Showing 1–50 of 68 results for author: Shapeev, A

Interatomic Interaction Models for Magnetic Materials: Recent Advances

Authors: Tatiana S. Kostiuchenko, Alexander V. Shapeev, Ivan S. Novikov

Abstract: Atomistic modeling is a widely employed theoretical method of computational materials science. It has found particular utility in the study of magnetic materials. Initially, magnetic empirical interatomic potentials or spin-polarized density functional theory (DFT) served as the primary models for describing interatomic interactions in atomistic simulations of magnetic systems. Furthermore, in rec… ▽ More Atomistic modeling is a widely employed theoretical method of computational materials science. It has found particular utility in the study of magnetic materials. Initially, magnetic empirical interatomic potentials or spin-polarized density functional theory (DFT) served as the primary models for describing interatomic interactions in atomistic simulations of magnetic systems. Furthermore, in recent years, a new class of interatomic potentials known as magnetic machine-learning interatomic potentials (magnetic MLIPs) has emerged. These MLIPs combine the computational efficiency, in terms of CPU time, of empirical potentials with the accuracy of DFT calculations. In this review, our focus lies on providing a comprehensive summary of the interatomic interaction models developed specifically for investigating magnetic materials. We also delve into the various problem classes to which these models can be applied. Finally, we offer insights into the future prospects of interatomic interaction model development for the exploration of magnetic materials. △ Less

Submitted 1 July, 2024; v1 submitted 21 May, 2024; originally announced May 2024.

Fitting to magnetic forces improves the reliability of magnetic Moment Tensor Potentials

Authors: Alexey S. Kotykhov, Konstantin Gubaev, Vadim Sotskov, Christian Tantardini, Max Hodapp, Alexander V. Shapeev, Ivan S. Novikov

Abstract: We propose a novel method for fitting machine-learning interatomic potentials with magnetic degrees of freedom, namely, magnetic Moment Tensor Potentials (mMTP). The main feature of the methodology consists in fitting mMTP to magnetic forces (negative derivatives of energies with respect to magnetic moments) derived from spin-polarized density functional theory calculations. We test our method on… ▽ More We propose a novel method for fitting machine-learning interatomic potentials with magnetic degrees of freedom, namely, magnetic Moment Tensor Potentials (mMTP). The main feature of the methodology consists in fitting mMTP to magnetic forces (negative derivatives of energies with respect to magnetic moments) derived from spin-polarized density functional theory calculations. We test our method on the bcc Fe-Al system with different composition. Specifically, we calculate formation energies, equilibrium lattice parameter, and total cell magnetization. Our findings demonstrate a precise match between values calculated with mMTP and those obtained by DFT at zero temperature. Additionally, using molecular dynamics, we estimate the finite temperature lattice parameter and capture the cell expansion as was previously revealed in experiment. We demonstrate that mMTPs fitted to magnetic forces, increase the relaxation reliability, which is the percent of successfully relaxed structures (i.e. with almost zero forces, stresses, and magnetic moments after the optimization of geometry). Eventually, we show that the proposed methodology can provide an accurate and reliable mMTP with reduced number of computationally complex spin-polarized density functional theory calculations. △ Less

Submitted 17 May, 2024; v1 submitted 11 May, 2024; originally announced May 2024.

A moment tensor potential for lattice thermal conductivity calculations of alpha and beta phases of Ga2O3

Authors: Nikita Rybin, Alexander Shapeev

Abstract: Calculations of heat transport in crystalline materials have recently become mainstream, thanks to machine-learned interatomic potentials that allow for significant computational cost reductions while maintaining the accuracy of first-principles calculations. Moment tensor potentials (MTP) are among the most efficient and accurate models in this regard. In this study, we demonstrate the applicatio… ▽ More Calculations of heat transport in crystalline materials have recently become mainstream, thanks to machine-learned interatomic potentials that allow for significant computational cost reductions while maintaining the accuracy of first-principles calculations. Moment tensor potentials (MTP) are among the most efficient and accurate models in this regard. In this study, we demonstrate the application of MTP to the calculation of the lattice thermal conductivity of alpha and beta Ga2O3. Although MTP is commonly employed for lattice thermal conductivity calculations, the advantages of applying the active learning methodology for potential generation is often overlooked. Here, we emphasize its importance and illustrate how it enables the generation of a robust and accurate interatomic potential while maintaining a moderate-sized training dataset. △ Less

Submitted 29 February, 2024; originally announced March 2024.

Thermophysical Properties of Molten FLiNaK: a Moment Tensor Potential Approach

Authors: Rybin Nikita, Dmitrii Maksimov, Yuriy Zaikov, Alexander Shapeev

Abstract: Fluoride salts demonstrate significant potential for applications in next-generation nuclear reactors, necessitating a comprehensive understanding of their thermophysical properties for technological advancements. Experimental measurement of these properties poses challenges, due to factors such as high temperatures, impurity control, and corrosion. Consequently, precise computational modeling met… ▽ More Fluoride salts demonstrate significant potential for applications in next-generation nuclear reactors, necessitating a comprehensive understanding of their thermophysical properties for technological advancements. Experimental measurement of these properties poses challenges, due to factors such as high temperatures, impurity control, and corrosion. Consequently, precise computational modeling methods become essential for predicting the thermophysical properties of molten salts. In this work, we performed molecular dynamics (MD) simulations of several thermophysical properties of the eutectic salt mixture LiF-NaF-KF (FLiNaK) melt, including density, self-diffusion coefficients, viscosity, and thermal conductivity. We demonstrated the successful application of moment tensor potentials (MTP) as an accurate model for interatomic interactions in FLiNaK. Our results on thermophysical properties calculations exhibit strong agreement with experimental data. An important aspect of our methodology is the incorporation of an active learning scheme, which enables the generation of a robust and accurate potential, while maintaining a moderate-sized training dataset. △ Less

Submitted 28 February, 2024; originally announced February 2024.

Mechanical properties of single and polycrystalline solids from machine learning

Authors: Faridun N. Jalolov, Evgeny V. Podryabinkin, Artem R. Oganov, Alexander V. Shapeev, Alexander G. Kvashnin

Abstract: Calculations of elastic and mechanical characteristics of non-crystalline solids are challenging due to high computation cost of $ab$ $initio$ methods and low accuracy of empirical potentials. We propose a computational technique towards efficient calculations of mechanical properties of polycrystals, composites, and multi-phase systems from atomistic simulation with high accuracy and reasonable c… ▽ More Calculations of elastic and mechanical characteristics of non-crystalline solids are challenging due to high computation cost of $ab$ $initio$ methods and low accuracy of empirical potentials. We propose a computational technique towards efficient calculations of mechanical properties of polycrystals, composites, and multi-phase systems from atomistic simulation with high accuracy and reasonable computational cost. It is based on using actively learned machine learning interatomic potentials (MLIPs) trained on a local fragments of the polycrystalline system for which forces, stresses and energies are computed by using $ab$ $initio$ calculations. Developed approach is used for calculation the dependence of elastic moduli of polycrystalline diamond on the grain size. This technique allows one to perform large-scale calculations of mechanical properties of complex solids of various compositions and structures with high accuracy making the transition from ideal (single crystal) systems to more realistic ones. △ Less

Submitted 26 September, 2023; originally announced September 2023.

Bayesian inference of composition-dependent phase diagrams

Authors: Timofei Miryashkin, Olga Klimanova, Vladimir Ladygin, Alexander Shapeev

Abstract: Phase diagrams serve as a highly informative tool for materials design, encapsulating information about the phases that a material can manifest under specific conditions. In this work, we develop a method in which Bayesian inference is employed to combine thermodynamic data from molecular dynamics (MD), melting point simulations, and phonon calculations, process these data, and yield a temperature… ▽ More Phase diagrams serve as a highly informative tool for materials design, encapsulating information about the phases that a material can manifest under specific conditions. In this work, we develop a method in which Bayesian inference is employed to combine thermodynamic data from molecular dynamics (MD), melting point simulations, and phonon calculations, process these data, and yield a temperature-concentration phase diagram. The employed Bayesian framework yields us not only the free energies of different phases as functions of temperature and concentration but also the uncertainties of these free energies originating from statistical errors inherent to finite-length MD trajectories. Furthermore, it extrapolates the results of the finite-atom calculations to the infinite-atom limit and facilitates the choice of temperature, chemical potentials, and the number of atoms conducting the next simulation with which will be the most efficient in reducing the uncertainty of the phase diagram. The developed algorithm was successfully tested on two binary systems, Ge-Si and K-Na, in the full range of concentrations and temperatures. △ Less

Submitted 3 September, 2023; originally announced September 2023.

Accurate melting point prediction through autonomous physics-informed learning

Authors: Olga Klimanova, Timofei Miryashkin, Alexander Shapeev

Abstract: We present an algorithm for computing melting points by autonomously learning from coexistence simulations in the NPT ensemble. Given the interatomic interaction model, the method makes decisions regarding the number of atoms and temperature at which to conduct simulations, and based on the collected data predicts the melting point along with the uncertainty, which can be systematically improved w… ▽ More We present an algorithm for computing melting points by autonomously learning from coexistence simulations in the NPT ensemble. Given the interatomic interaction model, the method makes decisions regarding the number of atoms and temperature at which to conduct simulations, and based on the collected data predicts the melting point along with the uncertainty, which can be systematically improved with more data. We demonstrate how incorporating physical models of the solid-liquid coexistence evolution enhances the algorithm's accuracy and enables optimal decision-making to effectively reduce predictive uncertainty. To validate our approach, we compare the results of 20 melting point calculations from the literature to the results of our calculations, all conducted with same interatomic potentials. Remarkably, we observe significant deviations in about one-third of the cases, underscoring the need for accurate and reliable algorithms for materials property calculations. △ Less

Submitted 13 October, 2023; v1 submitted 23 June, 2023; originally announced June 2023.

