arXiv subjects

Antoine Loew

Publications and source records attributed to Antoine Loew.

6 recordsLinked to original sources

arXiv2025

AI-Driven Expansion and Application of the Alexandria Database

We present a novel multi-stage workflow for computational materials discovery that achieves a 99% success rate in identifying compounds within 100 meV/atom of thermodynamic stability, with a threefold improvement over previous approaches. By combining the Matra-Genoa generative model, Orb-v2 universal machine learning interatomic potential, and ALIGNN graph neural network for energy prediction, we generated 119 million candidate structures and added 1.3 million DFT-validated compounds to the ALEXANDRIA database, including 74 thousand new stable materials. The expanded ALEXANDRIA database now contains 5.8 million structures with 175 thousand compounds on the convex hull. Predicted structural disorder rates (37-43%) match experimental databases, unlike other recent AI-generated datasets. Analysis reveals fundamental patterns in space group distributions, coordination environments, and phase stability networks, including sub-linear scaling of convex hull connectivity. We release the complete dataset, including sAlex25 with 14 million out-of-equilibrium structures containing forces and stresses for training universal force fields. We demonstrate that fine-tuning a GRACE model on this data improves benchmark accuracy. All data, models, and workflows are freely available under Creative Commons licenses.

cond-mat.mtrl-sci↗

arXiv2025

Universal Machine Learning Potentials under Pressure

Universal machine learning interatomic potentials (uMLIPs) represent arguably the most successful application of machine learning to materials science, demonstrating remarkable performance across diverse applications. However, critical blind spots in their reliability persist. Here, we address one such significant gap by systematically investigating the accuracy of uMLIPs under extreme pressure conditions from 0 to 150 GPa. Our benchmark reveals that while these models excel at standard pressure, their predictive accuracy deteriorates considerably as pressure increases. This decline in performance originates from fundamental limitations in the training data rather than in algorithmic constraints. In fact, we show that through targeted fine-tuning on high-pressure configurations, the robustness of the models can be easily increased. These findings underscore the importance of identifying and addressing overlooked regimes in the development of the next generation of truly universal interatomic potentials.

cond-mat.mtrl-sci↗

arXiv2025

Universal Machine Learning Potential for Systems with Reduced Dimensionality

We present a benchmark designed to evaluate the predictive capabilities of universal machine learning interatomic potentials across systems of varying dimensionality. Specifically, our benchmark tests zero- (molecules, atomic clusters, etc.), one- (nanowires, nanoribbons, nanotubes, etc.), two- (atomic layers and slabs) and three-dimensional (bulk materials) compounds. The benchmark reveals that while all tested models demonstrate excellent performance for three-dimensional systems, accuracy degrades progressively for lower-dimensional structures. The best performing models for geometry optimization are orbital version 2, equiformerV2, and the equivariant Smooth Energy Network, with the equivariant Smooth Energy Network also providing the most accurate energies. Our results indicate that the best models yield, on average, errors in the atomic positions in the range of 0.01-0.02 angstrom and errors in the energy below 10~meV/atom across all dimensionalities. These results demonstrate that state-of-the-art universal machine learning interatomic potentials have reached sufficient accuracy to serve as direct replacements for density functional theory calculations, at a small fraction of the computational cost, in simulations spanning the full range from isolated atoms to bulk solids. More significantly, the best performing models already enable efficient simulations of complex systems containing subsystems of mixed dimensionality, opening new possibilities for modeling realistic materials and interfaces.

cond-mat.mtrl-sci↗

arXiv2025

Accelerating point defect photo-emission calculations with machine learning interatomic potentials

We introduce a computational framework leveraging universal machine learning interatomic potentials (MLIPs) to dramatically accelerate the calculation of photoluminescence (PL) spectra of atomic or molecular emitters with ab initio accuracy. By replacing the costly density functional theory (DFT) computation of phonon modes with much faster MLIP phonon mode calculations, our approach achieves speed improvements exceeding an order of magnitude with minimal precision loss. We benchmark the approach using a dataset comprising ab initio emission spectra of 791 color centers spanning various types of crystal point defects in different charge and magnetic states. The method is also applied to a molecular emitter adsorbed on a hexagonal boron nitride surface. Across all the systems, we find excellent agreement for both the Huang-Rhys factor and the PL lineshapes. This application of universal MLIPs bridges the gap between computational efficiency and spectroscopic fidelity, opening pathways to high-throughput screening of defect-engineered materials. Our work not only demonstrates accelerated calculation of PL spectra with DFT accuracy, but also makes such calculations tractable for more complex materials.

cond-mat.mtrl-sci↗

arXiv2025

LAMBench: A Benchmark for Large Atomistic Models

Large Atomistic Models (LAMs) have undergone remarkable progress recently, emerging as universal or fundamental representations of the potential energy surface defined by the first-principles calculations of atomistic systems. However, our understanding of the extent to which these models achieve true universality, as well as their comparative performance across different models, remains limited. This gap is largely due to the lack of comprehensive benchmarks capable of evaluating the effectiveness of LAMs as approximations to the universal potential energy surface. In this study, we introduce LAMBench, a benchmarking system designed to evaluate LAMs in terms of their generalizability, adaptability, and applicability. These attributes are crucial for deploying LAMs as ready-to-use tools across a diverse array of scientific discovery contexts. We benchmark ten state-of-the-art LAMs released prior to August 1, 2025, using LAMBench. Our findings reveal a significant gap between the current LAMs and the ideal universal potential energy surface. They also highlight the need for incorporating cross-domain training data, supporting multi-fidelity modeling, and ensuring the models' conservativeness and differentiability. As a dynamic and extensible platform, LAMBench is intended to continuously evolve, thereby facilitating the development of robust and generalizable LAMs capable of significantly advancing scientific research. The LAMBench code is open-sourced at https://github.com/deepmodeling/lambench, and an interactive leaderboard is available at https://www.aissquare.com/openlam?tab=Benchmark.

physics.comp-ph↗

arXiv2024

Universal Machine Learning Interatomic Potentials are Ready for Phonons

There has been an ongoing race for the past several years to develop the best universal machinelearning interatomic potential. This progress has led to increasingly accurate models for predictingenergy, forces, and stresses, combining innovative architectures with big data. Here, we benchmarkthese models on their ability to predict harmonic phonon properties, which are critical for under-standing the vibrational and thermal behavior of materials. Using around 10 000 ab initio phononcalculations, we evaluate model performance across various phonon-related parameters to test theuniversal applicability of these models. The results reveal that some models achieve high accuracyin predicting harmonic phonon properties. However, others still exhibit substantial inaccuracies,even if they excel in the prediction of the energy and the forces for materials close to dynamicalequilibrium. These findings highlight the importance of considering phonon-related properties inthe development of universal machine learning interatomic potentials.

cond-mat.mtrl-sci↗