PulseAugur
EN
LIVE 09:57:50

Machine learning for DFTB: Transferability of functionals analyzed

By PulseAugur Editorial · [1 sources] ·

A new research paper investigates the transferability of machine-learned exchange-correlation functionals to density-functional tight binding (DFTB) methods. The study reveals that improvements in machine-learned functionals do not always directly translate to better DFTB parameterizations, particularly for covalent semiconductors. The researchers found that the transferability depends on the specific material properties and the nature of the potential, with some properties inheriting better than others. They propose using a 'transfer ratio' as a preliminary test to predict the success of parameterization campaigns. AI

IMPACT This research clarifies limitations in applying ML improvements to DFTB, potentially guiding future development of more accurate and efficient materials simulation methods.

RANK_REASON The cluster contains a scientific paper detailing novel research findings in computational materials science. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

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

Machine learning for DFTB: Transferability of functionals analyzed

COVERAGE [1]

  1. arXiv cs.LG TIER_1 English(EN) · Can Polat, Mustafa Kurban, Erchin Serpedin, Hasan Kurban ·

    When do machine-learned exchange-correlation improvements inherit into density-functional tight binding?

    arXiv:2608.14875v1 Announce Type: cross Abstract: Machine-learned exchange-correlation functionals correct band gaps at near-semilocal cost, while density-functional tight binding reaches the $10^3$-$10^6$-atom regime; combining them assumes that a better parent yields a better p…

RELATED TOPICS