FPGA Accelerators for Computing Interatomic Potential-Based Molecular Dynamics Simulation for Gold Nanoparticles:Exploring Different Communication Protocols

Molecular Dynamics(MD)simulation for computing Interatomic Potential(IAP)is a very important High-Performance Computing(HPC)application.MD simulation on particles of experimental relevance takes huge computation time,despite using an expensive high-end server.Heterogeneous computing,a combination of the Field Programmable Gate Array(FPGA)and a computer,is proposed as a solution to compute MD simulation efficiently.In such heterogeneous computation,communication between FPGA and Computer is necessary.One such MD simulation,explained in the paper,is the(Artificial Neural Network)ANN-based IAP computation of gold(Au_(147)&Au_(309))nanoparticles.MD simulation calculates the forces between atoms and the total energy of the chemical system.This work proposes the novel design and implementation of an ANN IAP-based MD simulation for Au_(147)&Au_(309) using communication protocols,such as Universal Asynchronous Receiver-Transmitter(UART)and Ethernet,for communication between the FPGA and the host computer.To improve the latency of MD simulation through heterogeneous computing,Universal Asynchronous Receiver-Transmitter(UART)and Ethernet communication protocols were explored to conduct MD simulation of 50,000 cycles.In this study,computation times of 17.54 and 18.70 h were achieved with UART and Ethernet,respectively,compared to the conventional server time of 29 h for Au_(147) nanoparticles.The results pave the way for the development of a Lab-on-a-chip application.

Modified embedded-atom method interatomic potentials for Mg-Al-Ca and Mg-Al-Zn ternary systems

Al,Ca,and Zn are representative commercial alloying elements for Mg alloys.To investigate the effects of these elements on the deformation and recrystallization behaviors of Mg alloys,we develop interatomic potentials for the Al-Ca,Al-Zn,Mg-Al-Ca and Mg-Al-Zn systems based on the second nearest-neighbor modified embedded-atom method formalism.The developed potentials describe structural,elastic,and thermodynamic properties of compounds and solutions of associated alloy systems in reasonable agreement with experimental data and higher-level calculations.The applicability of these potentials to the present investigation is confirmed by calculating the generalized stacking fault energy for various slip systems and the segregation energy on twin boundaries of the Mg-Al-Ca and Mg-Al-Zn alloys,accompanied with the thermal expansion coefficient and crystal structure maintenance of stable compounds in those alloys.

The Effect of Interatomic Potentials on the Molecular Dynamics Simulation of Nanometric Machining

One of the major tasks in a molecular dynamics （MD） simulation is the selection of adequate potential functions, from which forces are derived. If the potentials do not model the behaviour of the atoms correctly, the results produced from the simulation would be useless. Three popular potentials, namely, Lennard-Jones （L J）, Morse, and embedded-atom method （EAM） potentials, were employed to model copper workpiece and diamond tool in nanometric machining. From the simulation results and further analysis, the EAM potential was found to be the most suitable of the three potentials. This is because it best describes the metallic bonding of the copper atoms; it demonstrated the lowest cutting force variation, and the potential energy is most stable for the EAM.

Construction of embedded-atom-method interatomic potentials for alkaline metals (Li,Na,and K) by lattice inversion

The lattice-inversion embedded-atom-method interatomic potential developed previously by us is extended to alkaline metals including Li,Na,and K.It is found that considering interatomic interactions between neighboring atoms of an appropriate distance is a matter of great significance in constructing accurate embedded-atom-method interatomic potentials,especially for the prediction of surface energy.The lattice-inversion embedded-atom-method interatomic potentials for Li,Na,and K are successfully constructed by taking the fourth-neighbor atoms into consideration.These angular-independent potentials markedly promote the accuracy of predicted surface energies,which agree well with experimental results.In addition,the predicted structural stability,elastic constants,formation and migration energies of vacancy,and activation energy of vacancy diffusion are in good agreement with available experimental data and first-principles calculations,and the equilibrium condition is satisfied.

Interatomic Potentials for NiZr Alloys from Experimental and Ab initio Calculations

We applied an approach to the development of many-body interatomic potentials for NiZr alloys,gaining an improved accuracy and reliability.The functional form of the potential is that of the embedded method,but it has been improved as follows. (1) The database used for the development of the potential includes both experimental data and a large set of energies of different structures of the alloys generated by Fab initio calculations. (2) The optimum parametrization of the potential for the given database is obtained by fitting. Using this approach we developed reliable interatomic potentials for Ni and Zr. The potential accurately reproduces basic equilibrium properties of the alloys.

