arXiv subjects

Max Hodapp

Publications and source records attributed to Max Hodapp.

13 recordsLinked to original sources

arXiv2026

AutoPot: Automated and massively parallelized construction of Machine-Learning Potentials

Machine-learning potentials (MLIPs) have been a breakthrough for computational physics in bringing the accuracy of quantum mechanics to atomistic modeling. To achieve near-quantum accuracy, it is necessary that neighborhoods contained in the training set are rather close to the ones encountered during a simulation. Yet, constructing a single training set that works well for all applications is, and likely will remain, infeasible, so, one strategy is to supplement training protocols for MLIPs with additional learning methods, such as active learning, or fine-tuning. This strategy, however, yields very complex training protocols that are difficult to implement efficiently, and cumbersome to interpret, analyze, and reproduce. To address the above difficulties, we propose AutoPot, a software for automating the construction and archiving of MLIPs. AutoPot is based on BlackDynamite, a software operating parametric tasks, e.g., running simulations, or single-point ab initio calculations, in a highly-parallelized fashion, and Motoko, an event-based workflow manager for orchestrating interactions between the tasks. The initial version of AutoPot supports selection of training configurations from large training candidate sets, and on-the-fly selection from molecular dynamics simulations, using Moment Tensor Potentials as implemented in MLIP-2, and single-point calculations of the selected training configurations using VASP. Another strength of AutoPot is its flexibility: BlackDynamite tasks and orchestrators are Python functions to which own existing code can be easily added and manipulated without writing complex parsers. Therefore, it will be straightforward to add other MLIP and ab initio codes, and manipulate the Motoko orchestrators to implement other training protocols.

physics.comp-ph↗

arXiv2025

Incorporating Coulomb interactions with fixed charges in Moment Tensor Potentials and Equivariant Tensor Network Potentials

In this work, we incorporate long-range electrostatic interactions in the form of the Coulomb model with fixed charges into the functional form of short-range machine-learning interatomic potentials (MLIPs), particularly in the Moment Tensor Potential and Equivariant Tensor Network potential. We show that explicit incorporation of the Coulomb interactions with fixed charges leads to a significant reduction of energy fitting errors, namely, more than four times, of short-range MLIPs trained on organic dimers of charged molecules. Furthermore, with our long-range models we demonstrate a significant improvement in the prediction of the binding curves of the organic dimers of charged molecules. Finally, we show that the results calculated with MLIPs are in good correspondence with those obtained with density functional theory for organic dimers of charged molecules.

physics.chem-ph↗

arXiv2025

Moment Tensor Potential and Equivariant Tensor Network Potential with explicit dispersion interactions

In this study, we investigate the effect of incorporating explicit dispersion interactions in the functional form of machine learning interatomic potentials (MLIPs), particularly in the Moment Tensor Potential and Equivariant Tensor Network potential for accurate modeling of liquid carbon tetrachloride, methane, and toluene. We show that explicit incorporation of dispersion interactions via D2 and D3 corrections significantly improves the accuracy of MLIPs when the cutoff radius is set to a commonly used value of 5 -- 6 \r{A}. We also show that for carbon tetrachloride and methane, a substantial improvement in accuracy can be achieved by extending the cutoff radius to 7.5 \r{A}. However, for accurate modeling of toluene, explicit incorporation of dispersion remains important. Furthermore, we find that MLIPs incorporating dispersion interactions via D2 reach a close level of accuracy to those incorporating D3, and D2 is suitable for accurate modeling of the systems in the study, while being less computationally expensive. We evaluated the accuracy of MLIPs in dimer binding curves compared to ab initio data and in predicting density and radial distribution functions compared to experiments.

physics.chem-ph↗

arXiv2024

Actively-trained magnetic Moment Tensor Potentials for mechanical, dynamical, and thermal properties of paramagnetic CrN

We present a protocol for automated fitting of magnetic Moment Tensor Potential explicitly including magnetic moments in its functional form. For the fitting of this potential we use energies, forces, stresses, and magnetic forces (negative derivatives of energies with respect to magnetic moments) of configurations selected with an active learning algorithm. These selected configurations are computed using constrained density functional theory, which enables calculating energies and their derivatives for both equilibrium and non-equilibrium (excited) magnetic states. We test our protocol on the system of B1-CrN and demonstrate that the automatically trained magnetic Moment Tensor Potential reproduces mechanical, dynamical, and thermal properties, of B1-CrN in the paramagnetic state with respect to density functional theory and experiments.

cond-mat.mtrl-sci↗

arXiv2024

Exact average many-body interatomic interaction model for random alloys

Understanding the physical origin of deformation mechanisms in random alloys requires an understanding of their average behavior and, equally important, the role of local fluctuations around the average. Material properties of random alloys can be computed using direct simulations on random configurations but some properties are very difficult to compute, for others it is not even fully understood how to compute them using random sampling, in particular, interaction energies between multiple defects. To that end, we develop an atomistic model that does the averaging on the level of interatomic potentials. Then, the problem of averaging via random sampling is bypassed since computing material properties on random configurations reduces to computing material properties on a single crystal, the average alloy. We develop our average model on the class of linear machine-learning interatomic potentials. To that end, using tools from higher-order statistics, we derive an analytic expansion of the average many-body per-atom energy in terms of average tensor products of the feature vectors that scales linearly with the size of an atomic neighborhood. In order to avoid forming higher-order tensors composed of products of feature vectors, we develop an implementation using equivariant tensor network potentials in which the feature vectors are contracted to small-sized tensors before taking the average. We show that our average model predicts the compact screw dislocation core structure in the NbMoTaW medium-entropy alloy, in agreement with density functional theory, as opposed to state-of-the-art average embedded atom method potentials that predict artificial polarized cores. Hence, we anticipate that our model will become useful for understanding mechanistic origins of material properties and for developing predictive models of mechanical properties of random alloys.

