New AI method enhances accuracy of chemical functional predictions

Researchers have developed a new method called Derivative Informed XC-Loss (DI-Loss) to improve the accuracy of machine-learned exchange-correlation functionals in computational chemistry. This technique incorporates information from the first and second derivatives of energy, leading to a significant reduction in energy errors and faster self-consistent field iterations. The improved functionals also show better performance in predicting excited states in downstream calculations. AI

IMPACT Enhances accuracy and efficiency of AI models used in computational chemistry simulations.

RANK_REASON The cluster contains a research paper detailing a new method for improving machine-learned functionals in computational chemistry. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• B3LYP/def2-SVP
• Derivative Informed XC-Loss
• Machine-learned exchange-correlation functionals

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →