Rigorous Error Bounds for Ewald Summation of Electrostatics at Planar Interfaces

Factorization in molecular modeling and belief propagation algorithms

Factorization reduces computational complexity, and is therefore an important tool in statistical machine learning of high dimensional systems. Conventional molecular modeling, including molecular dynamics and Monte Carlo simulations of molecular systems, is a large research field based on approximate factorization of molecular interactions. Recently, the local distribution theory was proposed to factorize joint distribution of a given molecular system into trainable local distributions. Belief propagation algorithms are a family of exact factorization algorithms for (junction) trees, and are extended to approximate loopy belief propagation algorithms for graphs with loops. Despite the fact that factorization of probability distribution is the common foundation, computational research in molecular systems and machine learning studies utilizing belief propagation algorithms have been carried out independently with respective track of algorithm development. The connection and differences among these factorization algorithms are briefly presented in this perspective, with the hope to intrigue further development of factorization algorithms for physical modeling of complex molecular systems.

A formula and numerical study on Ewald 1D summation

Ewald summation is famous for its successful applications in molecular simulations for systems under 2 dimensional periodic boundary condition (2D PBC, e.g., planar interfaces) and systems under 3D PBC (e.g., bulk). However, the extension to systems under 1D PBC (like porous structures and tubes) is largely hindered by the special functions in the formula. In this work, a simple approximation of Ewald 1D sum is introduced with its error rigorously controlled. To investigate the impacts on the efficiency and accuracy by different parts, a pairwise potential is calculated for a series of screening parameters ( α ) and radial distances ( ρ ) between two point charges. A mapping between the sum of trigonometric functions in Ewald 1D method and the sum of specific vectors further reveals the different converging speeds of different Fourier parts. When choosing α = 0.2 Å^-1 , it is appropriate to ignore the insignificant parts in the sum to accelerate the method.

Analytic theory of finite-size effects in supercell modeling of charged interfaces

The Ewald3D sum with the tinfoil boundary condition (e3dtf) evaluates the electrostatic energy of a finite unit cell inside an infinitely periodic supercell. Although it has been used as a de facto standard treatment of electrostatics for simulations of extended polar or charged interfaces, the finite-size effect on simulated properties has yet to be fully understood. There is, however, an intuitive way to quantify the average effect arising from the difference between the e3dtf and Coulomb potentials on the response of mobile charges to contact surfaces with fixed charges and/or to an applied external electric field. Although any charged interface formed by mobile countercharges that compensate the fixed charges fluctuates upon a change in the acting electric field, the distance between a pair of oppositely charged interfaces is found to be nearly stationary, which allows an analytic finite-size correction to the amount of countercharges. Application of the theory to solvated electric double layers (insulator/electrolyte interfaces) predicts that the state of complete charge compensation is invariant with respect to solvent permittivities, which is confirmed by a proper analysis of simulation data in the literature.

On the connections and differences among three mean-field approximations: a stringent test

This letter attempts to clarify the meaning of three closely related mean-field approximations: random phase approximation (RPA), local molecular field (LMF) approximation, and symmetry-preserving mean-field (SPMF) approximation, and their use of reliability and validity in the field of theory and simulation of liquids when the long-ranged component of the intermolecular interaction plays an important role in determining density fluctuations and correlations. The RPA in the framework of classical density functional theory (DFT) neglects the higher order correlations in the bulk and directly applies the long-ranged part of the potential to correct the pair direct correlation function of the short-ranged system while the LMF approach introduces a nonuniform mimic system under a reconstructed static external potential that accounts for the average effect arising from the long-ranged component of the interaction. Furthermore, the SPMF approximation takes the viewpoint of LMF but instead instantaneously averages the long-ranged component of the potential over the degrees of freedom in the direction with preserved symmetry. The formal connections and the particular differences of the viewpoint among the three approximations are explained and their performances in producing structural properties of liquids are stringently tested using an exactly solvable model. We demonstrate that the RPA treatment often yields uncontrolled poor results for pair distribution functions of the bulk system. On the other hand, the LMF theory produces quite reasonably structural correlations when the pair distribution in the bulk is converted to the singlet particle distribution in the nonuniform system. It turns out that the SPMF approach outperforms the other two at all densities and under extreme conditions where the long-ranged component significantly contributes to the structural correlations.

The effect of electrostatic boundaries in molecular simulations: symmetry matters

Artifacts arise when the long-ranged electrostatic interaction is inappropriately treated in molecular simulations of electrolytes. When the usual Ewald3D sum method with the tinfoil boundary condition (e3dtf) is used for simulations of an interfacial liquid under an external electric field, a straightforward analysis of the liquid structure often suggests unphysical dielectric properties as a consequence of the inaccurate treatment of the electrostatics. In order to understand the underlying mechanism that leads to this apparent violation of thermodynamics, we now derive a new equation in the weak-field limit that, in a mean field view, accounts for the average effect arising from the difference between e3dtf and the sophisticated Ewald2D sum method (e2d). Numerical simulations of a water system in slab geometry confirm the validity of the weak-field limit equation for a series of parameter setup associated with e3dtf. Moreover, a similar procedure applied to a spherically confined water system suggests that corrections to the seemingly inappropriate treatment of the electrostatics in fact vanish. This cancellation of the boundary effect due to symmetry immediately sheds light on the long-lasting problem of the validity of the ad hoc application of e3dtf for bulk systems. In total, we argue that artifacts arising from e3dtf are often predictable and analytical corrections to the straightforward analysis might be applied to reveal consistent thermodynamic properties in liquid simulations.

Molecular insight into the dynamical adsorption behavior of nanoscale water droplets on a heterogeneous surface

A heterogeneous surface is constructed by adding one hydrophilic patch at the center of a hydrophobic surface, and the dynamical adsorption process of nanoscale water droplets is investigated adopting molecular dynamics simulations.

Symmetry-preserving mean field theory for electrostatics at interfaces

A novel method is developed for complex nonuniform electrostatics in computer simulations of molecular liquids at interfaces.