2018
DOI: 10.1021/acs.jctc.8b00879
|View full text |Cite
|
Sign up to set email alerts
|

Excited-State Diffusion Monte Carlo Calculations: A Simple and Efficient Two-Determinant Ansatz

Abstract: We perform excited-state variational Monte Carlo and diffusion Monte Carlo calculations using a simple and efficient wave function ansatz. This ansatz follows the recent variation-after-response formalism, accurately approximating a configuration interaction singles wave function as a sum of only two non-orthogonal Slater determinants, and further including important orbital relaxation. The ansatz is used to perform diffusion Monte Carlo calculations with large augmented basis sets, comparing to benchmarks fro… Show more

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
28
0

Year Published

2019
2019
2023
2023

Publication Types

Select...
7
1

Relationship

1
7

Authors

Journals

citations
Cited by 24 publications
(28 citation statements)
references
References 70 publications
0
28
0
Order By: Relevance
“…We used a simple single-determinant scheme to obtain the band gaps from DMC; therefore, our results should be treated as an upper bound due to the fixed node bias, which may not fully cancel between the ground and excited states. Multideterminant wavefunctions can be used to optimize the excited-state nodal surface and control fixed node bias, although studies for extended systems are very limited [131][132][133], because of the significant computational resources that would be required using DMC. A very recent work on VMC, however, shows that nodal surface errors can be minimized using orbital rotations on the singledeterminant wavefunction [134].…”
Section: Electronic Structure Quasiparticle and Optical Gapsmentioning
confidence: 99%
“…We used a simple single-determinant scheme to obtain the band gaps from DMC; therefore, our results should be treated as an upper bound due to the fixed node bias, which may not fully cancel between the ground and excited states. Multideterminant wavefunctions can be used to optimize the excited-state nodal surface and control fixed node bias, although studies for extended systems are very limited [131][132][133], because of the significant computational resources that would be required using DMC. A very recent work on VMC, however, shows that nodal surface errors can be minimized using orbital rotations on the singledeterminant wavefunction [134].…”
Section: Electronic Structure Quasiparticle and Optical Gapsmentioning
confidence: 99%
“…These developments also open very interesting prospects for the application of QMC to geometry relaxation in the excited state, where most electronic structure methods either lack the required accuracy or are computationally quite expensive due to their scaling with system size. To date, there are very few studies to assess the ability of QMC to predict excited-state geometries [22][23][24][25], while most of the relatively limited literature on excited-state QMC calculations is primarily concerned with vertical excitation energies [26][27][28][29][30][31][32][33][34][35][36][37][38][39][40]. Importantly, all these studies are characterized by the use of very different wave functions ranging from the simple ansatz of a CI singles wave function to complete active space (CAS) expansions, sometimes truncated or only partially optimized in the presence of the Jastrow factor due to the limitations previously faced in sampling and optimizing large numbers of determinants.…”
Section: Introductionmentioning
confidence: 99%
“…is observation is in agreement with the recent results of Blunt and Neuscamman on the same system. 125 As pointed out by Hammond and coworkers, 126 when the trial wave function does not include a Jastrow factor, the non-local pseudopotential can be localized analytically and the usual numerical quadrature over the angular part of the non-local pseudopotential can be eschewed. In practice, the calculation of the localized part of the pseudopotential represents only a small overhead (about 15%) with respect to a calculation without pseudopotentials (and the same number of electrons).…”
Section: Basismentioning
confidence: 99%