Comments: 12 figures

A machine learning potential-based generative algorithm for on-lattice crystal structure prediction

Authors: Vadim Sotskov, Alexander V. Shapeev, Evgeny V. Podryabinkin

Abstract: We propose a method for crystal structure prediction based on a new structure generation algorithm and on-lattice machine learning interatomic potentials. Our algorithm generates the atomic configurations assigning atomic species to sites of the given lattice, and uses cluster expansion or low-rank potential to evaluate their energy. We demonstrate two benefits of such approach. First, our structu… ▽ More We propose a method for crystal structure prediction based on a new structure generation algorithm and on-lattice machine learning interatomic potentials. Our algorithm generates the atomic configurations assigning atomic species to sites of the given lattice, and uses cluster expansion or low-rank potential to evaluate their energy. We demonstrate two benefits of such approach. First, our structure generation algorithm offers a ``smart'' configurational space sampling, targeting low-energy structures which significantly reduces computational costs. Second, the application of machine learning interatomic potentials significantly reduces the number of DFT calculations. We discuss how our algorithm resembles the latent diffusion models for image generation. We demonstrate the efficiency of our method by constructing the convex hull of Nb-Mo-Ta-W system, including binary and ternary Nb-W and Mo-Ta-W subsystems. We found new binary, ternary, and quaternary stable structures that are not reported in the AFLOW database which we choose as our baseline. Due to the computational efficiency of our method we anticipate that it can pave the way towards efficient high-throughput discovery of multicomponent materials. △ Less

Submitted 6 June, 2023; originally announced June 2023.

MLIP-3: Active learning on atomic environments with Moment Tensor Potentials

Authors: Evgeny Podryabinkin, Kamil Garifullin, Alexander Shapeev, Ivan Novikov

Abstract: Nowadays, academic research relies not only on sharing with the academic community the scientific results obtained by research groups while studying certain phenomena, but also on sharing computer codes developed within the community. In the field of atomistic modeling these were software packages for classical atomistic modeling, later -- quantum-mechanical modeling, and now with the fast growth… ▽ More Nowadays, academic research relies not only on sharing with the academic community the scientific results obtained by research groups while studying certain phenomena, but also on sharing computer codes developed within the community. In the field of atomistic modeling these were software packages for classical atomistic modeling, later -- quantum-mechanical modeling, and now with the fast growth of the field of machine-learning potentials, the packages implementing such potentials. In this paper we present the MLIP-3 package for constructing moment tensor potentials and performing their active training. This package builds on the MLIP-2 package (Novikov et al. (2020), The MLIP package: moment tensor potentials with MPI and active learning. Machine Learning: Science and Technology, 2(2), 025002.), however with a number of improvements, including active learning on atomic neighborhoods of a possibly large atomistic simulation. △ Less

Submitted 28 August, 2023; v1 submitted 25 April, 2023; originally announced April 2023.

Equivariant Tensor Networks

Authors: Max Hodapp, Alexander Shapeev

Abstract: Machine-learning interatomic potentials (MLIPs) have made a significant contribution to the recent progress in the fields of computational materials and chemistry due to MLIPs' ability of accurately approximating energy landscapes of quantum-mechanical models while being orders of magnitude more computationally efficient. However, the computational cost and number of parameters of many state-of-th… ▽ More Machine-learning interatomic potentials (MLIPs) have made a significant contribution to the recent progress in the fields of computational materials and chemistry due to MLIPs' ability of accurately approximating energy landscapes of quantum-mechanical models while being orders of magnitude more computationally efficient. However, the computational cost and number of parameters of many state-of-the-art MLIPs increases exponentially with the number of atomic features. Tensor (non-neural) networks, based on low-rank representations of high-dimensional tensors, have been a way to reduce the number of parameters in approximating multidimensional functions, however, it is often not easy to encode the model symmetries into them. In this work we develop a formalism for rank-efficient equivariant tensor networks (ETNs), i.e., tensor networks that remain invariant under actions of SO(3) upon contraction. All the key algorithms of tensor networks like orthogonalization of cores and DMRG carry over to our equivariant case. Moreover, we show that many elements of modern neural network architectures like message passing, pulling, or attention mechanisms, can in some form be implemented into the ETNs. Based on ETNs, we develop a new class of polynomial-based MLIPs that demonstrate superior performance over existing MLIPs for multicomponent systems. △ Less

Submitted 17 April, 2023; originally announced April 2023.

Comments: preprint

Machine-learning Driven Synthesis of TiZrNbHfTaC5 High-Entropy Carbide

Authors: Alexander Ya. Pak, Vadim Sotskov, Arina A. Gumovskaya, Yuliya Z. Vassilyeva, Zhanar S. Bolatova, Yulia A. Kvashnina, Gennady Ya. Mamontov, Alexander V. Shapeev, Alexander G. Kvashnin

Abstract: Synthesis of high-entropy carbides (HEC) requires high temperatures that can be provided by electric arc plasma method. However, the formation temperature of a single-phase sample remains unknown. Moreover, under some temperatures multi-phase structures can emerge. In this work we developed an approach for a controllable synthesis of HEC TiZrNbHfTaC5 based on theoretical and experimental technique… ▽ More Synthesis of high-entropy carbides (HEC) requires high temperatures that can be provided by electric arc plasma method. However, the formation temperature of a single-phase sample remains unknown. Moreover, under some temperatures multi-phase structures can emerge. In this work we developed an approach for a controllable synthesis of HEC TiZrNbHfTaC5 based on theoretical and experimental techniques. We used canonical Monte Carlo (CMC) simulations with the machine learning interatomic potentials to determine the temperature conditions for the formation of single-phase and multi-phase samples. In full agreement with the theory, the single-phase sample, produced with electric arc discharge, was observed at 2000 K. Below 1200 K the sample decomposed into (Ti-Nb-Ta)C and a mixture of (Zr-Hf-Ta)C, (Zr-Nb-Hf)C, (Zr-Nb)C, and (Zr-Ta)C. Our results demonstrate the conditions for the formation of HEC and we anticipate that our approach can pave the way towards targeted synthesis of multicomponent materials. △ Less

Submitted 9 December, 2022; originally announced December 2022.

Comments: 16 pages, 8 figures

Modelling Surface Segregation in Compositionally Complex Alloys with Ab-Initio Accuracy

Authors: Alberto Ferrari, Vadim Sotskov, Alexander V. Shapeev, Fritz Körmann

Abstract: Compositionally complex alloys or concentrated solid solutions are the latest frontier in catalyst design, but mixing different elements in one catalyst may result in surface segregation. Atomistic simulations can predict segregation patterns, but standard approaches based on mean-field models, cluster expansion, or classical interatomic potentials are often limited for the description of multicom… ▽ More Compositionally complex alloys or concentrated solid solutions are the latest frontier in catalyst design, but mixing different elements in one catalyst may result in surface segregation. Atomistic simulations can predict segregation patterns, but standard approaches based on mean-field models, cluster expansion, or classical interatomic potentials are often limited for the description of multicomponent alloys. We present machine learning potentials that can describe surface segregation with near DFT accuracy. The method is used to study a complex Co-Cu-Fe-Mo-Ni quinary alloy. For this alloy, an unexpected segregation of Co, which has a relatively high surface energy, is observed. We rationalize this surprising mechanism in terms of simple transition-metal chemistry. △ Less

Submitted 8 December, 2022; originally announced December 2022.

Comments: 9 pages, 5 figures, 2 tables

AI-accelerated Materials Informatics Method for the Discovery of Ductile Alloys

Authors: Ivan Novikov, Olga Kovalyova, Alexander Shapeev, Max Hodapp

Abstract: In computational materials science, a common means for predicting macroscopic (e.g., mechanical) properties of an alloy is to define a model using combinations of descriptors that depend on some material properties (elastic constants, misfit volumes, etc.), representative for the macroscopic behavior. The material properties are usually computed using special quasi-random structures (SQSs), in tan… ▽ More In computational materials science, a common means for predicting macroscopic (e.g., mechanical) properties of an alloy is to define a model using combinations of descriptors that depend on some material properties (elastic constants, misfit volumes, etc.), representative for the macroscopic behavior. The material properties are usually computed using special quasi-random structures (SQSs), in tandem with density functional theory (DFT). However, DFT scales cubically with the number of atoms and is thus impractical for a screening over many alloy compositions. Here, we present a novel methodology which combines modeling approaches and machine-learning interatomic potentials. Machine-learning interatomic potentials are orders of magnitude faster than DFT, while achieving similar accuracy, allowing for a predictive and tractable high-throughput screening over the whole alloy space. The proposed methodology is illustrated by predicting the room temperature ductility of the medium-entropy alloy Mo-Nb-Ta. △ Less

Submitted 14 October, 2022; originally announced October 2022.

