5 November 2020 in Journal of Physics: Condensed Matter
Condensed matter physics (CMP) seeks to understand the microscopic interactions of matter at the quantum and atomistic levels, and describes how these interactions result in both mesoscopic and macroscopic properties. CMP overlaps with many other important branches of science, such as chemistry, materials science, statistical physics, and high-performance computing. With the advancements in modern machine learning (ML) technology, a keen interest in applying these algorithms to further CMP research has created a compelling new area of research at the intersection of both fields. In this review, we aim to explore the main areas within CMP, which have successfully applied ML techniques to further research, such as the description and use of ML schemes for potential energy surfaces, the characterization of topological phases of matter in lattice systems, the prediction of phase transitions in off-lattice and atomistic simulations, the interpretation of ML theories with physics-inspired frameworks and the enhancement of simulation methods with ML algorithms. We also discuss in detail the main challenges and drawbacks of using ML methods on CMP problems, as well as some perspectives for future developments.
25 June 2024 in The Journal of Chemical Physics
The development of new materials typically involves a process of trial and error, guided by insights from past experimental and theoretical findings. The inverse design approach for soft-matter systems has the potential to optimize specific physical parameters, such as particle interactions, particle shape, or composition and packing fraction. This optimization aims to facilitate the spontaneous formation of specific target structures through self-assembly. In this study, we expand upon a recently introduced inverse design protocol for monodisperse systems to identify the required conditions and interactions for assembling crystal and quasicrystal phases within a binary mixture of two distinct species. This method utilizes an evolution algorithm to identify the optimal state point and interaction parameters, enabling the self-assembly of the desired structure. In addition, we employ a convolutional neural network (CNN) that classifies different phases based on their diffraction patterns, serving as a fitness function for the desired structure. Using our protocol, we successfully inverse design two-dimensional crystalline structures, including a hexagonal lattice and a dodecagonal quasicrystal, within a non-additive binary mixture of hard disks. Finally, we introduce a symmetry-based order parameter that leverages the encoded symmetry within the diffraction pattern. This order parameter circumvents the need for training a CNN and is used as a fitness function to inverse design an octagonal quasicrystal.
1 July 2022 in Physics of Fluids
The Ornstein–Zernike equation is solved for the hard-sphere and square-well fluids using a diverse selection of closure relations; the attraction range of the square-well is chosen to be . In particular, for both fluids, we mainly focus on the solution based on a three-parameter version of the Verlet closure relation [Verlet, “Integral equations for classical fluids. II. Hard spheres again,” Mol. Phys. 42, 1291–1302 (1981)]. To find the free parameters of the latter, an unconstrained optimization problem is defined as a condition of thermodynamic consistency based on the compressibility and solved using Evolutionary Algorithms. For the hard-sphere fluid, the results show good agreement when compared with mean-field equations of state and accurate computer simulation results; at high densities, i.e., close to the freezing transition, expected (small) deviations are seen. In the case of the square-well fluid, good agreement is observed at low and high densities when compared with event-driven molecular dynamics computer simulations. For intermediate densities, the explored closure relations vary in terms of accuracy. Our findings suggest that a modification of the optimization problem to include, for example, additional thermodynamic consistency criteria could improve the results for the type of fluid here explored.
4 November 2025 in The Journal of Chemical Physics
We present a comprehensive study of the structure, formation, and dynamics of a one-component model system that self-assembles into an icosahedral quasicrystal (IQC). Using molecular dynamics simulations combined with unsupervised machine learning techniques, we identify and characterize the unique structural motifs of IQCs, including icosahedral and dodecahedral arrangements, and quantify the evolution of local environments during the IQC formation process. Our analysis reveals that the formation of the IQC is driven by the emergence of distinct local clusters that serve as precursors to the fully developed quasicrystalline phase. In addition, we examine the dynamics of the system across a range of temperatures, identifying transitions from vibrationally restricted motion to activated diffusion and uncovering signatures of dynamic heterogeneity inherent to the quasicrystalline state. To directly connect structure and dynamics, we use a machine-learning-based order parameter to quantify the presence of distinct local environments across temperatures. We find that regions with high structural order, as captured by specific machine-learned classes, correlate with suppressed self-diffusion and minimal dynamical heterogeneity, consistent with phason-like motion within the IQC. In contrast, regions with lower structural order exhibit enhanced collective motion and increased dynamical heterogeneity. These results establish a quantitative framework for understanding the coupling between structural organization and dynamical processes in quasicrystals, providing new insights into the mechanisms governing IQC stability and dynamics.
22 April 2025 in The Journal of Chemical Physics
The hard-sphere potential has become a cornerstone in the study of both molecular and complex fluids. Despite its mathematical simplicity, its implementation in fixed time step molecular simulations remains a formidable challenge due to the discontinuity at contact. To avoid the issues associated with the ill-defined force at contact, a continuous potential has recently been proposed—here referred to as the pseudo-hard-sphere potential (pHS) [Báez et al., J. Chem, Phys. 149, 164907 (2018)]. This potential is constructed to match the second virial coefficient of the hard-sphere potential and is expected to mimic its thermodynamic properties. However, this hypothesis has only been partially validated within the fluid region of the phase diagram for hard-sphere dispersions in two and three dimensions. In this contribution, we examine the ability of the continuous pHS potential to reproduce the equation of state of a hard-sphere fluid, not only in the fluid phase but also across the fluid–solid coexistence region. Our focus is primarily on the phase diagram of hard-sphere systems in three and four dimensions; however, we also report on the feasibility of the pHS to reproduce the long time dynamics of a three-dimensional colloidal dispersion. We compare the thermodynamic properties obtained from Brownian dynamics simulations of the pHS potential with those derived from refined event-driven simulations of the corresponding hard-sphere potential. Furthermore, we provide a comparative analysis with theoretical equations of state based on both mean-field and integral equation approximations.