2020
DOI: 10.1007/s00214-020-02601-3
|View full text |Cite
|
Sign up to set email alerts
|

Benchmarking lattice energy of a model 1D molecular HF crystal

Abstract: State-of-the-art benchmark lattice energies of 1D hydrogen fluoride model crystal are presented. Many-body expanded coupled-cluster CCSD(T) extrapolated to the complete basis set, and thermodynamic limit results in −7.5 ± 0.1 kcal/mol per molecule. One-determinant fixed-node diffusion Monte Carlo ( −7.5 ± 0.1 kcal/mol) explicitly confirms its ability to produce competitive results in periodic setting.

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
2

Citation Types

0
7
0

Year Published

2020
2020
2023
2023

Publication Types

Select...
3

Relationship

3
0

Authors

Journals

citations
Cited by 3 publications
(7 citation statements)
references
References 27 publications
0
7
0
Order By: Relevance
“…We ask if δ ( M , D ) can be decomposed, so that its reconstruction δ ( A ' A ″ , B ' B ″ , A ' A ″ B ' B ″ ), or an approximation trueδ˜(),,AABBAABB, can be based on fragments instead of full‐sized models, in a way similar to the techniques used for an interaction energy (IE) decomposition 37–42 . For our dimer complex consisting of four disjoint fragments, an exact decomposition of δ to many‐body non‐additive contributions contains pairs of fragments, triplets and four‐body interactions, and we can write δ=iA,AjB,Bδfalse˜ij2+iA,Aj>iA,A,B,Bk>jB,Bδfalse˜ijk3+iA,Aj>iA,A,B,Bk>jA,A,B,Bl>kB,Bδfalse˜ijkl4, where we require A ′ < A ′′ < B ′ < B ′′ , or briefly, δ=δfalse˜2+…”
Section: Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…We ask if δ ( M , D ) can be decomposed, so that its reconstruction δ ( A ' A ″ , B ' B ″ , A ' A ″ B ' B ″ ), or an approximation trueδ˜(),,AABBAABB, can be based on fragments instead of full‐sized models, in a way similar to the techniques used for an interaction energy (IE) decomposition 37–42 . For our dimer complex consisting of four disjoint fragments, an exact decomposition of δ to many‐body non‐additive contributions contains pairs of fragments, triplets and four‐body interactions, and we can write δ=iA,AjB,Bδfalse˜ij2+iA,Aj>iA,A,B,Bk>jB,Bδfalse˜ijk3+iA,Aj>iA,A,B,Bk>jA,A,B,Bl>kB,Bδfalse˜ijkl4, where we require A ′ < A ′′ < B ′ < B ′′ , or briefly, δ=δfalse˜2+…”
Section: Methodsmentioning
confidence: 99%
“…We ask if δ(M, D) can be decomposed, so that its reconstruction δ(A ' A 00 , B ' B 00 , A ' A 00 B ' B 00 ), or an approximationδ A 0 A 00 , B 0 B 00 , A 0 A 00 B 0 B 00 ð Þ , can be based on fragments instead of full-sized models, in a way similar to the techniques used for an interaction energy (IE) decomposition. [37][38][39][40][41][42] For our dimer complex consisting of four disjoint fragments, an exact decomposition of δ to many-body non-additive contributions contains pairs of fragments, triplets and four-body interactions, and we can write…”
Section: Methodsmentioning
confidence: 99%
“…18,28 For this reason, the scalable SD FNDMC method has been able to compete vs "gold-standard" CCSD(T)/CBS to within 0.1 kcal/mol in a number of small archetypal noncovalent systems, like, e.g., dimers of H 2 O, NH 3 , CH 4 , or HF, 19,27 formaldehyde and formic acid, 38 the etheneethyne complex, 39 or even in extended systems like, e.g., the infinite 1D model crystal of HF molecules. 40 Limits of SD FNDMC in context of NCI start to delineate. For instance, in some systems like, e.g., the HCN dimer (HCN:HCN) 39 or the parallel displaced benzene dimer, 41 the method provides unsigned relative errors (REs) exceeding the high-standard accuracy target required for the "benchmark method" label (2% in RE).…”
Section: Introductionmentioning
confidence: 99%
“…Unless the node is the exact one, an approximate node causes bias in total energy. For instance, in atoms with a fixed N , SD FN bias grows with a nuclear charge Z . , In addition to FN bias, SD FNDMC using effective core potentials (ECPs) also suffers from bias stemming from additional approximations required to evaluate ECPs within FNDMC, namely partial localization (T-moves , ) or full localization (locality approximation, LA) of the nonlocal ECP operator. , Experience shows that the related biases (FN and localization errors) may cancel out in energy differences, so that the remaining errors in FNDMC energy differences become negligible. , For this reason, the scalable SD FNDMC method has been able to compete vs “gold-standard” CCSD­(T)/CBS to within 0.1 kcal/mol in a number of small archetypal noncovalent systems, like, e.g., dimers of H 2 O, NH 3 , CH 4 , or HF, , formaldehyde and formic acid, the ethene-ethyne complex, or even in extended systems like, e.g., the infinite 1D model crystal of HF molecules . Limits of SD FNDMC in context of NCI start to delineate.…”
Section: Introductionmentioning
confidence: 99%
“…In this respect, a promising ab initio approach that can readily treat periodic systems without the need to resort to many-body expansion techniques is the fixed-node diffusion Monte Carlo (FNDMC), a quantum Monte Carlo (QMC) method. 41,[52][53][54][55] An additional benefit of this method for (here essential) noncovalent systems is its accuracy, often comparable to that of the coupled cluster singles, doubles, and perturbative triples CCSD(T) method. 54,56,57 Although h-BN has been thoroughly studied computationally as well as experimentally, for some of its basic characteristics such as the exfoliation energy, no experimental estimates are currently available and theoretical models provide inconsistent results (21-39 meV/ atom).…”
Section: Introductionmentioning
confidence: 99%