Comments: preprint accepted for publication

Journal ref: Journal of Materials Research (2022)

Towards reliable calculations of thermal rate constants: ring polymer molecular dynamics for the OH + HBr $\to$ Br + H$_2$O reaction

Authors: Ivan S. Novikov, Edgar M. Makarov, Alexander V. Shapeev, Yury V. Suleimanov

Abstract: We combined Moment Tensor Potential (MTP) and Ring Polymer Molecular Dynamics (RPMD) for calculating the thermal rate constants of the OH + HBr system. We used the active learning (AL) algorithm for constructing a training set during RPMD. We compared the obtained RPMD-AL-MTP rate constants with the ones previously calculated using the quasi-classical trajectories (QCT) and the POTLIB potential en… ▽ More We combined Moment Tensor Potential (MTP) and Ring Polymer Molecular Dynamics (RPMD) for calculating the thermal rate constants of the OH + HBr system. We used the active learning (AL) algorithm for constructing a training set during RPMD. We compared the obtained RPMD-AL-MTP rate constants with the ones previously calculated using the quasi-classical trajectories (QCT) and the POTLIB potential energy surface, and with the experimental ones. We demonstrated that the RPMD rate constants were systematically closer to the experimental rate constants than the QCT ones at 200 K, 300 K, and 500 K. △ Less

Submitted 26 June, 2024; v1 submitted 3 June, 2022; originally announced June 2022.

Mechanical, optical, and thermoelectric properties of semiconducting ZnIn2X4 (X= S, Se, Te) monolayers

Authors: Mohammad Ali Mohebpour, Bohayra Mortazavi, Timon Rabczuk, Xiaoying Zhuang, Alexander V. Shapeev, Meysam Bagheri Tagani

Abstract: Mechanical stability of the ZnIn2X4 monolayers. The ZnIn2S4 and ZnIn2Se4 are semiconductors with direct band gaps of 3.94 and 2.77 eV, respectively whereas the ZnIn2Te4 shows an indirect band gap of 1.84 eV at the G0W0 level. The optical properties achieved from the solution of the Bethe-Salpeter equation predict the exciton binding energy of the ZnIn2S4, ZnIn2Se4, and ZnIn2Te4 monolayers to be 0.… ▽ More Mechanical stability of the ZnIn2X4 monolayers. The ZnIn2S4 and ZnIn2Se4 are semiconductors with direct band gaps of 3.94 and 2.77 eV, respectively whereas the ZnIn2Te4 shows an indirect band gap of 1.84 eV at the G0W0 level. The optical properties achieved from the solution of the Bethe-Salpeter equation predict the exciton binding energy of the ZnIn2S4, ZnIn2Se4, and ZnIn2Te4 monolayers to be 0.51, 0.41, and 0.34 eV, respectively, suggesting the high stability of the excitonic states against thermal dissociation. Using the iterative solutions of the Boltzmann transport equation accelerated by machine learning interatomic potentials, the room-temperature lattice thermal conductivity of the ZnIn2S4, ZnIn2Se4, and ZnIn2Te4 monolayers is predicted to be remarkably low as 5.8, 2.0, and 0.4 W/mK, respectively. Due to the low lattice thermal conductivity, high thermopower, and large figure of merit, we propose the ZnIn2Se4 and ZnIn2Te4 monolayers as promising candidates for thermoelectric energy conversion systems. This study provides an extensive vision concerning the intrinsic physical properties of the ZnIn2X4 nanosheets and highlights their characteristics for energy conversion and optoelectronics applications. △ Less

Submitted 26 February, 2022; originally announced February 2022.

Comments: 11 pages, 6 figures

Automated optimization of convergence parameters in plane wave density functional theory calculations via a tensor decomposition-based uncertainty quantification

Authors: Jan Janssen, Edgar Makarov, Tilmann Hickel, Alexander V. Shapeev, Jörg Neugebauer

Abstract: First principles approaches have revolutionized our ability in using computers to predict, explore and design materials. A major advantage commonly associated with these approaches is that they are fully parameter free. However, numerically solving the underlying equations requires to choose a set of convergence parameters. With the advent of high-throughput calculations it becomes exceedingly imp… ▽ More First principles approaches have revolutionized our ability in using computers to predict, explore and design materials. A major advantage commonly associated with these approaches is that they are fully parameter free. However, numerically solving the underlying equations requires to choose a set of convergence parameters. With the advent of high-throughput calculations it becomes exceedingly important to achieve a truly parameter free approach. Utilizing uncertainty quantification (UQ) and tensor decomposition we derive a numerically highly efficient representation of the statistical and systematic error in the multidimensional space of the convergence parameters. Based on this formalism we implement a fully automated approach that requires as input the target accuracy rather than convergence parameters. The performance and robustness of the approach are shown by applying it to a large set of elements crystallizing in a cubic fcc lattice. △ Less

Submitted 7 December, 2021; originally announced December 2021.

Exploring thermal expansion of carbon-based nanosheets by machine-learning interatomic potentials

Authors: Bohayra Mortazavi, Ali Rajabpour, Xiaoying Zhuang, Timon Rabczuk, Alexander V. Shapeev

Abstract: Examination of thermal expansion of two-dimensional (2D) nanomaterials is a challenging theoretical task with either ab-initio or classical molecular dynamics simulations. In this regard, while ab-initio molecular dynamics (AIMD) simulations offer extremely accurate predictions, but they are excessively demanding from computational point of view. On the other side, classical molecular dynamics sim… ▽ More Examination of thermal expansion of two-dimensional (2D) nanomaterials is a challenging theoretical task with either ab-initio or classical molecular dynamics simulations. In this regard, while ab-initio molecular dynamics (AIMD) simulations offer extremely accurate predictions, but they are excessively demanding from computational point of view. On the other side, classical molecular dynamics simulations can be conducted with affordable computational costs, but without predictive accuracy needed to study novel materials and compositions. Herein, we explore the thermal expansion of several carbon-based nanosheets on the basis of machine-learning interatomic potentials (MLIPs). We show that passively trained MLIPs over inexpensive AIMD trajectories enable the examination of thermal expansion of complex nanomembranes over wide range of temperatures. Passively fitted MLIPs could also with outstanding accuracy reproduce the phonon dispersion relations predicted by density functional theory calculations. Our results highlight that the devised methodology on the basis of passively trained MLIPs is computationally efficient and versatile to accurately examine the thermal expansion of complex and novel materials and compositions using the molecular dynamics simulations. △ Less

Submitted 20 October, 2021; originally announced October 2021.

Journal ref: Carbon 2022

Validation of moment tensor potentials for fcc and bcc metals using EXAFS spectra

Authors: Alexander V. Shapeev, Dmitry Bocharov, Alexei Kuzmin

Abstract: Machine-learning potentials for materials, namely the moment tensor potentials (MTPs), were validated using experimental EXAFS spectra for the first time. The MTPs for four metals (bcc W and Mo, fcc Cu and Ni) were obtained by the active learning algorithm of fitting to the results of the calculations using density functional theory (DFT). The MTP accuracy was assessed by comparing metal K-edge EX… ▽ More Machine-learning potentials for materials, namely the moment tensor potentials (MTPs), were validated using experimental EXAFS spectra for the first time. The MTPs for four metals (bcc W and Mo, fcc Cu and Ni) were obtained by the active learning algorithm of fitting to the results of the calculations using density functional theory (DFT). The MTP accuracy was assessed by comparing metal K-edge EXAFS spectra obtained experimentally and computed from the results of molecular dynamics (MD) simulations. The sensitivity of the method to various aspects of the MD and DFT models was demonstrated using Ni as an example. Good agreement was found for W, Mo and Cu using the recommended PAW pseudopotentials, whereas a more accurate pseudopotential with 18 valence electrons was required for Ni to achieve a similar agreement. The use of EXAFS spectra allows one to estimate the MTP ability in reproducing both average and dynamic atomic structures. △ Less

Submitted 11 August, 2021; originally announced August 2021.

Journal ref: Computational Materials Science 210 (2022) 111028

Machine-learning potentials enable predictive $\textit{and}$ tractable high-throughput screening of random alloys

Authors: Max Hodapp, Alexander Shapeev

Abstract: We present an automated procedure for computing stacking fault energies in random alloys from large-scale simulations using moment tensor potentials (MTPs) with the accuracy of density functional theory (DFT). To that end, we develop an algorithm for training MTPs on random alloys. In the first step, our algorithm constructs a set of ~10000 or more training candidate configurations with 50-100 ato… ▽ More We present an automated procedure for computing stacking fault energies in random alloys from large-scale simulations using moment tensor potentials (MTPs) with the accuracy of density functional theory (DFT). To that end, we develop an algorithm for training MTPs on random alloys. In the first step, our algorithm constructs a set of ~10000 or more training candidate configurations with 50-100 atoms that are representative for the atomic neighborhoods occurring in the large-scale simulation. In the second step, we use active learning to reduce this set to ~100 most distinct configurations - for which DFT energies and forces are computed and on which the potential is ultimately trained. We validate our algorithm for the MoNbTa medium-entropy alloy by showing that the MTP reproduces the DFT $\frac{1}{4}[111]$ unstable stacking fault energy over the entire compositional space up to a few percent. Contrary to state-of-the-art methods, e.g., the coherent potential approximation (CPA) or special quasi-random structures (SQSs), our algorithm naturally accounts for relaxation, is not limited by DFT cell sizes, and opens opportunities to efficiently investigate follow-up problems, such as chemical ordering. In a broader sense, our algorithm can be easily modified to compute related properties of random alloys, for instance, misfit volumes, or grain boundary energies. Moreover, it forms the basis for an efficient construction of MTPs to be used in large-scale simulations of multicomponent systems. △ Less

Submitted 19 October, 2021; v1 submitted 12 July, 2021; originally announced July 2021.

