AI BriefingExpanded Skala Accessibility: A Faster Path to Predictive DFT
Key point
Microsoft Research released Skala 1.1 with improved accuracy and expanded integration with major simulation codes.
Details
Microsoft Research released Skala 1.1, a deep learning-based exchange-correlation functional. Trained on 2.5 times more data than previous versions, this model shows significantly improved accuracy in key molecular simulation tasks such as thermochemistry, reaction dynamics, and molecular structure prediction.
Skala 1.1 achieves higher accuracy than the most expensive global hybrid functionals at the computational cost of meta-GGA functionals. It ranked first in 32 of the 55 categories in the widely used GMTKN55 benchmark, with a weighted mean error of 2.8 kcal/mol.
To expand accessibility, Skala is currently available in CP2K, with integrations underway for Psi4, FHI-aims, ORCA, and VASP. This enables scientists and industry to leverage next-generation DFT accuracy in the codes they rely on daily.
Additionally, Microsoft introduced a living benchmark to track the computational performance of continuously optimized Skala releases. This allows the community to measure and accelerate improvements in accuracy and efficiency.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Why LLM Research Agents Generalize Without Benchmark Overfitting: 'Compressibility' Is Key