse - graining requirement**: Traditional all - atom molecular dynamics (AA - MD) or ab - initio molecular dynamics (AIMD) methods have high computational costs and are difficult to simulate complex molecular systems on large spatio - temporal scales. - **Characteristics of constrained systems**: In constrained systems such as nano - channels, fluid structures and molecular forces exhibit unique properties, which are different from bulk - phase systems, so special modeling methods are required. 2. **Existing challenges** - **Limitations of bottom - up methods**: Existing bottom - up coarse - graining methods (such as iterative Boltzmann inversion IBI, relative entropy minimization RE, etc.) are based on statistical mechanics principles, but they have deficiencies in identifying global minima and sensitivity to initial conditions, and it is difficult to achieve transferability under different thermodynamic states. - **Limitations of top - down methods**: Top - down methods rely on experimental data fitting, but can only match macroscopic properties and cannot describe the microstructure in detail. 3. **Proposed data - driven method** - **Deep - learning model**: The authors use DNN to learn the mapping relationship between wall - fluid and fluid - fluid forces and LJ potential parameters. By training DNN, the inverse liquid - state problem can be approximately solved, thereby obtaining coarse - grained parameters suitable for constrained systems. - **Application of transfer learning**: Through transfer learning, DNN can predict the single - site LJ potential of multi - atomic simple liquids in slit channels, effectively replicating the fluid structure and molecular forces of the target AA system, especially when charge interactions are not dominant. 4. **Combined with relative entropy minimization** - **Synergistic effect**: Combining the data - driven method with the relative entropy minimization method significantly enhances the robustness and convergence of the iterative RE method. The initial potential generated by DNN is used as the input of the RE framework, making the two methods complementary to each other and improving the overall performance. ### Formula summary - **Single - particle density**: \[ \rho^{(1)}(\mathbf{r})=\frac{1}{Z_N}\sum_{i = 1}^N\int\exp\left[-V_N(\mathbf{r}_1,\ldots,\mathbf{r}_i,\ldots,\mathbf{r}_N)-\sum_{j = 1}^N U(\mathbf{r}_j)\right]d\mathbf{r}_1\cdots d\hat{\mathbf{r}}_i\cdots d\mathbf{r}_N \] - **External potential energy (single - wall)**: \[ U(z)=\frac{2}{3\pi}\rho_w\sigma_{wf}^3\epsilon_{wf}\left[\frac{2}{15}\left(\frac{\sigma_{wf}}{z}\right)^9-\left(\frac{\sigma_{wf}}{z}\right)^3\right] \] - **Average fluid - fluid force**: \[ \bar{f}_{ff}(z)=-\int\rho^{(1)}(z)g(z',z,R)\frac{\partial V(z',z,R)}{\partial z}dR dz' \] - **Average wall - fluid force**: \[ \bar{f}_{wf}(z)

A Data-Driven Approach to Coarse-Graining Simple Liquids in Confinement

Temperature-transferable coarse-graining of ionic liquids with dual graph convolutional neural networks

Accumulative Coarse-Graining of Simple Structural Descriptors Enables Accurately Predicting the Dynamics of Metallic Liquids

A coarse-grained deep neural network model for liquid water

A New Approach For Learning Coarse-Grained Potentials with Application to Immiscible Fluids

Transferable Coarse Graining Via Contrastive Learning of Graph Neural Networks

Large Scale and Quantum Accurate Molecular Dynamics Simulation:Liquid Iron under Extreme Condition

Neural Network-Assisted Model of Interfacial Fluids with Explicit Coarse-Grained Molecular Structures.

Reinforcement learning based hybrid bond-order coarse-grained interatomic potentials for exploring mesoscale aggregation in liquid–liquid mixtures

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

Machine Learning of coarse-grained Molecular Dynamics Force Fields

Improving the reliability of machine learned potentials for modeling inhomogenous liquids

Phase behaviour of coarse-grained fluids

High-Dimensional Neural Network Potential for Liquid Electrolyte Simulations.

Intelligent dissipative particle dynamics: Bridging mesoscopic models from microscopic simulations via deep neural networks

Data-driven dynamical coarse-graining for condensed matter systems

Machine learning assisted coarse-grained molecular dynamics modeling of meso-scale interfacial fluids

DiffGLE: Differentiable Coarse-Grained Dynamics using Generalized Langevin Equation

Coarse-Grained Modeling of Liquid-Liquid Phase Separation in Cells: Challenges and Opportunities

Coarse Graining of Multiple Water Molecules with Intelligent Dissipative Particle Dynamics