Comments: contains some additional validation of the training algorithm and the potentials

Journal ref: Phys. Rev. Materials 5, 113802 (2021)

Bayesian learning of thermodynamic integration and numerical convergence for accurate phase diagrams

Authors: V. Ladygin, I. Beniya, E. Makarov, A. Shapeev

Abstract: Accurate phase diagram calculation from molecular dynamics requires systematic treatment and convergence of statistical averages. In this work we propose a Gaussian process regression based framework for reconstructing the free energy functions using data of various origin. Our framework allows for propagating statistical uncertainty from finite molecular dynamics trajectories to the phase diagram… ▽ More Accurate phase diagram calculation from molecular dynamics requires systematic treatment and convergence of statistical averages. In this work we propose a Gaussian process regression based framework for reconstructing the free energy functions using data of various origin. Our framework allows for propagating statistical uncertainty from finite molecular dynamics trajectories to the phase diagram and automatically performing convergence with respect to simulation parameters. Furthermore, our approach provides a way for automatic optimal sampling in the simulation parameter space based on Bayesian optimization approach. We validate our methodology by constructing phase diagrams of two model systems, the Lennard-Jones and soft-core potential, and compare the results with existing works studies and our coexistence simulations. Finally, we construct the phase diagram of lithium at temperatures above 300 K and pressures below 30 GPa from a machine-learning potential trained on ab initio data. Our approach performs well when compared to coexistence simulations and experimental results. △ Less

Submitted 1 November, 2021; v1 submitted 19 April, 2021; originally announced April 2021.

MSC Class: 37C20 (Primary); 60G15; 62-04 (Secondary) ACM Class: J.2; G.3; I.6.5

Journal ref: Phys. Rev. B 104, 104102 (2021)

Free energy of (CoxMn1-x)3O4 mixed phases from machine-learning-enhanced ab initio calculations

Authors: Suzanne K. Wallace, Anton S. Bochkarev, Ambroise van Roekeghem, Javier Carrasco, Alexander Shapeev, Natalio Mingo

Abstract: (CoxMn1-x)3O4 is a promising candidate material for solar thermochemical energy storage. A high-temperature model for this system would provide a valuable tool for evaluating its potential. However, predicting phase diagrams of complex systems with ab initio calculations is challenging due to the varied sources affecting the free energy, and with the prohibitive amount of configurations needed in… ▽ More (CoxMn1-x)3O4 is a promising candidate material for solar thermochemical energy storage. A high-temperature model for this system would provide a valuable tool for evaluating its potential. However, predicting phase diagrams of complex systems with ab initio calculations is challenging due to the varied sources affecting the free energy, and with the prohibitive amount of configurations needed in the configurational entropy calculation. In this work, we compare three different machine learning (ML) approaches for sampling the configuration space of (CoxMn1-x)3O4, including a simpler ML approach, which would be suitable for application in high-throughput studies. We use experimental data for a feature of the phase diagram to assess the accuracy of model predictions. We find that with some methods, data pre-treatment is needed to obtain accurate predictions due to inherently composition-imbalanced training data for a mixed phase. We highlight that the important entropy contributions depend on the physical regimes of the system under investigation and that energy predictions with ML models are more challenging at compositions where there are energetically competing ground state crystal structures. Similar methods to those outlined here can be used to screen other candidate materials for thermochemical energy storage △ Less

Submitted 6 March, 2021; originally announced March 2021.

Journal ref: Phys. Rev. Materials 5, 035402, 2021

Exceptional piezoelectricity, high thermal conductivity and stiffness and promising photocatalysis in two-dimensional MoSi2N4 family confirmed by first-principles

Authors: Bohayra Mortazavi, Brahmanandam Javvaji, Fazel Shojaei, Timon Rabczuk, Alexander V. Shapeev, Xiaoying Zhuang

Abstract: Chemical vapor deposition has been most recently employed to fabricate centimeter-scale high-quality single-layer MoSi2N4 (Science; 2020;369; 670). Motivated by this exciting experimental advance, herein we conduct extensive first-principles based simulations to explore the stability, mechanical properties, lattice thermal conductivity, piezoelectric and flexoelectric response, and photocatalytic… ▽ More Chemical vapor deposition has been most recently employed to fabricate centimeter-scale high-quality single-layer MoSi2N4 (Science; 2020;369; 670). Motivated by this exciting experimental advance, herein we conduct extensive first-principles based simulations to explore the stability, mechanical properties, lattice thermal conductivity, piezoelectric and flexoelectric response, and photocatalytic and electronic features of MA2Z4 (M = Cr, Mo, W; A = Si, Ge; Z = N, P) monolayers. The considered nanosheets are found to exhibit dynamical stability and remarkably high mechanical properties. Moreover, they show diverse electronic properties from antiferromagnetic metal to half metal and to semiconductors with band gaps ranging from 0.31 to 2.57 eV. Among the studied nanosheets, the MoSi2N4 and WSi2N4 monolayers yield appropriate band edge positions, high electron and hole mobilities, and strong visible light absorption, highly promising for applications in optoelectronics and photocatalytic water splitting. The MoSi2N4 and WSi2N4 monolayers are also predicted to show outstandingly high lattice thermal conductivity of 440 and 500 W/mK, respectively. For the first time we show that machine learning interatomic potentials trained over small supercells can be employed to examine the flexoelectric and piezoelectric properties of complex structures. As the most exciting finding, WSi2N4, CrSi2N4 and MoSi2N4 are found to exhibit the highest piezoelectric coefficients, outperforming all other-known 2D materials. Our results highlight that MA2Z4 nanosheets not only undoubtedly outperform the transition metal dichalcogenides family but also can compete with graphene for applications in nanoelectronics, optoelectronic, energy storage/conversion and thermal management systems. △ Less

Submitted 29 December, 2020; originally announced December 2020.

Journal ref: Nano Energy 2021

Assessing Parameters for Ring Polymer Molecular Dynamics Simulations at Low Temperatures: DH+H Chemical Reaction

Authors: Ivan S. Novikov, Yury V. Suleimanov, Alexander V. Shapeev

Abstract: Ring polymer molecular dynamics (RPMD) is an accurate method for calculating thermal chemical reaction rates. It has recently been discovered that low-temperature calculations are strongly affected by the simulation parameters. Here, for the thermally activated reaction DH + H -> D + H2, we calculate the RPMD rate constants at T = 50, 100, and 300 K and demonstrate that for T >= 100 K the standard… ▽ More Ring polymer molecular dynamics (RPMD) is an accurate method for calculating thermal chemical reaction rates. It has recently been discovered that low-temperature calculations are strongly affected by the simulation parameters. Here, for the thermally activated reaction DH + H -> D + H2, we calculate the RPMD rate constants at T = 50, 100, and 300 K and demonstrate that for T >= 100 K the standard input parameters yield accurate results, but at low temperatures (e.g., 50 K) one must increase the asymptotic distance and force constant, and decrease the umbrella integration step. △ Less

Submitted 31 March, 2021; v1 submitted 28 December, 2020; originally announced December 2020.

Magnetic Moment Tensor Potentials for collinear spin-polarized materials reproduce different magnetic states of bcc Fe

Authors: Ivan Novikov, Blazej Grabowski, Fritz Kormann, Alexander Shapeev

Abstract: We present the magnetic Moment Tensor Potentials (mMTPs), a class of machine-learning interatomic potentials, accurately reproducing both vibrational and magnetic degrees of freedom as provided, e.g., from first-principles calculations. The accuracy is achieved by a two-step minimization scheme that coarse-grains the atomic and the spin space. The performance of the mMTPs is demonstrated for the p… ▽ More We present the magnetic Moment Tensor Potentials (mMTPs), a class of machine-learning interatomic potentials, accurately reproducing both vibrational and magnetic degrees of freedom as provided, e.g., from first-principles calculations. The accuracy is achieved by a two-step minimization scheme that coarse-grains the atomic and the spin space. The performance of the mMTPs is demonstrated for the prototype magnetic system bcc iron, with applications to phonon calculations for different magnetic states, and molecular dynamics simulations with fluctuating magnetic moments. △ Less

Submitted 10 December, 2021; v1 submitted 23 December, 2020; originally announced December 2020.

Accelerating first-principles estimation of thermal conductivity by machine-learning interatomic potentials: A MTP/ShengBTE solution

Authors: Bohayra Mortazavi, Evgeny P. Podryabinkin, Ivan S. Nvikovb, Timon Rabczuk, Xiaoying Zhuang, Alexander V. Shapeev

Abstract: Accurate evaluation of the thermal conductivity of a material can be a challenging task from both experimental and theoretical points of view. In particular for the nanostructured materials, the experimental measurement of thermal conductivity is associated with diverse sources of uncertainty. As a viable alternative to experiment, the combination of density functional theory (DFT) simulations and… ▽ More Accurate evaluation of the thermal conductivity of a material can be a challenging task from both experimental and theoretical points of view. In particular for the nanostructured materials, the experimental measurement of thermal conductivity is associated with diverse sources of uncertainty. As a viable alternative to experiment, the combination of density functional theory (DFT) simulations and the solution of Boltzmann transport equation is currently considered as the most trusted approach to examine thermal conductivity. The main bottleneck of the aforementioned method is to acquire the anharmonic interatomic force constants using the computationally demanding DFT calculations. In this work we propose a substantially accelerated approach for the evaluation of anharmonic interatomic force constants via employing machine-learning interatomic potentials (MLIPs) trained over short ab-initio molecular dynamics trajectories. The remarkable accuracy of the proposed accelerated method is confirmed by comparing the estimated thermal conductivities of several bulk and two-dimensional materials with those computed by the full-DFT approach. The MLIP-based method proposed in this study can be employed as a standard tool, which would substantially accelerate and facilitate the estimation of lattice thermal conductivity in comparison with the commonly used full-DFT solution. △ Less

Submitted 6 September, 2020; originally announced September 2020.

