A geometry-enhanced graph neural network for learning the smoothness of glassy dynamics from static structure.

Modeling the dynamics of glassy systems has been challenging in physics for several decades. Recent studies have shown the efficacy of Graph Neural Networks (GNNs) in capturing particle dynamics from the graph structure of glassy systems. However, current GNN methods do not take the dynamic patterns established by neighboring particles explicitly into account. In contrast to these approaches, this paper introduces a novel dynamical parameter termed "smoothness" based on the theory of graph signal processing, which explores the dynamic patterns from a graph perspective. Present graph-based approaches encode structural features without considering smoothness constraints, leading to a weakened correlation between structure and dynamics, particularly on short timescales. To address this limitation, we propose a Geometry-enhanced Graph Neural Network (Geo-GNN) to learn the smoothness of dynamics. Results demonstrate that our method outperforms state-of-the-art baselines in predicting glassy dynamics. Ablation studies validate the effectiveness of each proposed component in capturing smoothness within dynamics. These findings contribute to a deeper understanding of the interplay between glassy dynamics and static structure.

Correlation between the topologically close-packed structure and the deformation behavior of metallic Cu64.5Zr35.5.

The topologically close-packed (TCP) structural characteristics in a model metallic glass (MG) of Cu64.5Zr35.5 have been investigated by molecular dynamics simulations. A group of structural indicators based on the largest standard cluster (LaSC) have been correlated with the non-affine displacement (D2) of atoms, so as to reveal the hidden correlation between local structures and deformation behavior of Cu64.5Zr35.5 during compression. It was found that the 15 types of Top-10 LaSCs are all TCP-like ones, among which the most numerous icosahedron (Z12 and 1-Z12) decreases in population sharply and moderately during respectively the elastic and yield region of compression; while in the fluid-flow region, the number of all Top-10 LaSCs tends to be almost constant. Low-D2 atoms prefer to link with each other; while medium-D2 atoms act as transition structures between backbone areas and deformation areas. Most interestingly, the deformation response of TCP-like atoms is not only determined by its nearest neighbor characteristics, but also depends on the linkage with other atoms. In addition, icosahedral atoms with a higher degree of medium range five-fold symmetry (MRFFS) are more resistant to the stress-induced deformation. Therefore, the TCP characteristics, including its nearest neighbor feature and the inter-connection between TCP LaSCs, are closely related with the deformation behavior of atoms, especially the MRFFS (up to 5 layers) of icosahedral atoms. These findings shed new light on the understanding of the relationship between microstructure and deformation response of MGs, which will promote the development of deformation theory of MGs.

A Simple Parallel Chaotic Circuit Based on Memristor.

This paper reports a simple parallel chaotic circuit with only four circuit elements: a capacitor, an inductor, a thermistor, and a linear negative resistor. The proposed system was analyzed with MATLAB R2018 through some numerical methods, such as largest Lyapunov exponent spectrum (LLE), phase diagram, Poincaré map, dynamic map, and time-domain waveform. The results revealed 11 kinds of chaotic attractors, 4 kinds of periodic attractors, and some attractive characteristics (such as coexistence attractors and transient transition behaviors). In addition, spectral entropy and sample entropy are adopted to analyze the phenomenon of coexisting attractors. The theoretical analysis and numerical simulation demonstrate that the system has rich dynamic characteristics.

A Hidden Chaotic System with Multiple Attractors.

This paper reports a hidden chaotic system without equilibrium point. The proposed system is studied by the software of MATLAB R2018 through several numerical methods, including Largest Lyapunov exponent, bifurcation diagram, phase diagram, Poincaré map, time-domain waveform, attractive basin and Spectral Entropy. Seven types of attractors are found through altering the system parameters and some interesting characteristics such as coexistence attractors, controllability of chaotic attractor, hyperchaotic behavior and transition behavior are observed. Particularly, the Spectral Entropy algorithm is used to analyze the system and based on the normalized values of Spectral Entropy, the state of the studied system can be identified. Furthermore, the system has been implemented physically to verify the realizability.

Different structural transitions of rapidly supercooled tantalum melt under pressure.

Molecular dynamics (MD) simulations have been performed to study the effects of pressure (P) on the crystallization of tantalum (Ta) under different pressures from [0, 100] GPa. The average potential energy of atoms in the system, the pair distribution function and largest standard cluster analysis (LSCA) have been employed to analyze the structure evolution. It was found that the solidified state at 100 K changes from the complex crystal (ß-Ta) through the body-centered cubic (bcc) crystal (α-Ta) to the hexagonal close-packed (hcp) crystal with increasing pressure. At P ≤ 3 GPa, the favorable state is ß-Ta, which is composed of Z12, Z14 and Z15 atoms, and crystallization starts at the same temperature of crystallization (Tc = 1897 K), while there is a stochastic relationship between the crystallinity and pressure. At P ∈ [3, 57.5] GPa, the melt is always crystallized into rather perfect α-Ta, and Tc is nearly linear to pressure. However, when P > 57.5 GPa, a quite perfect bcc crystal is first formed and then transforms to a hcp crystal via a solid-solid (bcc-hcp) phase transition. Moreover, if the new hcp atoms formed in the bcc stage are arranged in regular grains, the bcc-hcp transition would take a multiple-intermediate-state pathway else, a single-intermediate-state pathway is the possibilty. Additionaly, the parameter δs readily reflects the crystallinity of the ß-Ta, and smaller the value of δs, higher is the crystallinity of the ß-Ta. Finally, during the bcc-hcp transition under high pressure, the volume reduction is due to the rearrangement of atoms rather than the reduction in the atomic radius; a slight increase in the number of nearest neighboring pairs results in a significant increase of the system energy.

Removal of Artifacts from EEG Signals: A Review.

Electroencephalogram (EEG) plays an important role in identifying brain activity and behavior. However, the recorded electrical activity always be contaminated with artifacts and then affect the analysis of EEG signal. Hence, it is essential to develop methods to effectively detect and extract the clean EEG data during encephalogram recordings. Several methods have been proposed to remove artifacts, but the research on artifact removal continues to be an open problem. This paper tends to review the current artifact removal of various contaminations. We first discuss the characteristics of EEG data and the types of different artifacts. Then, a general overview of the state-of-the-art methods and their detail analysis are presented. Lastly, a comparative analysis is provided for choosing a suitable methods according to particular application.

Topologically close-packed characteristic of amorphous tantalum.

The structural evolution of tantalum (Ta) during rapid cooling was investigated by molecular dynamics simulation, in terms of the system energy, the pair distribution function and the largest standard cluster analysis. It was found that the critical cooling rate for vitrification was about R ≥ 0.25 K ps-1, two orders lower than other metals (such as Au, Ag, Al, Zr and Zn) and that the meta-stable σ phase (ß-Ta) not only appears on the pathway from liquid to the stable body-centred cubic crystal, but is also easily obtained at room temperature as a long-lived metastable phase with some probability. The most interesting point is that the liquid, amorphous and ß-Ta phases share a nontrivial structural homology; the intrinsic topologically close-packed (TCP) structures in liquids are inherited and developed in different ways, resulting in amorphous or crystalline solids, respectively. With highly local packing fractions and geometrical incompatibility with the global close-packed (such as hcp, fcc and bcc) crystals, TCP structures inevitably result in structural heterogeneity and favour vitrification. As a superset of icosahedrons, TCP structures are ubiquitous in metallic melts, and just before the onset of crystallization reach their maximal number, which is much bigger in Ta than in other poor-GFA metals; so we argue that the strong forming ability of TCP local structures significantly enhances the glass forming ability of pure metals. These findings open up a new perspective that could have a profound impact on the research into metallic glasses.

The role of liquid-liquid transition in glass formation of CuZr alloys.

Some glass-forming liquids have different liquid phases that have the same composition but different structure, density and entropy. Based on experimental and molecular dynamics simulation, we here report thermodynamic, dynamic, and structural evidence of the liquid-liquid transition (LLT) in ten Cu-Zr glass-forming liquids well above the liquidus temperature. We find that for Cu-Zr alloys, the LLT is beneficial to glass formation, and there is a close relationship between the relative transition strength (RTS) of the LLT and the critical thickness representing the glass forming ability (GFA): the bigger the value of the RTS parameters, the stronger the GFA of Cu-Zr alloys. This work not only uncovers the role of the LLT of melts in the glass formation of solids, but also sheds light on the inheritance of properties of glassy solids from the aspect of the detectable dynamics of high-temperature melts.

