News from Japan

Updates every hour. Last Updated: 27-Oct-2026 02:16 ET (27-Oct-2026 06:16 GMT/UTC)

27-Apr-2026

 

Machine learning offers faster, more reliable analysis of Fermi surfaces

Tokyo University of Science
Peer-Reviewed Publication

Interpreting Fermi surface data requires specialized expertise and is vulnerable to noise, creating a bottleneck in materials research. Researchers now demonstrate a machine learning approach that can analyze simulated Fermi surface images of the alloy Co₂MnGa_xGe_1-x and automatically identify compositions where the Fermi surface morphology changes, even in blurred or noisy data. The method reduces the need for manual screening and could accelerate the discovery of advanced electronic materials.

Journal

Scientific Reports

Funder

Japan Science and Technology Agency, Japan Society for the Promotion of Science