From Gradient Elasticity to Gradient Interatomic Potentials: The Case-Study of Gradient London Potential

Motivated by the special theory of gradient elasticity (GradEla), a proposal is advanced for extending it to construct gradient models for interatomic potentials, commonly used in atomistic simulations. Our focus is on London’s quantum mechanical potential which is an analytical expression valid until a certain characteristic distance where “attractive” molecular interactions change character and become “repulsive” and cannot be described by the classical form of London’s potential. It turns out that the suggested internal length gradient (ILG) generalization of London’s potential generates both an “attractive” and a “repulsive” branch, and by adjusting the corresponding gradient parameters, the behavior of the empirical Lennard-Jones potentials is theoretically captured.

A deep learning interatomic potential suitable for simulating radiation damage in bulk tungsten

So far, it has been a challenge for existing interatomic potentials to accurately describe a wide range of physical properties and maintain reasonable efficiency. In this work, we develop an interatomic potential for simulating radiation damage in body-centered cubic tungsten by employing deep potential, a neural network-based deep learning model for representing the potential energy surface. The resulting potential predicts a variety of physical properties consistent with first-principles calculations, including phonon spectrum, thermal expansion, generalized stacking fault energies, energetics of free surfaces, point defects, vacancy clusters, and prismatic dislocation loops. Specifically, we investigated the elasticity-related properties of prismatic dislocation loops, i.e., their dipole tensors, relaxation volumes, and elastic interaction energies. This potential is found to predict the maximal elastic interaction energy between two 1/2 <1 1 1> loops better than previous potentials, with a relative error of only 7.6%. The predicted threshold displacement energies are in reasonable agreement with experimental results, with an average of 128 eV. The efficiency of the present potential is also comparable to the tabulated gaussian approximation potentials and modified embedded atom method potentials, meanwhile, can be further accelerated by graphical processing units. Extensive benchmark tests indicate that this potential has a relatively good balance between accuracy, transferability, and efficiency.

Systematic assessment of various universal machine-learning interatomic potentials

Machine-learning interatomic potentials have revolutionized materials modeling at the atomic scale.Thanks to these,it is now indeed possible to perform simulations of ab initio quality over very large time and length scales.More recently,various universal machine-learning models have been proposed as an out-of-box approach avoiding the need to train and validate specific potentials for each particular material of interest.In this paper,we review and evaluate four different universal machine-learning interatomic potentials(uMLIPs),all based on graph neural network architectures which have demonstrated transferability from one chemical system to another.The evaluation procedure relies on data both from a recent verification study of density-functional-theory implementations and from the Materials Project.Through this comprehensive evaluation,we aim to provide guidance to materials scientists in selecting suitable models for their specific research problems,offer recommendations for model selection and optimization,and stimulate discussion on potential areas for improvement in current machinelearning methodologies in materials science.

A continuum thermal stress theory for crystals based on interatomic potentials

This paper presents a new continuum thermal stress theory for crystals based on interatomic potentials.The effect of finite temperature is taken into account via a harmonic model.An EAM potential for copper is adopted in this paper and verified by computing the effect of the temperature on the specific heat,coefficient of thermal expansion and lattice constant.Then we calculate the elastic constants of copper at finite temperature.The calculation results are in good agreement with experimental data.The thermal stress theory is applied to an anisotropic crystal graphite,in which the Brenner potential is employed.Temperature dependence of the thermodynamic properties,lattice constants and thermal strains for graphite is calculated.The calculation results are also in good agreement with experimental data.

Microscopic Phase-Field Simulation for the Influence of Interatomic Potential on the Precipitation Process of Ni_(75)Al_(14)Mo_(11) Alloy

Abstract The process of γ＇ phase precipitating from Ni75Al14MO11 is studied by a computational simulation technique based on microscopic phase-field kinetics model. We studied the phase transformation with the purpose of clarifying the influence of the nearest interatomic potential V Ni-Al （the nearest interatomic potential） on the precipitation process of γ＇ phase. The result demonstrates that there are two kinds of ordered phases, respective Llo and L12 in the early stage, and Llo phase transforms into L12 phase subsequently. For L12 phase, Ni atoms mainly occupy α site （face center positions）, while Al and Mo atoms occupy fl sites （the vertex positions）. When VNi-Al is increased by 10 MeV, the occupation probability of Ni atoms on α sites and Al atoms on β sites are enhanced. Enhanced VNi-Al facilitates clustering and ordering of Al atom, which promotes the formation of the γ＇ phase. At last, the simulation result was discussed by employing the thermodynamic stability.