A novel crystallization pathway for SiGe alloy rapid cooling.

Understanding the structural evolution of covalent systems under rapid cooling is very important to establish a comprehensive solidification theory. Herein, we conducted molecular dynamics simulations to investigate the crystallization of silicon-germanium (SiGe) alloys. It was found that during crystallization, the saturation and orientation of covalent bonds are satisfied in order, resulting in three phase transitions. The saturation is satisfied during a continuous phase transition that occurs in the super-cooled liquid state. When the orientation was satisfied at the local scale, a novel state, the critical-nuclei crystalline (CNC) phase was obtained, where the local diamond structures increase in number with time and ultimately stabilize at an average size at the critical value. Finally with a coordinated rearrangement of atoms, the orientation is satisfied globally and a stable diamond crystal is produced. For SiGe alloys this CNC phase is universal and rather stable, and the stable temperature range has a certain relationship with the cooling rate and number fraction of atoms. This novel pathway is believed to be universal for such materials including carbon. The CNC state can explain the observation that diamond can be obtained without high pressure. These findings will significantly advance the understanding of the mechanism of phase transition, particularly for covalently bonded materials.

The effect of pressure on the crystallization of rapidly supercooled zirconium melts.

Molecular dynamics simulations have been performed to explore the effect of pressure (P) on the crystallization of zirconium (Zr) under rapid cooling. The structural evolutions have been analysed in terms of the system energy, the pair distribution function and the largest standard cluster analysis. It was found that at the cooling rate of 1.0 × 1011 K s-1, which can crystallize Zr melts into hcp crystals via the bcc intermediate state under zero pressure, the critical pressure (Pc) for vitrification is about 28.75 GPa, and the larger the pressure, the higher the glass transition temperature Tg. At P < Pc the Ostwald's step rule is applied to Zr melts. Crystallization of rapidly super-cooled Zr melts under pressure always begins with the bcc phase and ends in the hcp crystal; the higher the pressure, the lower the onset temperature (Tc) of crystallization. Unlike the single-intermediate-state crystallization (SisC) under zero pressure, multiple-intermediate-state crystallization (MisC) is usually observed under pressure. Structural analysis reveals that if nucleation is essentially completed at the end of the first crystalline (bcc-dominated) stage, MisC will occur; otherwise, SisC occurs. The origin of such an observation is also discussed from the effect of pressure upon the thermodynamics and kinetics factors. These findings are useful for comprehensively understanding the solidification of metals under pressure.

Improving the magnetic moment of Ca2Ge and promoting the conversion of semiconductors to diluted magnetic semiconductors using Mn-doping.

The dilute magnetic properties of materials have important potential applications in the field of electronic science and technology. Intrinsic Ca2Ge is a new environmentally friendly semiconductor material, and exhibits cubic and orthorhombic phases. The crystal structure characteristics of Ca2Ge indicate that the modulation of its dilute magnetic properties can theoretically be achieved by doping with magnetic elements. The study of band structures shows that Ca2Ge is a semiconductor, while Mn doped Ca2Ge is a semi-metal. The results of density of states and atomic population analysis show that Mn doped Ca2Ge exhibits ferrimagnetism with a magnetic moment of 5 µB, and the orbital splitting energy of the Mn atom is 1.0 eV. Mn-doping changes the cubic crystal field of Ca2Ge, and the charge transfer and electron polarization of Ca d and Ge p orbitals are affected by Mn atoms. The Ca d orbital is split into dzz, dzy, dzx, dxx-yy and dxy orbitals, and the contribution of spin of each d split orbital to the magnetic moment of the Ca d orbital is in the order dxy > dyz > dxz > dxx-yy > dzz. The Ge p orbital is split into px, py and pz orbitals, and the spin contribution of each p orbital to the magnetic moment of the Ge p orbital is in the order py > pz > px. The analysis of atom populations shows that the charge transfer and spin of Ca and Ge change with Mn doping, and the difference between spin up and spin down increases, improving the magnetism of Ca2Ge and forming a dilute magnetic semiconductor.

Toward Interpreting the Thermally Activated ß Dynamics in Metallic Glass Using the Structural Constraint Neural Network.