Comments: arXiv admin note: text overlap with arXiv:2006.06794

Journal ref: Computer Physics Communications 2021

The MLIP package: Moment Tensor Potentials with MPI and Active Learning

Authors: Ivan S. Novikov, Konstantin Gubaev, Evgeny V. Podryabinkin, Alexander V. Shapeev

Abstract: The subject of this paper is the technology (the "how") of constructing machine-learning interatomic potentials, rather than science (the "what" and "why") of atomistic simulations using machine-learning potentials. Namely, we illustrate how to construct moment tensor potentials using active learning as implemented in the MLIP package, focusing on the efficient ways to sample configurations for th… ▽ More The subject of this paper is the technology (the "how") of constructing machine-learning interatomic potentials, rather than science (the "what" and "why") of atomistic simulations using machine-learning potentials. Namely, we illustrate how to construct moment tensor potentials using active learning as implemented in the MLIP package, focusing on the efficient ways to sample configurations for the training set, how expanding the training set changes the error of predictions, how to set up ab initio calculations in a cost-effective manner, etc. The MLIP package (short for Machine-Learning Interatomic Potentials) is available at https://mlip.skoltech.ru/download/. △ Less

Submitted 16 July, 2020; originally announced July 2020.

Machine-Learning Interatomic Potentials Enable First-Principles Multiscale Modeling of Lattice Thermal Conductivity in Graphene/Borophene Heterostructures

Authors: Bohayra Mortazavi, Evgeny V. Podryabinkin, Stephan Roche, Timon Rabczuk, Xiaoying Zhuang, Alexander V. Shapeev

Abstract: One of the ultimate goals of computational modeling in condensed matter is to be able to accurately compute materials properties with minimal empirical information. First-principles approaches such as the density functional theory (DFT) provide the best possible accuracy on electronic properties but they are limited to systems up to a few hundreds, or at most thousands of atoms. On the other hand,… ▽ More One of the ultimate goals of computational modeling in condensed matter is to be able to accurately compute materials properties with minimal empirical information. First-principles approaches such as the density functional theory (DFT) provide the best possible accuracy on electronic properties but they are limited to systems up to a few hundreds, or at most thousands of atoms. On the other hand, classical molecular dynamics (CMD) simulations and finite element method (FEM) are extensively employed to study larger and more realistic systems, but conversely depend on empirical information. Here, we show that machine-learning interatomic potentials (MLIPs) trained over short ab-initio molecular dynamics trajectories enable first-principles multiscale modeling, in which DFT simulations can be hierarchically bridged to efficiently simulate macroscopic structures. As a case study, we analyze the lattice thermal conductivity of coplanar graphene/borophene heterostructures, recently synthesized experimentally (Sci. Adv. 2019; 5: eaax6444), for which no viable classical modeling alternative is presently available. Our MLIP-based approach can efficiently predict the lattice thermal conductivity of graphene and borophene pristine phases, the thermal conductance of complex graphene/borophene interfaces and subsequently enable the study of effective thermal transport along the heterostructures at continuum level. This work highlights that MLIPs can be effectively and conveniently employed to enable first-principles multiscale modeling via hierarchical employment of DFT/CMD/FEM simulations, thus expanding the capability for computational design of novel nanostructures. △ Less

Submitted 11 June, 2020; originally announced June 2020.

Journal ref: Materials Horizons 2020

High thermal conductivity in semiconducting Janus and non-Janus diamanes

Authors: Mostafa Raeisi, Bohayra Mortazavi, Evgeny V. Podryabinkin, Fazel Shojaei, Xiaoying Zhuang, Alexander V. Shapeev

Abstract: Most recently, F-diamane monolayer was experimentally realized by the fluorination of bilayer graphene. In this work we elaborately explore the electronic and thermal conductivity responses of diamane lattices with homo or hetero functional groups, including: non-Janus C2H, C2F and C2Cl diamane and Janus counterparts of C4HF, C4HCl and C4FCl. Noticeably, C2H, C2F, C2Cl, C4HF, C4HCl and C4FCl diama… ▽ More Most recently, F-diamane monolayer was experimentally realized by the fluorination of bilayer graphene. In this work we elaborately explore the electronic and thermal conductivity responses of diamane lattices with homo or hetero functional groups, including: non-Janus C2H, C2F and C2Cl diamane and Janus counterparts of C4HF, C4HCl and C4FCl. Noticeably, C2H, C2F, C2Cl, C4HF, C4HCl and C4FCl diamanes are found to show electronic diverse band gaps of, 3.86, 5.68, 2.42, 4.17, 0.86, and 2.05 eV, on the basis of HSE06 method estimations. The thermal conductivity of diamane nanosheets was acquired using the full iterative solutions of the Boltzmann transport equation, with substantially accelerated calculations by employing machine-learning interatomic potentials in obtaining the anharmonic force constants. According to our results, the room temperature lattice thermal conductivity of graphene and C2H, C2F, C2Cl, C4HF, C4HCl and C4FCl diamane monolayers are estimated to be 3636, 1145, 377, 146, 454, 244 and 196 W/mK, respectively. The underlying mechanisms resulting in significant effects of functional groups on the thermal conductivity of diamane nanosheets were thoroughly explored. Our results highlight the substantial role of functional groups on the electronic and thermal conduction responses of diamane nanosheets. △ Less

Submitted 7 June, 2020; originally announced June 2020.

Journal ref: Carbon 2020

Nanoporous C3N4, C3N5 and C3N6 nanosheets; Novel strong semiconductors with low thermal conductivities and appealing optical/electronic properties

Authors: Bohayra Mortazavi, Fazel Shojaei, Masoud Shahrokhi, Maryam Azizi, Timon Rabczuk, Alexander V. Shapeev, Xiaoying Zhuang

Abstract: Carbon nitride two-dimensional (2D) materials are among the most attractive class of nanomaterials, with wide range of application prospects. As a continuous progress, most recently, two novel carbon nitride 2D lattices of C3N5 and C3N4 have been successfully experimentally realized. Motivated by these latest accomplishments and also by taking into account the well-known C3N4 triazine-based graphi… ▽ More Carbon nitride two-dimensional (2D) materials are among the most attractive class of nanomaterials, with wide range of application prospects. As a continuous progress, most recently, two novel carbon nitride 2D lattices of C3N5 and C3N4 have been successfully experimentally realized. Motivated by these latest accomplishments and also by taking into account the well-known C3N4 triazine-based graphitic carbon nitride structures, we predicted two novel C3N6 and C3N4 counterparts. We then conducted extensive density functional theory simulations to explore the thermal stability, mechanical, electronic and optical properties of these novel nanoporous carbon-nitride nanosheets. According to our results all studied nanosheets were found to exhibit desirable thermal stability and mechanical properties. Non-equilibrium molecular dynamics simulations on the basis of machine learning interatomic potentials predict ultralow thermal conductivities for these novel nanosheets. Electronic structure analyses confirm direct band gap semiconducting electronic character and optical calculations reveal the ability of these novel 2D systems to adsorb visible range of light. Extensive first-principles based results by this study provide a comprehensive vision on the stability, mechanical, electronic and optical responses of C3N4, C3N5 and C3N6 as novel 2D semiconductors and suggest them as promising candidates for the design of advanced nanoelectronics and energy storage/conversion systems. △ Less

Submitted 6 June, 2020; originally announced June 2020.

Journal ref: Carbon 2020

Exploring Phononic Properties of Two-Dimensional Materials using Machine Learning Interatomic Potentials

Authors: Bohayra Mortazavi, Ivan S. Novikov, Evgeny V. Podryabinkin, Stephan Roche, Timon Rabczuk, Alexander V. Shapeev, Xiaoying Zhuang

Abstract: Phononic properties are commonly studied by calculating force constants using the density functional theory (DFT) simulations. Although DFT simulations offer accurate estimations of phonon dispersion relations or thermal properties, but for low-symmetry and nanoporous structures the computational cost quickly becomes very demanding. Moreover, the computational setups may yield nonphysical imaginar… ▽ More Phononic properties are commonly studied by calculating force constants using the density functional theory (DFT) simulations. Although DFT simulations offer accurate estimations of phonon dispersion relations or thermal properties, but for low-symmetry and nanoporous structures the computational cost quickly becomes very demanding. Moreover, the computational setups may yield nonphysical imaginary frequencies in the phonon dispersion curves, impeding the assessment of phononic properties and the dynamical stability of the considered system. Here, we compute phonon dispersion relations and examine the dynamical stability of a large ensemble of novel materials and compositions. We propose a fast and convenient alternative to DFT simulations which derived from machine-learning interatomic potentials passively trained over computationally efficient ab-initio molecular dynamics trajectories. Our results for diverse two-dimensional (2D) nanomaterials confirm that the proposed computational strategy can reproduce fundamental thermal properties in close agreement with those obtained via the DFT approach. The presented method offers a stable, efficient, and convenient solution for the examination of dynamical stability and exploring the phononic properties of low-symmetry and porous 2D materials. △ Less

Submitted 11 May, 2020; originally announced May 2020.

Journal ref: Applied Materials Today, 2020

Elinvar effect in $β-$Ti simulated by on-the-fly trained moment tensor potential

Authors: Alexander V. Shapeev, Evgeny V. Podryabinkin, Konstantin Gubaev, Ferenc Tasnádi, Igor A. Abrikosov

Abstract: A combination of quantum mechanics calculations with machine learning (ML) techniques can lead to a paradigm shift in our ability to predict materials properties from first principles. Here we show that on-the-fly training of an interatomic potential described through moment tensors provides the same accuracy as state-of-the-art {\it ab inito} molecular dynamics in predicting high-temperature elas… ▽ More A combination of quantum mechanics calculations with machine learning (ML) techniques can lead to a paradigm shift in our ability to predict materials properties from first principles. Here we show that on-the-fly training of an interatomic potential described through moment tensors provides the same accuracy as state-of-the-art {\it ab inito} molecular dynamics in predicting high-temperature elastic properties of materials with two orders of magnitude less computational effort. Using the technique, we investigate high-temperature bcc phase of titanium and predict very weak, Elinvar, temperature dependence of its elastic moduli, similar to the behavior of the so-called GUM Ti-based alloys [T. Sato {\ it et al.}, Science {\bf 300}, 464 (2003)]. Given the fact that GUM alloys have complex chemical compositions and operate at room temperature, Elinvar properties of elemental bcc-Ti observed in the wide temperature interval 1100--1700 K is unique. △ Less

Submitted 9 May, 2020; originally announced May 2020.

