THOR AI revolutionizes material physics with lightning-fast computations

Researchers from the University of New Mexico and Los Alamos National Laboratory have unveiled a groundbreaking AI system called THOR. The new computational method tackles complex material physics problems with unprecedented speed. Early tests show it can deliver results in seconds that once took thousands of hours to compute.

THOR AI uses tensor network techniques to simplify intricate calculations in material science. By breaking down problems into smaller, manageable parts, it avoids the 'curse of dimensionality' that has long hindered direct computations of configurational integrals—key equations for modelling particle interactions and predicting material properties.

The system also employs tensor interpolation and automatically detects symmetries in crystalline structures. This allows it to analyse materials like copper, high-pressure argon, and tin during phase transitions. Remarkably, its results match those from far more resource-heavy simulations but are computed up to 400 times faster. Beyond standalone performance, THOR AI can integrate with modern machine learning models. This flexibility enables deeper analysis of how materials behave under different physical conditions. The breakthrough arrives as global research accelerates in AI-driven material science, with institutions like MIT, ETH Zurich, and Stanford advancing similar projects between 2024 and 2026.

The development of THOR AI could transform how new materials are discovered and studied. By drastically reducing computation time without sacrificing accuracy, it opens doors for faster innovation in physics, chemistry, and engineering. Research teams worldwide are now exploring ways to apply the method to real-world challenges.