It is crucial to unravel the structural factors influencing the dynamics of the amorphous solids. Deep learning aids in navigating these complexities, while transparency issues persist. Drawing inspiration from the successful application of prototype neural networks in image analysis, this study introduces a novel machine learning approach to address interpretability challenges in glassy research. Distinguishing from traditional machine learning models, the proposed neural network tries to learn distant structural motifs for solid-like atoms and liquid-like atoms. Such learned structural motifs constrain the underlying structural space and thus can serve as a breakthrough in explaining how structural differences impact dynamics. We further used the proposed model to explore the correlation between the local structure and activation energy in the CuZr alloys. Building upon this interpretable model, we demonstrated significant structural differences among atoms with different activation energies. Our interpretable model is a data-driven solution that provides a pathway to reveal the origin of structural heterogeneity in amorphous alloys.

Deep endpoints focusing network under geometric constraints for end-to-end biometric measurement in fetal ultrasound images.

Biometric measurements in fetal ultrasound images are one of the most highly demanding medical image analysis tasks that can directly contribute to diagnosing fetal diseases. However, the natural high-speckle noise and shadows in ultrasound data present big challenges for automatic biometric measurement. Almost all the existing dominant automatic methods are two-stage models, where the key anatomical structures are segmented first and then measured, thus bringing segmentation and fitting errors. What is worse, the results of the second-stage fitting are completely dependent on the good performance of first-stage segmentation, i.e., the segmentation error will lead to a larger fitting error. To this end, we propose a novel end-to-end biometric measurement network, abbreviated as E2EBM-Net, that directly fits the measurement parameters. E2EBM-Net includes a cross-level feature fusion module to extract multi-scale texture information, a hard-soft attention module to improve position sensitivity, and center-focused detectors jointly to achieve accurate localizing and regressing of the measurement endpoints, as well as a loss function with geometric cues to enhance the correlations. To our knowledge, this is the first AI-based application to address the biometric measurement of irregular anatomical structures in fetal ultrasound images with an end-to-end approach. Experiment results showed that E2EBM-Net outperformed the existing methods and achieved the state-of-the-art performance.

Analysis of grain size and five-fold twins during rapid solidification processes inTi3Al alloy.

The wide application of titanium aluminum (Ti-Al) intermetallic compounds for aerospace and automotive fields has accelerated the research and development of Ti3Al alloy. In this study, simulation is adopted to investigate the crystallization behavior during rapid solidification of Ti3Al alloys using molecular dynamics at different cooling rates of 1010 K s-1, 1011 K s-1, 1012 K s-1, and 1013 K s-1. The evolution of microstructures is characterized by taking advantage of the average potential energy, the pair distribution function and visualization. The results show that the system has formed a microstructural configuration with the face-centered cubic structure as the main structure and the hexagonal close-packed structure as the supplement. An increase in the cooling rate will reduce the grain size of the alloy, which in turn will increase the number of grains. At the cooling rate at which the alloy can crystallize, the system forms five-fold twin structures. Meanwhile, we obtain a deeper insight into the properties of five-fold twins in terms of atoms on different sites, and establish a standard model of the same specification for comparison to get the commonality and differences of the five-fold twins between the standard and the solidified. In addition, the evolution of dislocation densities and distribution of dislocation lines in the system under different conditions are analyzed. This study further explores crystallization behavior on the atomic scale and it is hoped that this research will contribute to expanding the understanding of Ti3Al alloy during the growth process.

Atomic structure evolutions and mechanisms of the crystallization pathway of liquid Al during rapid cooling.

The solidification of pure aluminum has been studied by a large-scale molecular dynamic simulation. The potential energy, position D, height H, and width W of the first peak and valley of PDF curves, and the local structures were investigated. It was found that the FCC-crystallization ability of pure Al is so strong that still local crystal regions exist in the amorphized solid. As the temperature decreases, besides the counter-intuitive increase in D p (D of the first peak), H p increases monotonically; W p, D v, and H v decrease monotonically; only W v first decreases and then increases. They all change critically when phase transition happens. After the nucleation, orientation-disordered HCP-regions, as the grain boundaries or defects of FCC crystals, rapidly transform into FCC structures, and then the surviving HCP-regions regularize into few parallel layers or orientation-disordered HCP-regions. If parallel layers result in dislocation pinning, structural evolution terminates; otherwise, it continues. These findings will have a positive impact on the development of the solidification and nucleation theory.

Theoretical study of cellulose II nanocrystals with different exposed facets.

Derived from the most abundant natural polymer, cellulose nanocrystal materials have attracted attention in recent decades due to their chemical and mechanical properties. However, still unclear is the influence of different exposed facets of the cellulose nanocrystals on the physicochemical properties. Herein, we first designed cellulose II nanocrystals with different exposed facets, the hydroxymethyl conformations distribution, hydrogen bond (HB) analysis, as well as the relative structural stability of these models (including crystal facets {A, B, O} and Type-A models vary in size) are theoretically investigated. The results reveal that the HB network of terminal anhydroglucose depends on the adjacent chain's contact sites in nanocrystals exposed with different facets. Compared to nanocrystals exposed with inclined facet, these exposed with flat facet tend to be the most stable. Therefore, the strategy of tuning exposed crystal facets will guide the design of novel cellulose nanocrystals with various physicochemical properties.

Geometric analysis of shape transition for two-layer carbon-silicon nanotubes.

The two-layer nanotubes consisted of carbon atoms on the outside layer and silicon atoms on the inside layer (CNT@SiNT) show a series of diversity in the shape transitions, for instance transforming from a circle through an oval to a rectangle. In this paper, we investigate this geometric change from three perspectives. In the first aspect, we stationary time, followed by quantize in the three-dimensional Z-axis of nanotubes. In the second aspect, we stationary Z-axis, followed by quantize in the time. Finally, we tracked distance of nanotubes flattest section and roundest section. At the stationary time, the overall image of different Z-axis distance distributions is similar to a plan view of multiple ice creams, regardless of whether CNT or SiNT are on the same Z-axis, their slice plans are circle or rectangle of the projection of the Z-axis section on the XOY plane. In the stationary Z-axis, the nanotubes periodically change from a circle to an oval, and then from an oval to a rectangle at different times. Most remarkably, the distance value of deformation which we track the flattest and roundest is a constant value, and in the same distance period, there is only one roundest circle and one longest rectangle at different section and different time. The geometric analysis provided theoretical reference for the preparation of various devices and semiconductor nano-heterojunctions.

Formation and evolution of metastable bcc phase during solidification of liquid Ag: a molecular dynamics simulation study.

On the basis of the quantum Sutton-Chen potential, the rapid solidification processes of liquid silver have been studied by molecular dynamics simulation for four cooling rates. By means of several analysis methods, the competitions and transitions between microstructures during the cooling processes have been analyzed intensively. It is found that there are two phase transitions in all simulation processes. The first one is from liquid state to metastable (transitional) body-centered cubic (bcc) phase. The initial crystallization temperature T(ic) increases with the decrease of the cooling rate. The second one is from the transitional bcc phase to the final solid phase. This study validates the Ostwald's step rule and provides evidence for the prediction that the metastable bcc phase forms first from liquid. Further analyses reveal that the final solid at 273 K can be a mixture of hexagonal close-packed (hcp) and face-centered cubic (fcc) microstructures with various proportions of the two, and the slower the cooling rate is, the higher proportion the fcc structure occupies.

Structural evolutions and hereditary characteristics of icosahedral nano-clusters formed in Mg70Zn30 alloys during rapid solidification processes.

To investigate the structural evolution and hereditary mechanism of icosahedral nano-clusters formed during rapid solidification, a molecular dynamics (MD) simulation study has been performed for a system consisting of 107 atoms of liquid Mg70Zn30 alloy. Adopting Honeycutt-Anderson (HA) bond-type index method and cluster type index method (CTIM-3) to analyse the microstructures in the system it is found that for all the nano-clusters including 2~8 icosahedral clusters in the system, there are 62 kinds of geometrical structures, and those can be classified, by the configurations of the central atoms of basic clusters they contained, into four types: chain-like, triangle-tailed, quadrilateral-tailed and pyramidal-tailed. The evolution of icosahedral nano-clusters can be conducted by perfect heredity and replacement heredity, and the perfect heredity emerges when temperature is slightly less than Tm then increase rapidly and far exceeds the replacement heredity at Tg; while for the replacement heredity, there are three major modes: replaced by triangle (3-atoms), quadrangle (4-atoms) and pentagonal pyramid (6-atoms), rather than by single atom step by step during rapid solidification processes.