Evaluation of tungsten interatomic potentials for radiation damage simulations

The structure and properties of materials under neutron irradiation are an important basis in future fusion reactors.In the absence of fusion neutron sources for irradiation experiments,it is increasingly important and urgent to carry out neutron irradiation simulations on fusion reactor materials and then establish complete databases of defect properties and collisional cascades,where the first and foremost step is to select suitable interatomic potentials for atomistic-level simulations.In this work,six typic interatomic potentials for tungsten(W)are evaluated and reviewed systematically for radiation damage simulations.The relative lattice stability and elastic constants of bulk W are considered first with those potentials;then,the properties of point defects and defect clusters at interstitial sites and vacancies are obtained by molecular statics/dynam-ics simulations.The formation energies of interstitial/vacancy clusters,1/2<111>and<100>dislocation loops in W and the threshold displacement energies along different directions are also determined.In addition,the extended defects are further investigated,such as free surfaces and the energy profiles of 1/2<111>{110}and 1/2<111>{112}stacking faults.The current results provide a reference for selecting W potentials to simulate the radiation damage.

Machine-learning-based interatomic potentials for advanced manufacturing

This paper summarizes the progress of machine-learning-based interatomic potentials and their applications in advanced manufacturing.Interatomic potential is essential for classical molecular dynamics.The advancements made in machine learning(ML)have enabled the development of fast interatomic potential with ab initio accuracy.The accelerated atomic simulation can greatly transform the design principle of manufacturing technology.The most widely used supervised and unsupervised ML methods are summarized and compared.Then,the emerging interatomic models based on ML are discussed:Gaussian approximation potential,spectral neighbor analysis potential,deep potential molecular dynamics,SCHNET,hierarchically interacting particle neural network,and fast learning of atomistic rare events.

Machine learning accelerated carbon neutrality research using big data-from predictive models to interatomic potentials

Carbon neutrality has been proposed as a solution for the current severe energy and climate crisis caused by the overuse of fossil fuels, and machine learning(ML) has exhibited excellent performance in accelerating related research owing to its powerful capacity for big data processing. This review presents a detailed overview of ML accelerated carbon neutrality research with a focus on energy management, screening of novel energy materials, and ML interatomic potentials(MLIPs), with illustrations of two selected MLIP algorithms: moment tensor potential(MTP) and neural equivariant interatomic potential(NequIP). We conclude by outlining the important role of ML in accelerating the achievement of carbon neutrality from global-scale energy management, unprecedented screening of advanced energy materials in massive chemical space, to the revolution of atomicscale simulations of MLIPs, which has the bright prospect of applications.

From the Green's Function in Tight-Binding Representation to Interatomic Many-body Potentials

In this paper, we deduce the analytical form of many-body interatomic potentials based on the Green's function in tight-binding representation. The many-body potentials are expressed as the functions of the hopping integrals which are the physical origin of cohesion of atoms. For thesimple case of s-valent system, the inversion of the many-body potentials are discussed in detail by using the lattice inversion method.

Inverse Monte Carlo study on effective interaction potential of Ag-Rh alloy from pair correlation functions

This paper presents an inverse Monte Carlo method to reconstruct pair interaction potential from pair correlation function. This approach adopts an iterative algorithm on interaction potential to fit known pair correlation function by compelling deviations of canonical average to meet with Hamiltonian parameters on a basis of statistical mechanism. The effective interaction potential between particles in liquid Ag Rh alloys has been calculated with the inverse Monte Carlo method. It demonstrates an effective and simple way to obtain the effective potential of complex melt systems.

The ternary Ni–Al–Co embedded-atom-method potential for γ/γ Ni-based single-crystal superalloys: Construction and application

An Ni-AI-Co system embedded-atom-method potential is constructed for the γ（Ni）/γ＇（Ni3A1） superalloy based on experiments and first-principles calculations. The stacking fault energies （SFEs） of the Ni（Co, A1） random solid solutions are calculated as a function of the concentrations of Co and A1. The calculated SFEs decrease with increasing concentrations of Co and A1, which is consistent with the experimental results. The embedding energy term in the present potential has an important influence on the SFEs of the random solid solutions. The cross-slip processes of a screw dislocation in homogenous Ni（Co） solid solutions are simulated using the present potential and the nudged elastic band method. The cross-slip activation energies increase with increasing Co concentration, which implies that the creep resistance of γ（Ni） may be improved by the addition of Co.