Comments: 15 pages, 4 figures

In operando active learning of interatomic interaction during large-scale simulations

Authors: Max Hodapp, Alexander Shapeev

Abstract: A well-known drawback of state-of-the-art machine-learning interatomic potentials is their poor ability to extrapolate beyond the training domain. For small-scale problems with tens to hundreds of atoms this can be solved by using active learning which is able to select atomic configurations on which a potential attempts extrapolation and add them to the ab initio-computed training set. In this se… ▽ More A well-known drawback of state-of-the-art machine-learning interatomic potentials is their poor ability to extrapolate beyond the training domain. For small-scale problems with tens to hundreds of atoms this can be solved by using active learning which is able to select atomic configurations on which a potential attempts extrapolation and add them to the ab initio-computed training set. In this sense an active learning algorithm can be viewed as an on-the-fly interpolation of an ab initio model. For large-scale problems, possibly involving tens of thousands of atoms, this is not feasible because one cannot afford even a single density functional theory computation with such a large number of atoms. This work marks a new milestone toward fully automatic ab initio-accurate large-scale atomistic simulations. We develop an active learning algorithm that identifies local subregions of the simulation region where the potential extrapolates. Then the algorithm constructs periodic configurations out of these local, non-periodic subregions, sufficiently small to be computable with plane-wave density functional theory codes, in order to obtain accurate ab initio energies. We benchmark our algorithm on the problem of the screw dislocation motion in bcc tungsten and show that our algorithm reaches ab initio accuracy, down to typical magnitudes of numerical noise in DFT codes. We show that our algorithm reproduces material properties such as core structure, Peierls barrier, and Peierls stress. This unleashes new capabilities for computational materials science toward applications which have currently been out of scope if approached solely by ab initio methods. △ Less

Submitted 21 September, 2020; v1 submitted 27 April, 2020; originally announced April 2020.

Comments: Published version

Journal ref: Mach. Learn.: Sci. Technol. 1 045005 (2020)

Ab initio analysis of some Ge-based 2D nanomaterials

Authors: Ali Ghojavand, S. Javad Hashemifar, Mahdi Tarighi Ahmadpour, Alexander V. Shapeev, Amir Alhaji, Qaem Hassanzada

Abstract: The structural, electronic and dynamical properties of a group of 2D germanium-based compounds, including GeC, GeN, GeO, GeSi, GeS, GeSe, and germanene, are investigated by employing first-principles calculations. The most stable structure of each of these systems is identified after considering the most probable configurations and performing accurate phonon calculations. We introduce a new phase… ▽ More The structural, electronic and dynamical properties of a group of 2D germanium-based compounds, including GeC, GeN, GeO, GeSi, GeS, GeSe, and germanene, are investigated by employing first-principles calculations. The most stable structure of each of these systems is identified after considering the most probable configurations and performing accurate phonon calculations. We introduce a new phase of germanene, which we name the tile germanene, which is significantly more stable than the known hexagonal germanene. We apply the modern modified Becke-Johnson (mBJ) and DFT1/2 schemes to obtain an accurate band structure for our selected 2D materials. It is seen that GeO and GeC exhibit the highest band gaps of more than 3 eV in this group of materials. Moreover, we argue that, in contrast to the semi-metallic nature of hexagonal germanene, the tile germanene is a very good conductor. The band edges of our semiconducting 2D materials are accurately aligned to the vacuum level to address the potential photocatalytic application of this system for water splitting and carbon dioxide reduction. The optical properties, including dielectric functions, refractive index, reflectivity, and Loss function of the samples are investigated in the framework of the Bethe-Salpeter approach. △ Less

Submitted 23 January, 2020; v1 submitted 21 January, 2020; originally announced January 2020.

Comments: 7 pages, 6 figures, 4 tables

Ring Polymer Molecular Dynamics and Active Learning of Moment Tensor Potential for Gas-Phase Barrierless Reactions: Application to S + H2

Authors: Ivan S. Novikov, Alexander V. Shapeev, Yury V. Suleimanov

Abstract: Ring polymer molecular dynamics (RPMD) has proven to be an accurate approach for calculating thermal rate coefficients of various chemical reactions. For wider application of this methodology, efficient ways to generate the underlying full-dimensional potential energy surfaces (PESs) and the corresponding energy gradients are required. Recently, we have proposed a fully automated procedure based o… ▽ More Ring polymer molecular dynamics (RPMD) has proven to be an accurate approach for calculating thermal rate coefficients of various chemical reactions. For wider application of this methodology, efficient ways to generate the underlying full-dimensional potential energy surfaces (PESs) and the corresponding energy gradients are required. Recently, we have proposed a fully automated procedure based on combining the original RPMDrate code with active learning for PES on-the-fly using moment tensor potential and successfully applied it to two representative thermally activated chemical reactions [I. S. Novikov, Y. V. Suleimanov, A. V. Shapeev, Phys. Chem. Chem. Phys., 29503-29512 (2018)]. In this work, using a prototype insertion chemical reaction S + H$_2$, we show that this procedure works equally well for another class of chemical reactions. We find that the corresponding PES can be generated by fitting to less than 1500 automatically generated structures while the RPMD rate coefficients show deviation from the reference values within the typical convergence error of RPMDrate. We note that more structures are accumulated during the real-time propagation of the dynamic factor (the recrossing factor) as opposed to the previous study. We also observe that relatively flat free energy profile of the along the reaction coordinate before entering the complex-formation well can cause issues with locating the maximum of the free energy surface for less converged PESs. However, the final RPMD rate coefficient is independent of the position of the dividing surface that makes it invulnerable to this problem, kee** the total number of necessary structures within a few thousand. Our work concludes that, in future, the proposed methodology can be applied to realistic complex chemical reactions with various energy profiles. △ Less

Submitted 18 November, 2019; v1 submitted 13 September, 2019; originally announced September 2019.

Prediction of C7N6 and C9N4: Stable and strong porous carbon-nitride nanosheets with attractive electronic and optical properties

Authors: Bohayra Mortazavi, Masoud Shahrokhi, Alexander V Shapeev, Timon Rabczuk, Xiaoying Zhuang

Abstract: In this work, three novel porous carbon-nitride nanosheets with C7N6, C9N4 and C10N3 stoichiometries are predicted. First-principles simulations were accordingly employed to evaluate stability and explore the mechanical, electronic and optical properties. Phonon dispersions confirm the dynamical stability of all predicted nanosheets. Nonetheless, ab-initio molecular dynamics results indicate that… ▽ More In this work, three novel porous carbon-nitride nanosheets with C7N6, C9N4 and C10N3 stoichiometries are predicted. First-principles simulations were accordingly employed to evaluate stability and explore the mechanical, electronic and optical properties. Phonon dispersions confirm the dynamical stability of all predicted nanosheets. Nonetheless, ab-initio molecular dynamics results indicate that only C7N6 and C9N4 are thermally stable. C7N6, C9N4 and C10N3 nanosheets were predicted to exhibit high elastic modulus of 212, 202 and 208 N/m and maximum tensile strengths of 14.1, 22.4 and 15.8 N/m, respectively. C7N6 monolayer was confirmed to be a direct band-gap semiconductor, with a 2.25 eV gap according to the HSE06 method estimation. Interestingly, C9N4 and C10N3 monolayers show metallic character. The first absorption peaks of optical spectra reveal that C7N6 nanosheet can absorb the visible light, whereas C9N4 and C10N3 monolayers can absorb the Infrared range of light. Moreover, the absorption coefficient and optical conductivity of predicted nanosheets in the visible range of light are larger than those of the graphene. The results provided by this study confirm the stability and highlight very promising properties of C7N6 and C9N4 nanosheets, which may serve as promising candidates for numerous advanced technologies. △ Less

Submitted 16 August, 2019; v1 submitted 8 August, 2019; originally announced August 2019.

Journal ref: Journal of Materials Chemistry C 2019

A Performance and Cost Assessment of Machine Learning Interatomic Potentials

Authors: Yunxing Zuo, Chi Chen, Xiangguo Li, Zhi Deng, Yiming Chen, Jörg Behler, Gábor Csányi, Alexander V. Shapeev, Aidan P. Thompson, Mitchell A. Wood, Shyue ** Ong

Abstract: Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of ML-IAPs based on four local environment descriptors --- Behler-Parrinello symmetry functions, smooth overlap of atomic positions… ▽ More Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of ML-IAPs based on four local environment descriptors --- Behler-Parrinello symmetry functions, smooth overlap of atomic positions (SOAP), the Spectral Neighbor Analysis Potential (SNAP) bispectrum components, and moment tensors --- using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model, and consequently computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications. △ Less

Submitted 24 July, 2019; v1 submitted 20 June, 2019; originally announced June 2019.

