arXiv subjects

Kareem Abdelmaqsoud

Publications and source records attributed to Kareem Abdelmaqsoud.

5 recordsLinked to original sources

arXiv2026

UMA: A Family of Universal Models for Atoms

The ability to quickly and accurately compute properties from atomic simulations is critical for advancing a large number of applications in chemistry and materials science including drug discovery, energy storage, and semiconductor manufacturing. To address this need, Meta FAIR presents a family of Universal Models for Atoms (UMA), designed to push the frontier of speed, accuracy, and generalization. UMA models are trained on half a billion unique 3D atomic structures (the largest training runs to date) by compiling data across multiple chemical domains, e.g. molecules, materials, and catalysts. We develop empirical scaling laws to help understand how to increase model capacity alongside dataset size to achieve the best accuracy. The UMA small and medium models utilize a novel architectural design we refer to as mixture of linear experts that enables increasing model capacity without sacrificing speed. For example, UMA-medium has 1.4B parameters but only ~50M active parameters per atomic structure. We evaluate UMA models on a diverse set of applications across multiple domains and find that, remarkably, a single model without any fine-tuning can perform similarly or better than specialized models. We are releasing the UMA code, weights, and associated data to accelerate computational workflows and enable the community to continue to build increasingly capable AI models.

cs.LG↗

arXiv2026

Computational Design of Ductile Additively Manufactured Tungsten-Based Refractory Alloys

Tungsten exhibits exceptional temperature and radiation resistance, making it well-suited for applications in extreme environments such as nuclear fusion reactors. Additive manufacturing offers geometrical design freedom and rapid prototyping capabilities for these applications, provided the intrinsic brittleness and low printability of tungsten can be overcome. Designing tungsten alloys with improved ductility, and thus printability in additive manufacturing, can be accelerated using a computationally derived performance predictor to screen out brittle compositions. Calculations of the Pugh ratio using density functional theory may serve this purpose, given its correlation with ductility. This process can be made more efficient through the use of machine learning interatomic potentials to accelerate density functional theory calculations. Here, we demonstrate that machine learning interatomic potentials can effectively identify optimal alloy compositions in the W-Ta-Nb system along the melting point-Pugh ratio Pareto front. The trend in Pugh ratio as a function of tungsten fraction is explained in terms of the electronic density of states at the Fermi level. Experimental validation reveals a strong correlation between the computed Pugh ratio and the observed crack fractions in additively manufactured alloys. Notably, the two alloys predicted to have the highest Pugh ratio values, W20Ta70Nb10 and W30Ta60Nb10, exhibit no intergranular microcracking in experiments.

cond-mat.mtrl-sci↗

arXiv2026

Electronic structure and elasticity of the Ta-W solid solution

The brittleness or ductility of metals has long been attributed to their elastic constants, with high Poisson ratio, or equivalently high Pugh ratio, favoring greater ductility. Growing evidence links ductility with their electronic structure. Consequently, it is desirable to understand how the electronic structure affects the elastic constants. Here, we examine the Ta-W binary alloy system, which evolves from ductile character at Ta-rich compositions to brittleness at high W. We show that a change in slope of the composition-dependent shear modulus near the equiatomic composition coincides with an abrupt change in the Fermi level density of states. We relate the behaviors of the elastic constants to the characters of occupied electronic orbitals close to the Fermi level. Finally, we consider additional alloy systems from groups V and VI and show that qualitatively similar behavior occurs more broadly.

cond-mat.mtrl-sci↗

arXiv2025

The Open DAC 2025 Dataset for Sorbent Discovery in Direct Air Capture

Identifying useful sorbent materials for direct air capture (DAC) from humid air remains a challenge. We present the Open DAC 2025 (ODAC25) dataset, a significant expansion and improvement upon ODAC23 (Sriram et al., ACS Central Science, 10 (2024) 923), comprising nearly 60 million DFT single-point calculations for CO$_2$, H$_2$O, N$_2$, and O$_2$ adsorption in 15,000 MOFs. ODAC25 introduces chemical and configurational diversity through functionalized MOFs, high-energy GCMC-derived placements, and synthetically generated frameworks. ODAC25 also significantly improves upon the accuracy of DFT calculations and the treatment of flexible MOFs in ODAC23. Along with the dataset, we release new state-of-the-art machine-learned interatomic potentials trained on ODAC25 and evaluate them on adsorption energy and Henry's law coefficient predictions.

cond-mat.mtrl-sci↗

arXiv2025

Uncertainty Quantification in Graph Neural Networks with Shallow Ensembles

Machine-learned potentials (MLPs) have revolutionized materials discovery by providing accurate and efficient predictions of molecular and material properties. Graph Neural Networks (GNNs) have emerged as a state-of-the-art approach due to their ability to capture complex atomic interactions. However, GNNs often produce unreliable predictions when encountering out-of-domain data and it is difficult to identify when that happens. To address this challenge, we explore Uncertainty Quantification (UQ) techniques, focusing on Direct Propagation of Shallow Ensembles (DPOSE) as a computationally efficient alternative to deep ensembles. By integrating DPOSE into the SchNet model, we assess its ability to provide reliable uncertainty estimates across diverse Density Functional Theory datasets, including QM9, OC20, and Gold Molecular Dynamics. Our findings often demonstrate that DPOSE successfully distinguishes between in-domain and out-of-domain samples, exhibiting higher uncertainty for unobserved molecule and material classes. This work highlights the potential of lightweight UQ methods in improving the robustness of GNN-based materials modeling and lays the foundation for future integration with active learning strategies.

cs.LG↗