Research community previously filed under “Materials Chemistry”

Machine Learning Potentials and Altermagnets in Computational Materials Science

Mostly in the field High-Pressure Physics, Topological Materials, and Computational Potentials (44% of its papers), one of 4 communities there

Papers 13,424
Elements —
Keywords & sectors 2
Leading venue arXiv (Cornell University)

Machine Learning Potentials and Altermagnets in Computational Materials Science

This community develops artificial intelligence models to simulate atomic interactions and predict the properties of magnetic materials, bridging the gap between quantum mechanics and large-scale engineering simulations.

The work focuses on creating "interatomic potentials" and "force fields" that allow researchers to run "molecular dynamics" simulations at scales impossible for traditional "density functional theory." A significant portion of the research applies these tools to "two-dimensional" and "transition metal" systems, specifically investigating "magnetic anisotropy," "spin" behavior, and "altermagnetism." The recurring phrases highlight a dual focus: the development of "neural network" architectures like "graph neural networks" and "large language models" for data extraction, and their application to "electronic structure" prediction and "crystal structure" design. This enables the simulation of complex phenomena such as "phase transition" and "anomalous Hall" effects in materials ranging from "thin films" to "energetic materials."

The community’s output is distributed across a wide range of elements, with the highest concentration found in carbon research, which accounts for 10.79% of the community's papers, and hydrogen, at 7.51%. These elements are central to the organic and inorganic compounds being modeled. The total corpus comprises 13,424 papers, published most frequently in Physical Review B, The Journal of Physical Chemistry C, and npj Computational Materials.

Recent work continues to refine these computational tools, with new papers detailing "neural network potentials" for "energetic materials" and "graph atomic cluster expansion" for "machine learning interatomic potentials." Other recent studies apply these methods to "altermagnetic" systems, such as "2D d-wave altermagnet" spin sensing and "chiral magnons" in metallic structures, demonstrating the ongoing integration of advanced magnetic physics with AI-driven simulation.

Recurring themes in the literature

Phrases that recur across this community's paper titles -- a quick map of its sub-topics, drawn straight from the titles themselves.

  • machine learning 827
  • density functional 248
  • molecular dynamics 224
  • functional theory 193
  • neural network 183
  • electronic structure 147
  • transition metal 145
  • interatomic potentials 132
  • thin films 113
  • crystal structure 105
  • magnetic anisotropy 98
  • deep learning 95
  • neural networks 94
  • waals magnet 90
  • graph neural 86
  • learning interatomic 85
  • artificial intelligence 79
  • active learning 76
  • electronic magnetic 74
  • first-principles calculations 73
  • room temperature 71
  • large language 70
  • force fields 69
  • anomalous hall 69

Papers behind this description

Most cited
Nature Machine Intelligence · 2024 · 736 citations
Nature · 2023 · 930 citations
Nature Communications · 2024 · 724 citations
Nature Machine Intelligence · 2023 · 908 citations
Nature · 2024 · 678 citations
Newest
npj Computational Materials · 2025 · 56 citations
npj Computational Materials · 2026 · 35 citations
Computational Condensed Matter · 2025 · 37 citations
Nature Catalysis · 2025 · 35 citations

A sample from the 16 papers behind this description. Create a free account to see them all.