Machine-learned Interatomic Potentials for Alloys and Alloy Phase Diagrams

Authors: Conrad W. Rosenbrock, Konstantin Gubaev, Alexander V. Shapeev, Livia B. Pártay, Noam Bernstein, Gábor Csányi, Gus L. W. Hart

Abstract: We introduce machine-learned potentials for Ag-Pd to describe the energy of alloy configurations over a wide range of compositions. We compare two different approaches. Moment tensor potentials (MTP) are polynomial-like functions of interatomic distances and angles. The Gaussian Approximation Potential (GAP) framework uses kernel regression, and we use the Smooth Overlap of Atomic Positions (SOAP)… ▽ More We introduce machine-learned potentials for Ag-Pd to describe the energy of alloy configurations over a wide range of compositions. We compare two different approaches. Moment tensor potentials (MTP) are polynomial-like functions of interatomic distances and angles. The Gaussian Approximation Potential (GAP) framework uses kernel regression, and we use the Smooth Overlap of Atomic Positions (SOAP) representation of atomic neighbourhoods that consists of a complete set of rotational and permutational invariants provided by the power spectrum of the spherical Fourier transform of the neighbour density. Both types of potentials give excellent accuracy for a wide range of compositions and rival the accuracy of cluster expansion, a benchmark for this system. While both models are able to describe small deformations away from the lattice positions, SOAP-GAP excels at transferability as shown by sensible transformation paths between configurations, and MTP allows, due to its lower computational cost, the calculation of compositional phase diagrams. Given the fact that both methods perform as well as cluster expansion would but yield off-lattice models, we expect them to open new avenues in computational materials modeling for alloys. △ Less

Submitted 9 July, 2019; v1 submitted 18 June, 2019; originally announced June 2019.

Comments: 9 pages, 6 figures, 4 tables

Ab initio vibrational free energies including anharmonicity for multicomponent alloys

Authors: Blazej Grabowski, Yuji Ikeda, Fritz Körmann, Christoph Freysoldt, Andrew Ian Duff, Alexander Shapeev, Jörg Neugebauer

Abstract: A density-functional-theory based approach to efficiently compute numerically exact vibrational free energies - including anharmonicity - for chemically complex multicomponent alloys is developed. It is based on a combination of thermodynamic integration and a machine-learning potential. We demonstrate the performance of the approach by computing the anharmonic free energy of the prototypical five… ▽ More A density-functional-theory based approach to efficiently compute numerically exact vibrational free energies - including anharmonicity - for chemically complex multicomponent alloys is developed. It is based on a combination of thermodynamic integration and a machine-learning potential. We demonstrate the performance of the approach by computing the anharmonic free energy of the prototypical five-component VNbMoTaW refractory high entropy alloy. △ Less

Submitted 28 February, 2019; originally announced February 2019.

Deeper Connections between Neural Networks and Gaussian Processes Speed-up Active Learning

Authors: Evgenii Tsymbalov, Sergei Makarychev, Alexander Shapeev, Maxim Panov

Abstract: Active learning methods for neural networks are usually based on greedy criteria which ultimately give a single new design point for the evaluation. Such an approach requires either some heuristics to sample a batch of design points at one active learning iteration, or retraining the neural network after adding each data point, which is computationally inefficient. Moreover, uncertainty estimates… ▽ More Active learning methods for neural networks are usually based on greedy criteria which ultimately give a single new design point for the evaluation. Such an approach requires either some heuristics to sample a batch of design points at one active learning iteration, or retraining the neural network after adding each data point, which is computationally inefficient. Moreover, uncertainty estimates for neural networks sometimes are overconfident for the points lying far from the training sample. In this work we propose to approximate Bayesian neural networks (BNN) by Gaussian processes, which allows us to update the uncertainty estimates of predictions efficiently without retraining the neural network, while avoiding overconfident uncertainty prediction for out-of-sample points. In a series of experiments on real-world data including large-scale problems of chemical and physical modeling, we show superiority of the proposed approach over the state-of-the-art methods. △ Less

Submitted 27 February, 2019; originally announced February 2019.

Journal ref: Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence {IJCAI-19}, 2019

Moment tensor Potentials as a Promising Tool to Study Diffusion Processes

Authors: I. I. Novoselov, A. V. Yanilkin, A. V. Shapeev, E. V. Podryabinkin

Abstract: A recently proposed class of machine-learning interatomic potentials --- Moment tensor potentials (MTPs) --- is investigated in this work. MTPs are able to actively select configurations and parametrize the potential on-the-fly. It is shown that MTPs accurately reproduce energies, forces and stresses calculated ab initio. As a more comprehensive test, MTPs are employed to calculate vacancy diffusi… ▽ More A recently proposed class of machine-learning interatomic potentials --- Moment tensor potentials (MTPs) --- is investigated in this work. MTPs are able to actively select configurations and parametrize the potential on-the-fly. It is shown that MTPs accurately reproduce energies, forces and stresses calculated ab initio. As a more comprehensive test, MTPs are employed to calculate vacancy diffusion rates in Al, Mo and Si. We demonstrate that the results are in a good agreement with ab initio data for the materials considered. △ Less

Submitted 10 December, 2018; v1 submitted 7 December, 2018; originally announced December 2018.

Comments: 28 pages, 9 figures, 1 table, 42 references

Impact of lattice relaxations on phase transitions in a high-entropy alloy studied by machine-learning potentials

Authors: Tatiana Kostiuchenko, Fritz Körmann, Jörg Neugebauer, Alexander Shapeev

Abstract: Recently, high-entropy alloys (HEAs) have attracted wide attention due to their extraordinary materials properties. A main challenge in identifying new HEAs is the lack of efficient approaches for exploring their huge compositional space. Ab initio calculations have emerged as a powerful approach that complements experiment. However, for multicomponent alloys existing approaches suffer from the ch… ▽ More Recently, high-entropy alloys (HEAs) have attracted wide attention due to their extraordinary materials properties. A main challenge in identifying new HEAs is the lack of efficient approaches for exploring their huge compositional space. Ab initio calculations have emerged as a powerful approach that complements experiment. However, for multicomponent alloys existing approaches suffer from the chemical complexity involved. In this work we propose a method for studying HEAs computationally. Our approach is based on the application of machine-learning potentials based on ab initio data in combination with Monte Carlo simulations. The high efficiency and performance of the approach are demonstrated on the prototype bcc NbMoTaW HEA. The approach is employed to study phase stability, phase transitions, and chemical short-range order. The importance of including local relaxation effects is revealed: they significantly stabilize single-phase formation of bcc NbMoTaW down to room temperature. Finally, a so-far unknown mechanism that drives chemical order due to atomic relaxation at ambient temperatures is discovered. △ Less

Submitted 6 May, 2019; v1 submitted 25 October, 2018; originally announced October 2018.

Comments: 7 pages, 5 figures

Journal ref: npj Computational Materials, 5(1), 55 (2019)

Machine-learned multi-system surrogate models for materials prediction

Authors: Chandramouli Nyshadham, Matthias Rupp, Brayden Bekker, Alexander V. Shapeev, Tim Mueller, Conrad W. Rosenbrock, Gábor Csányi, David W. Wingate, Gus L. W. Hart

Abstract: Surrogate machine-learning models are transforming computational materials science by predicting properties of materials with the accuracy of ab initio methods at a fraction of the computational cost. We demonstrate surrogate models that simultaneously interpolate energies of different materials on a dataset of 10 binary alloys (AgCu, AlFe, AlMg, AlNi, AlTi, CoNi, CuFe, CuNi, FeV, NbNi) with 10 di… ▽ More Surrogate machine-learning models are transforming computational materials science by predicting properties of materials with the accuracy of ab initio methods at a fraction of the computational cost. We demonstrate surrogate models that simultaneously interpolate energies of different materials on a dataset of 10 binary alloys (AgCu, AlFe, AlMg, AlNi, AlTi, CoNi, CuFe, CuNi, FeV, NbNi) with 10 different species and all possible fcc, bcc and hcp structures up to 8 atoms in the unit cell, 15\,950 structures in total. We find that the deviation of prediction errors when increasing the number of simultaneously modeled alloys is less than 1\,meV/atom. Several state-of-the-art materials representations and learning algorithms were found to qualitatively agree on the prediction errors of formation enthalpy with relative errors of $<$2.5\% for all systems. △ Less

Submitted 20 May, 2019; v1 submitted 24 September, 2018; originally announced September 2018.

Comments: 12 pages, 7 figures

Journal ref: npj Computational Materials 5.1 (2019): 51

Improving accuracy of interatomic potentials: more physics or more data? A case study of silica

Authors: Ivan S. Novikov, Alexander V. Shapeev

Abstract: In this paper we test two strategies to improving the accuracy of machine-learning potentials, namely adding more fitting parameters thus making use of large volumes of available quantum-mechanical data, and adding a charge-equilibration model to account for ionic nature of the SiO2 bonding. To that end, we compare Moment Tensor Potentials (MTPs) and MTPs combined with the charge-equilibration (QE… ▽ More In this paper we test two strategies to improving the accuracy of machine-learning potentials, namely adding more fitting parameters thus making use of large volumes of available quantum-mechanical data, and adding a charge-equilibration model to account for ionic nature of the SiO2 bonding. To that end, we compare Moment Tensor Potentials (MTPs) and MTPs combined with the charge-equilibration (QEq) model (MTP+QEq) fitted to a density functional theory dataset of alpha-quartz SiO2-based structures. In order to make a meaningful comparison, in addition to the accuracy, we assess the uncertainty of predictions of each potential. It is shown that adding the QEq model to MTP does not make any improvement over the MTP potential alone, while adding more parameters does improve the accuracy and uncertainty of its predictions. △ Less

Submitted 16 November, 2018; v1 submitted 11 August, 2018; originally announced August 2018.

