Meta-generalized gradient approximation made in the Hartree gauge
ORAL
Abstract
In density functional theory (DFT), exact constraints—fundamental mathematical properties of
the exchange-correlation (XC) energy and its underlying XC hole—along with paradigm systems
such as the uniform electron gas and the hydrogen atom have been instrumental in developing
exchange-correlation (XC) density functional approximations (DFAs). However, since the spatial XC
energy density is not uniquely defined, its exact constraints can only be formulated within a chosen
gauge and are therefore seldom utilized in DFA construction. Here, we propose a meta-generalized
gradient approximation for the exchange energy, explicitly constructed within the Hartree gauge,
using the hydrogen atom’s exchange energy density for gauge alignment in core and asymptotic re-
gions. By formulating DFAs at the XC energy density level, this approach expands reference datasets
for machine learning and establishes a foundation for more accurate nonlocal density functionals
requiring gauge alignment.
the exchange-correlation (XC) energy and its underlying XC hole—along with paradigm systems
such as the uniform electron gas and the hydrogen atom have been instrumental in developing
exchange-correlation (XC) density functional approximations (DFAs). However, since the spatial XC
energy density is not uniquely defined, its exact constraints can only be formulated within a chosen
gauge and are therefore seldom utilized in DFA construction. Here, we propose a meta-generalized
gradient approximation for the exchange energy, explicitly constructed within the Hartree gauge,
using the hydrogen atom’s exchange energy density for gauge alignment in core and asymptotic re-
gions. By formulating DFAs at the XC energy density level, this approach expands reference datasets
for machine learning and establishes a foundation for more accurate nonlocal density functionals
requiring gauge alignment.
*We acknowledge the support of the U.S. DOE, Office of Science, Basic Energy Sciences Grant No. DE- SC0019350. This work also used Purdue Anvil CPU at Rosen Center for Advanced Computing (RCAC) through allocation DMR190076 from the Advanced Cyberinfras- tructure Coordination Ecosystem: Services & Support (ACCESS) program, which is supported by National Sci- ence Foundation grants #2138259, #2138286, #2138307, #2137603, and #2138296.
–
Presenters
-
Akilan Ramasamy
- Tulane University