Story perspectives

PRDNet and PRIME Revolutionize Crystalline Material Predictions

10/2/2025

43 8

Summary

Story summary

PRDNet and PRIME predict crystalline material properties, outperforming existing methods.
PRDNet uses diffraction data and graph methods to capture long-range atomic interactions, achieving state-of-the-art results.
It predicts formation energy with 0.028 eV/atom on Materials Project and 0.032 eV/atom on JARVIS-DFT.
PRIME combines real-space and reciprocal-space representations to predict formation energy, band gap, and mechanical properties with improved accuracy.

1 of 1

Story summary

PRDNet and PRIME predict crystalline material properties, outperforming existing methods.
PRDNet uses diffraction data and graph methods to capture long-range atomic interactions, achieving state-of-the-art results.
It predicts formation energy with 0.028 eV/atom on Materials Project and 0.032 eV/atom on JARVIS-DFT.
PRIME combines real-space and reciprocal-space representations to predict formation energy, band gap, and mechanical properties with improved accuracy.