Accelerating high-throughput searches for new alloys with active learning of interatomic potentials

Authors: Konstantin Gubaev, Evgeny V. Podryabinkin, Gus L. W. Hart, Alexander V. Shapeev

Abstract: We propose an approach to materials prediction that uses a machine-learning interatomic potential to approximate quantum-mechanical energies and an active learning algorithm for the automatic selection of an optimal training dataset. Our approach significantly reduces the amount of DFT calculations needed, resorting to DFT only to produce the training data, while structural optimization is perform… ▽ More We propose an approach to materials prediction that uses a machine-learning interatomic potential to approximate quantum-mechanical energies and an active learning algorithm for the automatic selection of an optimal training dataset. Our approach significantly reduces the amount of DFT calculations needed, resorting to DFT only to produce the training data, while structural optimization is performed using the interatomic potentials. Our approach is not limited to one (or a small number of) lattice types (as is the case for cluster expansion, for example) and can predict structures with lattice types not present in the training dataset. We demonstrate the effectiveness of our algorithm by predicting the convex hull for the following three systems: Cu-Pd, Co-Nb-V, and Al-Ni-Ti. Our method is three to four orders of magnitude faster than conventional high-throughput DFT calculations and explores a wider range of materials space. In all three systems, we found unreported stable structures compared to the AFLOW database. Because our method is much cheaper and explores much more of materials space than high-throughput methods or cluster expansion, and because our interatomic potentials have a systematically improvable accuracy compared to empirical potentials such as EAM, etc., it will have a significant impact in the discovery of new alloy phases, particularly those with three or more components. △ Less

Submitted 27 June, 2018; originally announced June 2018.

Dropout-based Active Learning for Regression

Authors: Evgenii Tsymbalov, Maxim Panov, Alexander Shapeev

Abstract: Active learning is relevant and challenging for high-dimensional regression models when the annotation of the samples is expensive. Yet most of the existing sampling methods cannot be applied to large-scale problems, consuming too much time for data processing. In this paper, we propose a fast active learning algorithm for regression, tailored for neural network models. It is based on uncertainty… ▽ More Active learning is relevant and challenging for high-dimensional regression models when the annotation of the samples is expensive. Yet most of the existing sampling methods cannot be applied to large-scale problems, consuming too much time for data processing. In this paper, we propose a fast active learning algorithm for regression, tailored for neural network models. It is based on uncertainty estimation from stochastic dropout output of the network. Experiments on both synthetic and real-world datasets show comparable or better performance (depending on the accuracy metric) as compared to the baselines. This approach can be generalized to other deep learning architectures. It can be used to systematically improve a machine-learning model as it offers a computationally efficient way of sampling additional data. △ Less

Submitted 5 July, 2018; v1 submitted 26 June, 2018; originally announced June 2018.

Comments: Report on AIST 2018; will be published in Springer LNCS series (Analysis of Images, Social Networks and Texts - 7th International Conference, AIST 2018)

Journal ref: Analysis of Images, Social Networks and Texts - 7th International Conference, AIST 2018, Lecture Notes in Computer Science book series (LNCS), volume 11179, pp. 247-258

Automated Calculation of Thermal Rate Coefficients using Ring Polymer Molecular Dynamics and Machine-Learning Interatomic Potentials with Active Learning

Authors: Ivan S. Novikov, Yury V. Suleimanov, Alexander V. Shapeev

Abstract: We propose a methodology for fully automated calculation of thermal rate coefficients of gas phase chemical reactions, which is based on combining the ring polymer molecular dynamics (RPMD) with the machine-learning interatomic potentials actively learning on-the-fly. Based on the original computational procedure implemented in the RPMDrate code, our methodology gradually and automatically constru… ▽ More We propose a methodology for fully automated calculation of thermal rate coefficients of gas phase chemical reactions, which is based on combining the ring polymer molecular dynamics (RPMD) with the machine-learning interatomic potentials actively learning on-the-fly. Based on the original computational procedure implemented in the RPMDrate code, our methodology gradually and automatically constructs the potential energy surfaces (PESs) from scratch with the data set points being selected and accumulated during the RPMDrate simulation. Such an approach ensures that our final machine-learning model provides reliable description of the PES which avoids artifacts during exploration of the phase space by RPMD trajectories. We tested our methodology on two representative thermally activated chemical reactions studied recently by RPMDrate at temperatures within the interval of 300--1000~K. The corresponding PESs were generated by fitting to only a few thousands automatically generated structures (less than 5000) while the RPMD rate coefficients retained the deviation from the reference values within the typical convergence error of RPMDrate. In future, we plan to apply our methodology to chemical reactions which proceed via complex-formation thus providing a completely general tool for calculating RPMD thermal rate coefficients for any polyatomic gas phase chemical reaction. △ Less

Submitted 15 November, 2018; v1 submitted 30 May, 2018; originally announced May 2018.

Accelerating crystal structure prediction by machine-learning interatomic potentials with active learning

Authors: Evgeny V. Podryabinkin, Evgeny V. Tikhonov, Alexander V. Shapeev, Artem R. Oganov

Abstract: In this letter we propose a new methodology for crystal structure prediction, which is based on the evolutionary algorithm USPEX and the machine-learning interatomic potentials actively learning on-the-fly. Our methodology allows for an automated construction of an interatomic interaction model from scratch replacing the expensive DFT with a speedup of several orders of magnitude. Predicted low-en… ▽ More In this letter we propose a new methodology for crystal structure prediction, which is based on the evolutionary algorithm USPEX and the machine-learning interatomic potentials actively learning on-the-fly. Our methodology allows for an automated construction of an interatomic interaction model from scratch replacing the expensive DFT with a speedup of several orders of magnitude. Predicted low-energy structures are then tested on DFT, ensuring that our machine-learning model does not introduce any prediction error. We tested our methodology on a problem of prediction of carbon allotropes, dense sodium structures and boron allotropes including those which have more than 100 atoms in the primitive cell. All the the main allotropes have been reproduced and a new 54-atom structure of boron have been found at very modest computational efforts. △ Less

Submitted 26 February, 2019; v1 submitted 21 February, 2018; originally announced February 2018.

Journal ref: Phys. Rev. B 99, 064114 (2019)

Machine learning of molecular properties: locality and active learning

Authors: Konstantin Gubaev, Evgeny V. Podryabinkin, Alexander V. Shapeev

Abstract: In recent years the machine learning techniques have shown a great potential in various problems from a multitude of disciplines, including materials design and drug discovery. The high computational speed on the one hand and the accuracy comparable to that of DFT on another hand make machine learning algorithms efficient for high-throughput screening through chemical and configurational space. Ho… ▽ More In recent years the machine learning techniques have shown a great potential in various problems from a multitude of disciplines, including materials design and drug discovery. The high computational speed on the one hand and the accuracy comparable to that of DFT on another hand make machine learning algorithms efficient for high-throughput screening through chemical and configurational space. However, the machine learning algorithms available in the literature require large training datasets to reach the chemical accuracy and also show large errors for the so-called outliers - the out-of-sample molecules, not well-represented in the training set. In the present paper we propose a new machine learning algorithm for predicting molecular properties that addresses these two issues: it is based on a local model of interatomic interactions providing high accuracy when trained on relatively small training sets and an active learning algorithm of optimally choosing the training set that significantly reduces the errors for the outliers. We compare our model to the other state-of-the-art algorithms from the literature on the widely used benchmark tests. △ Less

Submitted 11 April, 2018; v1 submitted 18 September, 2017; originally announced September 2017.

Comments: revised version

Accurate representation of formation energies of crystalline alloys with many components

Authors: Alexander Shapeev

Abstract: In this paper I propose a new model for representing the formation energies of multicomponent crystalline alloys as a function of atom types. In the cases when displacements of atoms from their equilibrium positions are not large, the proposed method has a similar accuracy as the state-of-the-art cluster expansion method, and a better accuracy when the fitting dataset size is small. The proposed m… ▽ More In this paper I propose a new model for representing the formation energies of multicomponent crystalline alloys as a function of atom types. In the cases when displacements of atoms from their equilibrium positions are not large, the proposed method has a similar accuracy as the state-of-the-art cluster expansion method, and a better accuracy when the fitting dataset size is small. The proposed model has only two tunable parameters - one for the interaction range and one for the interaction complexity. △ Less

Submitted 1 August, 2017; v1 submitted 10 December, 2016; originally announced December 2016.

Comments: 13 pages, 5 figures; a 23-component test case added and compared to compressive sensing, journal reference and DOI added

Journal ref: Computational Materials Science, Volume 139, November 2017, Pages 26-30

Active learning of linearly parametrized interatomic potentials

Authors: Evgeny V. Podryabinkin, Alexander V. Shapeev

Abstract: This paper introduces an active learning approach to the fitting of machine learning interatomic potentials. Our approach is based on the D-optimality criterion for selecting atomic configurations on which the potential is fitted. It is shown that the proposed active learning approach is highly efficient in training potentials on the fly, ensuring that no extrapolation is attempted and leading to… ▽ More This paper introduces an active learning approach to the fitting of machine learning interatomic potentials. Our approach is based on the D-optimality criterion for selecting atomic configurations on which the potential is fitted. It is shown that the proposed active learning approach is highly efficient in training potentials on the fly, ensuring that no extrapolation is attempted and leading to a completely reliable atomistic simulation without any significant decrease in accuracy. We apply our approach to molecular dynamics and structure relaxation, and we argue that it can be applied, in principle, to any other type of atomistic simulation. The software, test cases, and examples of usage are published at http://gitlab.skoltech.ru/shapeev/mlip/. △ Less

Submitted 22 August, 2017; v1 submitted 28 November, 2016; originally announced November 2016.

Comments: 12 pages, numerical tests added

Journal ref: Computational Materials Science, volume 140, pages 171-180, 2017