Atomic simulations of primary irradiation damage in U–Mo–Xe system

To shed a light on Xe bubble nucleation in U–Mo fuel from the view of primary irradiation damage,a reported U–Mo–Xe potential under the framework of embedded atom method has been modified within the range of short and intermediate atomic distance.The modified potential can better describe the interactions between energetic particles,and can accurately reproduce the threshold displacement energy surface calculated by the first-principles method.Then,molecular dynamics simulations of primary irradiation damage in U–Mo–Xe system have been conducted under different contents.The raise of Xe concentration brings about a remarkable promotion in residual defect quantity and generates bubbles in more overpressured state,which suggests an acceleration of irradiation damage under the accumulation of the fission gas.Meanwhile,the addition of Mo considerably reduces the residual defect count and hinders irradiation-induced Xe diffusion especially at high contents of Xe,corroborating the importance of high Mo content in mitigation of irradiation damage and swelling behavior in U–Mo fuel.In particular,the variation of irradiation damage with respect to contents suggests a necessity of taking into account the influence of local components on defect evolution in mesoscale simulations.

Thermal transport properties of two-dimensional boron dichalcogenides from a first-principles and machine learning approach

The investigation of thermal transport is crucial to the thermal management of modern electronic devices.To obtain the thermal conductivity through solution of the Boltzmann transport equation,calculation of the anharmonic interatomic force constants has a high computational cost based on the current method of single-point density functional theory force calculation.The recent suggested machine learning interatomic potentials(MLIPs)method can avoid these huge computational demands.In this work,we study the thermal conductivity of two-dimensional MoS_(2)-like hexagonal boron dichalcogenides(H-B_(2)VI_(2);V I=S,Se,Te)with a combination of MLIPs and the phonon Boltzmann transport equation.The room-temperature thermal conductivity of H-B_(2)S_(2)can reach up to 336 W·m^(-1)·K^(-1),obviously larger than that of H-B_(2)Se_(2)and H-B_(2)Te_(2).This is mainly due to the difference in phonon group velocity.By substituting the different chalcogen elements in the second sublayer,H-B_(2)VIV I′have lower thermal conductivity than H-B_(2)VI_(2).The room-temperature thermal conductivity of B2STe is only 11%of that of H-B_(2)S_(2).This can be explained by comparing phonon group velocity and phonon relaxation time.The MLIP method is proved to be an efficient method for studying the thermal conductivity of materials,and H-B_(2)S_(2)-based nanodevices have excellent thermal conduction.

Ageing process of pre-precipitation phase in Ni_(0.75)Al_(0.05)Fe_(0.2) alloy based on phase field method

By utilizing phase field method combined with analysis on free energy and interatomic potentials, pre-precipitation phase formation and transformation process of Ni0.75Al0.05Fe0.2 alloy in early precipitation stage during the ageing process under 1 000 K were studied. And free energy, microstructures, compositions and volume fractions of pre-precipitation phase and equilibrium phase were analyzed. The simulation results indicate that nonstoichiometric Llo pre-precipitation phase formed first, and then would gradually transform into L12 equilibrium phase. It is discovered that the phase transformation process was closely related to free energy and interatomic potentials. Additionally, it is revealed that free energy of Llo pre-precipitation phase was higher and interatomic potential was smaller than that of L12 equilibrium phase. Therefore, it is concluded that Llo phase was unstable, and phase transformation would occur to L12 which was more stable.

ON THE CONTINUUM MODELING OF CARBON NANOTUBES

We have recently proposed a nanoscale continuum theory for carbon nanotubes.The theory links continuum analysis with atomistic modeling by incor- porating interatomic potentials and atomic structures of carbon nanotubes directly into the constitutive law.Here we address two main issues involved in setting up the nanoscale continuum theory for carbon nanotubes,namely the multi-body in- teratomic potentials and the lack of centrosymmetry in the nanotube structure.We explain the key ideas behind these issues in establishing a nanoscale continuum theory in terms of interatomic potentials and atomic structures.