Double-hybrid quality reactivity in converged quantum environments at GGA-like speed: comparing local correlation and local quantum embedding accelerations

Double-hybrid (DH) density functionals provide the highest rung of DFT accuracy, but their MP2-like term can be computationally prohibitive, when including realistic chemical environments. Here, we extend our local embedded subsystem (LESS) framework, previously introduced for projector- and Huzinaga-equation-based quantum embedding for hybrid DFT-in-GGA, by implementing DH-in-DFT embedding with hybrid or GGA DFT environments. We also combine it with local MP2 (LMP2) and local DH (LDH) acceleration and analyze the performance of the resulting LDH-in-GGA and LESS-accelerated LDH-in-GGA variants compared to LESS DH-in-GGA. Benchmarks are reported for environment-inclusive organo-, enzymatic, and zeolite-catalyzed reactions (171–238 QM atoms), all exhibiting high DFT uncertainties at lower rungs. The LESS acceleration retains only 50–60% of the atomic orbital and 20–40% of the density-fitting basis functions for the high-level DH part, while yields uncertainties of only about 0.1 kcal/mol. Combined with mixed basis sets, LESS DH-in-GGA surpasses the speed of LMP2-based LDH for these realistic environment sizes, requiring 4-8 core hours, i.e., two orders of magnitude speedup over conventional DH DFT. Importantly, the cost of the DH part in LESS DH-in-DFT becomes comparable or less than the full system low-level DFT computation, even with GGA environment models. Our additional motivation emerges from the lower complexity of LESS Huzinaga embedding from the perspective of implementing derivative and spectroscopic properties. This technical benefit, in combination with its outstanding accuracy-over-cost performance, makes LESS-based quantum embedding highly competitive with local correlation as a route to enable DH quality simulations at routine cost, with proper chemical environments.

Authors

Institutions

Publication Details

Journal
ChemRxiv
Published
2026-11-17
DOI
https://doi.org/10.26434/chemrxiv.15012343/v1
Primary Topic
Advanced Chemical Physics Studies
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Double-hybrid quality reactivity in converged quantum environments at GGA-like speed: comparing local correlation and local quantum embedding accelerations

Dénes Berta, József Csóka, Péter R. Nagy, Gyula Samu
4 links
ChemRxiv
Advanced Chemical Physics Studies
preprint

Double-hybrid quality reactivity in converged quantum environments at GGA-like speed: comparing local correlation and local quantum embedding accelerations

Dénes Berta, József Csóka, Péter R. Nagy, Gyula Samu
preprint en

Abstract

Double-hybrid (DH) density functionals provide the highest rung of DFT accuracy, but their MP2-like term can be computationally prohibitive, when including realistic chemical environments. Here, we extend our local embedded subsystem (LESS) framework, previously introduced for projector- and Huzinaga-equation-based quantum embedding for hybrid DFT-in-GGA, by implementing DH-in-DFT embedding with hybrid or GGA DFT environments. We also combine it with local MP2 (LMP2) and local DH (LDH) acceleration and analyze the performance of the resulting LDH-in-GGA and LESS-accelerated LDH-in-GGA variants compared to LESS DH-in-GGA. Benchmarks are reported for environment-inclusive organo-, enzymatic, and zeolite-catalyzed reactions (171–238 QM atoms), all exhibiting high DFT uncertainties at lower rungs. The LESS acceleration retains only 50–60% of the atomic orbital and 20–40% of the density-fitting basis functions for the high-level DH part, while yields uncertainties of only about 0.1 kcal/mol. Combined with mixed basis sets, LESS DH-in-GGA surpasses the speed of LMP2-based LDH for these realistic environment sizes, requiring 4-8 core hours, i.e., two orders of magnitude speedup over conventional DH DFT. Importantly, the cost of the DH part in LESS DH-in-DFT becomes comparable or less than the full system low-level DFT computation, even with GGA environment models. Our additional motivation emerges from the lower complexity of LESS Huzinaga embedding from the perspective of implementing derivative and spectroscopic properties. This technical benefit, in combination with its outstanding accuracy-over-cost performance, makes LESS-based quantum embedding highly competitive with local correlation as a route to enable DH quality simulations at routine cost, with proper chemical environments.

ChemRxiv
Budapest University of Technology and Economics (HU), Hungarian Research Network (HU)
European High Performance Computing Joint Undertaking, Magyar Tudományos Akadémia, Nemzeti Kutatási Fejlesztési és Innovációs Hivatal
Advanced Chemical Physics Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon