The atomic cluster expansion is developed as a complete descriptor of the local environment, including multicomponent materials, and its relation to number other descriptors potentials discussed. effort for evaluating shown scale linearly with neighbors, irrespective order expansion. Application small Cu clusters demonstrates smooth convergence meV accuracy. By introducing nonlinear functions an interatomic potential obtained that comparable in accuracy state-of-the-art machine learning...

We report metallic NiPS3@NiOOH core–shell heterostructures as an efficient and durable electrocatalyst for the oxygen evolution reaction, exhibiting a low onset potential of 1.48 V (vs RHE) stable performance over 160 h. The atomically thin NiPS3 nanosheets are obtained by exfoliation bulk in presence ionic surfactant. OER mechanism was studied combination SECM, situ Raman spectroscopy, SEM, XPS measurements, which enabled direct observation formation heterostructure at electrode interface....

The presence of hydrogen may weaken the bonding iron atoms at grain boundaries, leading to intergranular embrittlement and thus failure bulk material. In this paper, we study interaction interstitials with close-packed open boundary structures in $\ensuremath{\alpha}$- $\ensuremath{\gamma}$-Fe using density-functional theory. We find that accommodation within boundaries strongly depends on local coordination available interstitial sites. Within larger sites are available, enhancing...

Abstract The atomic cluster expansion is a general polynomial of the energy in multi-atom basis functions. Here we implement performant C++ code that suitable for use large-scale atomistic simulations. We briefly review and give detailed expressions energies forces as well efficient algorithms their evaluation. demonstrate implemented shifts previously established Pareto front machine learning interatomic potentials toward faster more accurate calculations. Moreover, purpose...

We present an atomic cluster expansion (ACE) for carbon that improves over available classical and machine learning potentials. The ACE is parametrized from exhaustive set of important structures extended volume energy ranges, computed using density functional theory (DFT). Rigorous validation reveals accurately predicts a broad range properties both crystalline amorphous phases while being several orders magnitude more computationally efficient than models. demonstrate the predictive power...

Abstract The Atomic Cluster Expansion (ACE) provides a formally complete basis for the local atomic environment. ACE is not limited to representing energies as function of positions and chemical species, but can be generalized vectorial or tensorial properties incorporate further degrees freedom (DOF). This crucial magnetic materials with potential energy surfaces that depend on moments simultaneously. In this work, we employ formalism develop non-collinear parametrization prototypical...

Abstract Molecular dynamics simulation is an important tool in computational materials science and chemistry, the past decade it has been revolutionized by machine learning. This rapid progress learning interatomic potentials produced a number of new architectures just few years. Particularly notable among these are atomic cluster expansion, which unified many earlier ideas around atom-density-based descriptors, Neural Equivariant Interatomic Potentials (NequIP), message-passing neural...

New candidate ground states at 1:4, 1:2, and 1:1 compositions are identified in the well-known Fe-B system via a combination of ab initio high-throughput evolutionary searches. We show that proposed oP12-FeB2 stabilizes by break up 2D boron layers into 1D chains while oP10-FeB4 distortion 3D network. The uniqueness these configurations gives rise to set remarkable properties: is expected be first semiconducting metal diboride shown have potential for phonon-mediated superconductivity with Tc 15-20 K.

Density functional theory calculations have been used to study the mixing behavior of Fe-Cr alloys. The heats formation $\ensuremath{\Delta}{E}_{f}$ 65 structures in their magnetic ground states determined. A positive is found over most concentration range. From 0--12 % Cr a small negative down $\ensuremath{-}8\phantom{\rule{0.3em}{0ex}}\mathrm{meV}$/atom found. origin Fe-rich traced solution energy single atoms. At low concentration, atoms Fe repel each other, causing ordering. Cr-Cr...

To support and accelerate the development of simulation protocols in atomistic modelling, we introduce an integrated environment (IDE) for computational materials science called pyiron (http://pyiron.org). The IDE combines a web based source code editor, job management system build automation, hierarchical data solution. core components are objects on abstract class, which links application structures such as structures, projects, jobs, computing resources with persistent storage interactive...

The atomic cluster expansion (Drautz, Phys. Rev. B 99, 014104 (2019)) is extended in two ways, the modelling of vectorial and tensorial properties inclusion degrees freedom addition to positions atoms. In particular, species, magnetic moments charges are attached an that includes different on equal footing derived. Expressions for efficient evaluation forces torques given. Relations other methods discussed.

The atomic cluster expansion (ACE) provides a general, local, and complete representation of energies. Here we present an efficient framework for parametrization ACE models elements, alloys, molecules. To this end, first introduce general requirements physically meaningful description the interaction, in addition to usual equivariance requirements. We then demonstrate that can be converged systematically with respect two fundamental characteristics---the number complexity basis functions...

The rapid progress of machine learning interatomic potentials over the past couple years produced a number new architectures. Particularly notable among these are Atomic Cluster Expansion (ACE), which unified many earlier ideas around atom density-based descriptors, and Neural Equivariant Interatomic Potentials (NequIP), message passing neural network with equivariant features that showed state art accuracy. In this work, we construct mathematical framework unifies models: ACE is generalised...

The atomic cluster expansion (ACE) was proposed recently as a new class of data-driven interatomic potentials with formally complete basis set. Since the development any potential requires careful selection training data and thorough validation, an automation construction dataset well indication model's uncertainty are highly desirable. In this work, we compare performance two approaches for ACE models based on D-optimality criterion ensemble learning. While both show comparable predictions,...

The atomic cluster expansion provides local, complete basis functions that enable efficient parametrization of many-atom interactions. We extend the to incorporate graph functions. This naturally leads representations description semilocal interactions in physically and chemically transparent form. Simplification by tensor decomposition results an iterative procedure comprises current message-passing machine learning interatomic potentials. demonstrate accuracy efficiency for a number small...

Iron-chromium alloys are characterized by a complex phase diagram, the small negative enthalpy of mixing at low Cr concentrations in bcc $\ensuremath{\alpha}$-phase Fe, and inversion short-range order parameter. We present Monte Carlo simulations binary Fe-Cr alloy based on cluster expansion approximation for system. The set coefficients is validated against density functional calculations energies clusters chromium structure. show that limit concentration remains up to fairly high...

An analytic interatomic bond-order potential is derived that depends explicitly on the valence of transition-metal element. It generalizes second-moment Finnis-Sinclair and fourth-moment Carlsson potentials to include higher moments. We find sixth-moment approximation predicts not only structural trend from $\mathrm{hcp}\ensuremath{\rightarrow}\mathrm{bcc}\ensuremath{\rightarrow}\mathrm{hcp}\ensuremath{\rightarrow}\mathrm{fcc}$ observed across nonmagnetic $4d$ $5d$ series, but also different...

Structural transformations in Fe–C alloys are decisive for the mechanical properties of steels, but their modeling remains a challenge due to simultaneous changes Fe lattice and redistribution C. With combination orientation relationships between austenite, ferrite cementite, we identify metastable intermediate structure (MIS), which can serve as link three phases. Based on this framework, different mechanisms depending local conditions (C concentration, strain, magnetism) revealed from ab...