physics.comp-ph↗

arXiv2024

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

We developed a method for fitting machine-learning interatomic potentials with magnetic degrees of freedom, namely, magnetic Moment Tensor Potentials (mMTP). The main feature of our method consists in fitting mMTP to magnetic forces (negative derivatives of energies with respect to magnetic moments) as obtained spin-polarized density functional theory calculations. We test our method on the bcc Fe-Al system with different compositions. Specifically, we calculate formation energies, equilibrium lattice parameter, and total cell magnetization. Our findings demonstrate an accurate correspondence between the 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. Furthermore, we demonstrate that fitting to magnetic forces increases the reliability of structure relaxation (or, equilibration), in the sense of ensuring that every relaxation run ends up with a successfully relaxed structure (the failure may otherwise be caused by falsely driving a configuration away from the region covered in the training set).

cond-mat.mtrl-sci↗

arXiv2023

Ab initio framework for deciphering trade-off relationships in multi-component alloys

While first-principles methods have been successfully applied to characterize individual properties of multi-principal element alloys (MPEA), their use to search for optimal trade-offs between competing properties is hampered by high computational demands. In this work, we present a novel framework to explore Pareto-optimal compositions by integrating advanced ab initio-based techniques into a Bayesian multi-objective optimization workflow complemented with a simple analytical model providing straightforward analysis of trends. We benchmark the framework by applying it to solid solution strengthening and ductility of refractory MPEAs, with the parameters of the strengthening and ductility models being efficiently computed using a combination of the coherent-potential approximation method, accounting for finite-temperature effects, and actively-learned moment-tensor potentials parameterized with ab initio data. Properties obtained from ab initio calculations are subsequently used to extend predictions of all relevant material properties to a large class of refractory alloys with the help of the analytical model validated by the data and relying on a few element-specific parameters and universal functions that describe bonding between elements. Our findings offer new crucial insights into the traditional strength-vs-ductility dilemma of refractory MPEAs. The proposed framework is versatile and can be extended to other materials and properties of interest, enabling a predictive and tractable high-throughput screening of Pareto-optimal MPEAs over the entire composition space.

cond-mat.mtrl-sci↗

arXiv2023

Automated atomistic simulations of dissociated dislocations with ab initio accuracy

In (M Hodapp and A Shapeev 2020 Mach. Learn.: Sci. Technol. 1 045005), we have proposed an algorithm that fully automatically trains machine-learning interatomic potentials (MLIPs) during large-scale simulations, and successfully applied it to simulate screw dislocation motion in body-centered cubic tungsten. The algorithm identifies local subregions of the large-scale simulation region where the potential extrapolates, and then constructs periodic configurations of 100--200 atoms out of these non-periodic subregions that can be efficiently computed with plane-wave Density Functional Theory (DFT) codes. In this work, we extend this algorithm to dissociated dislocations with arbitrary character angles and apply it to partial dislocations in face-centered cubic aluminum. Given the excellent agreement with available DFT reference results, we argue that our algorithm has the potential to become a universal way of simulating dissociated dislocations in face-centered cubic and possibly also other materials, such as hexagonal closed-packed magnesium, and their alloys. Moreover, it can be used to construct reliable training sets for MLIPs to be used in large-scale simulations of curved dislocations.

physics.comp-ph↗

arXiv2023

Equivariant Tensor Network Potentials

Machine-learning interatomic potentials (MLIPs) have made a significant contribution to the recent progress in the fields of computational materials and chemistry due to the 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-based algorithms 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.

physics.comp-ph↗

arXiv2022

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

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.

cond-mat.mtrl-sci↗

arXiv2021

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

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.

cond-mat.mtrl-sci↗

arXiv2021

Efficient flexible boundary conditions for long dislocations

We present a novel efficient implementation of the flexible boundary condition (FBC) method, initially proposed by Sinclair et al., for large single-periodic problems. Efficiency is primarily achieved by constructing a hierarchical matrix ($\mathscr{H}$-matrix) representation of the periodic Green matrix, reducing the complexity for updating the boundary conditions of the atomistic problem from quadratic to almost linear in the number of pad atoms. In addition, our implementation is supported by various other tools from numerical analysis, such as a residual-based transformation of the boundary conditions to accelerate the convergence. We assess the method for a comprehensive set of examples, relevant for predicting mechanical properties, such as yield strength or ductility, including dislocation bow-out, dislocation-precipitate interaction, and dislocation cross-slip. The main result of our analysis is that the FBC method is robust, easy-to-use, and up to two orders of magnitude more efficient than the current state-of-the-art method for this class of problems, the periodic array of dislocations (PAD) method, in terms of the required number of per-atom force computations when both methods give similar accuracy. This opens new prospects for large-scale atomistic simulations - without having to worry about spurious image effects that plague classical boundary conditions.

physics.comp-ph↗

arXiv2020

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

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.

physics.comp-ph↗