In the quest to understand how structure and dynamics are connected in glasses, a number of machine learning based methods have been developed that predict dynamics in supercooled liquids. These methods include both increasingly complex machine learning techniques, and increasingly sophisticated descriptors used to describe the environment around particles. In many cases, both the chosen machine learning technique and choice of structural descriptors are varied simultaneously, making it hard to quantitatively compare the performance of different machine learning approaches. Here, we use three different machine learning algorithms -- linear regression, neural networks, and GNNs -- to predict the dynamic propensity of a glassy binary hard-sphere mixture using as structural input a recursive set of order parameters recently introduced by Boattini et al. [Phys. Rev. Lett. 127, 088007 (2021)]. As we show, when these advanced descriptors are used, all three methods predict the dynamics with nearly equal accuracy. However, the linear regression is orders of magnitude faster to train making it by far the method of choice.
The emergence of glassy dynamics and the glass transition in dense disordered systems is still not fully understood theoretically. Mode-coupling theory (MCT) has shown to be effective in describing some of the non-trivial features of glass formation, but it cannot explain the full glassy phenomenology due to the strong approximations on which it is based. Generalized mode-coupling theory (GMCT) is a hierarchical extension of the theory, which is able to outclass MCT by carefully describing the dynamics of higher order correlations in its generalized framework. Unfortunately, the theory has so …
Soil creeps imperceptibly downhill, but also fails catastrophically to create landslides. Despite the importance of these processes as hazards and in sculpting landscapes, there is no agreed upon model that captures the full range of behavior. Here we examine the granular origins of hillslope soil transport by Discrete Element Method simulations, and re-analysis of measurements in natural landscapes. We find creep for slopes below a critical gradient, where average particle velocity (sediment flux) increases exponentially with friction coefficient (gradient). At critical there is a continuous …
Characterizing the cage state of glassy systems and its sensitivity to frozen boundaries
Debye Institute of Nanomaterials Science, Utrecht University（乌得勒支大学德拜纳米材料科学研究所）；Université Paris-Saclay, CNRS, Laboratoire de Physique des Solides（巴黎萨克雷大学、法国国家科学研究中心、固体物理实验室）
Amorphous solids lack long-range order. Therefore identifying structural defects -- akin to dislocations in crystalline solids -- that carry plastic flow in these systems remains a daunting challenge. By comparing many different structural indicators in computational models of glasses, under a variety of conditions we carefully assess which of these indicators are able to robustly identify the structural defects responsible for plastic flow in amorphous solids. We further demonstrate that the density of defects changes as a function of material preparation and strain in a manner that is highly…
The binary Voronoi mixture is a fluid model whose interactions are local and many-body. Here we perform molecular-dynamics (MD) simulations of an equimolar mixture that is weakly polydisperse and additive. For the first time we study the structural relaxation of this mixture in the supercooled-liquid regime. From the simulations we determine the time- and temperature-dependent scattering functions for a large range of wave vectors, as well as the mean-square displacements of both particle species. We perform a detailed analysis of the dynamics by comparing the MD results with the first-princip…
The boson peak in structural and orientational glasses of simple alcohols: Specific heat at low temperatures
在这项工作中，我们审查比热实验，我们已经进行了不同的氢键玻璃在过去几年中。具体来说，我们已经测量了一组玻璃醇的低温比热Cp：正常和全氘代乙醇，1-和2-丙醇，和甘油。乙醇表现出非常有趣的多态性，在低温下呈现三种不同的固相：完全有序（单斜）晶体，取向无序（立方）晶体或“取向玻璃”，以及普通结构玻璃。通过测量和比较三个相在玻色子峰2-10 K以及低于1 K的隧穿态范围内的低温比热，我们能够定量地证实“玻璃化行为”不是非晶态固体独有的性质。另一方面，丙醇是最简单的一元醇，具有两种不同的立体异构体（1-丙醇和2-丙醇），这使我们能够直接研究原子的空间重排对玻璃普遍性质的影响。我们已经测量了这两种异构体的比热，发现它们之间有一个值得注意的数量差异。最后，还获得了玻璃态甘油的低温比热数据。在这里，我们提出了一个简单的方法的基础上的软势模型来分析低温比热测量，我们用这种方法的定量比较所有这些数据的玻璃醇和作为一个严格的测试的几个普遍的相关性和标度律在文献中提出的。特别地，我们发现玻色子峰的双折射模型[A. V. Granato，Phys. Rev. Lett. 68（1992）974]很好地解释了Cp/T^3出现最大值时的温度。
Active glassy systems can be thought of as simple model systems that imitate complex biological systems. Sometimes, it becomes crucial to estimate the amount of the activity present in such biological systems, such as predicting the progression rate of the cancer cells or the healing time of the wound. In this work, we study a model active glassy system to understand a possible quantification of the degree of activity from the collective, long-range phonon response in the system. We find that the four-point dynamic susceptibility, $χ_4(t)$ at the phonon timescale, grows with increased activity…
Topological Data Analysis combined with Machine Learning for Predicting Permeability of Porous Media
Department of Mathematical Sciences, New Jersey Institute of Technology（新泽西理工学院数学科学系）；Department of Physics, New Jersey Institute of Technology（新泽西理工学院物理系）
Predicting the local dynamics of supercooled liquids based purely on local structure is a key challenge in our quest for understanding glassy materials. Recent years have seen an explosion of methods for making such a prediction, often via the application of increasingly complex machine learning techniques. The best predictions so far have involved so-called Graph Neural Networks (GNN) whose accuracy comes at a cost of models that involve on the order of 10$^5$ fit parameters. In this Letter, we propose that the key structural ingredient to the GNN method is its ability to consider not only th…
