Density functional theory offers a very accurate way of computing materials properties from first principles. However, it is too expensive for modeling large-scale molecular systems whose are, in contrast, computed using interatomic potentials. The present paper considers, mathematical point view, the problem constructing potentials that approximate given quantum-mechanical interaction model. In particular, new class systematically improvable proposed, analyzed, and tested on an existing database.

Machine learning of the quantitative relationship between local environment descriptors and potential energy surface a system atoms has emerged as new frontier in development interatomic potentials (IAPs). Here, we present comprehensive evaluation machine IAPs (ML-IAPs) based on four descriptors—atom-centered symmetry functions (ACSF), smooth overlap atomic positions (SOAP), spectral neighbor analysis (SNAP) bispectrum components, moment tensors—using diverse data set generated using...

The subject of this paper is the technology (the 'how') constructing machine-learning interatomic potentials, rather than science 'what' and 'why') atomistic simulations using potentials. Namely, we illustrate how to construct moment tensor potentials active learning as implemented in MLIP package, focusing on efficient ways automatically sample configurations for training set, expanding set changes error predictions, up ab initio calculations a cost-effective manner, etc. package (short...

In this letter we propose a new methodology for crystal structure prediction, which is based on the evolutionary algorithm USPEX and machine-learning interatomic potentials actively learning on-the-fly. Our allows an automated construction of interaction model from scratch replacing expensive DFT with speedup several orders magnitude. Predicted low-energy structures are then tested DFT, ensuring that our does not introduce any prediction error. We problem carbon allotropes, dense sodium...

Abstract Density functional theory calculations are robust tools to explore the mechanical properties of pristine structures at their ground state but become exceedingly expensive for large systems finite temperatures. Classical molecular dynamics (CMD) simulations offer possibility study larger elevated temperatures, they require accurate interatomic potentials. Herein authors propose concept first‐principles multiscale modeling properties, where ab initio level accuracy is hierarchically...

One of the ultimate goals computational modeling in condensed matter is to be able accurately compute materials properties with minimal empirical information. First-principles approaches such as density functional theory (DFT) provide best possible accuracy on electronic but they are limited systems up a few hundreds, or at most thousands atoms. On other hand, classical molecular dynamics (CMD) simulations and finite element method (FEM) extensively employed study larger more realistic...

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 huge compositional space. Ab initio calculations emerged as a powerful approach that complements experiment. However, multicomponent existing suffer from chemical complexity involved. In this work we propose method studying computationally. Our based on application machine-learning...

In recent years, the machine learning techniques have shown great potent1ial in various problems from a multitude of disciplines, including materials design and drug discovery. The high computational speed on one hand accuracy comparable to that density functional theory another make algorithms efficient for high-throughput screening through chemical configurational space. However, available literature require large training datasets reach also show errors so-called outliers-the...

Two-dimensional transition metal carbides, that is, MXenes and especially Ti3C2, attract attention due to their excellent combination of properties. Ti3C2 nanosheets could be the material choice for future flexible electronics, energy storage, electromechanical nanodevices. There has been limited information available on mechanical properties which is essential utilization. We have fabricated studied using direct in situ tensile tests inside a transmission electron microscope, quantitative...

Abstract We introduce machine-learned potentials for Ag-Pd to describe the energy of alloy configurations over a wide range compositions. compare two different approaches. Moment tensor (MTPs) are polynomial-like functions interatomic distances and angles. The Gaussian approximation potential (GAP) framework uses kernel regression, we use smooth overlap atomic position (SOAP) representation neighborhoods that consist complete set rotational permutational invariants provided by power spectrum...

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

Surrogate machine-learning models are transforming computational materials science by predicting properties of with the accuracy ab initio methods at a fraction cost. We demonstrate surrogate that simultaneously interpolate energies different on dataset 10 binary alloys (AgCu, AlFe, AlMg, AlNi, AlTi, CoNi, CuFe, CuNi, FeV, NbNi) species and all possible fcc, bcc hcp structures up to 8 atoms in unit cell, 15\,950 total. find deviation prediction errors when increasing number modeled is less...

While lattice thermal conductivity is an important parameter for many technological applications, its calculation a time-consuming task, especially compounds with complex crystal structure. In this paper, we solve problem using machine learning interatomic potentials. These potentials trained on the density functional theory results and provide accurate description of dynamics. Additionally, active was applied to significantly reduce number expensive quantum-mechanical calculations required...

Abstract The unique and unanticipated properties of multiple principal component alloys have reinvigorated the field alloy design drawn strong interest across scientific disciplines. vast compositional parameter space makes these a area exploration by means computational design. However, as now method to compute efficiently, yet with high accuracy thermodynamic such has been missing. One underlying reasons is lack accurate efficient approaches vibrational free energies—including...

Significance Deforming a material to large extent without inelastic relaxation can result in unprecedented properties. However, the optimal deformation state is buried within vast continua of choices available strain space. Here we advance unique and powerful strategy circumvent conventional trial-and-error methods, adopt artificial intelligence techniques for rationally designing most energy-efficient pathway achieve desirable property such as electronic bandgap. The broad framework...

Carbon nitride nanomembranes are currently among the most appealing two-dimensional (2D) materials. As a nonstop endeavor in this field, novel 2D fused aromatic nanoporous network with C5N stoichiometry has been recently synthesized. Inspired by experimental advance and exciting physics of carbon nitrides, herein we conduct extensive density functional theory calculations to explore electronic, optical photocatalytic properties monolayer. In order examine dynamic stability evaluate...