Magnitude of pseudopotential localization errors in fixed node diffusion quantum Monte Carlo
Abstract
Growth in computational resources has lead to the application of real space diffusion quantum Monte Carlo to increasingly heavy elements. Although generally assumed to be small, we find that when using standard techniques, the pseudopotential localization error can be large, on the order of an electron volt for an isolated cerium atom. We formally show that the localization error can be reduced to zero with improvements to the Jastrow factor alone, and we define a metric of Jastrow sensitivity that may be useful in the design of pseudopotentials. We employ an extrapolation scheme to extract the bare fixed node energy and estimate the localization error in both the locality approximation and the Tmoves schemes for the Ce atom in charge states 3+/4+. The locality approximation exhibits the lowest Jastrow sensitivity and generally smaller localization errors than Tmoves although the locality approximation energy approaches the localization free limit from above/below for the 3+/4+ charge state. We find that energy minimized Jastrow factors including threebody electronelectronion terms are the most effective at reducing the localization error for both the locality approximation and Tmoves for the case of the Ce atom. Less complex or variance minimized Jastrows are generally less effective. Finally, ourmore »
 Authors:
 ORNL
 Publication Date:
 Research Org.:
 Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
 Sponsoring Org.:
 USDOE
 OSTI Identifier:
 1376370
 Alternate Identifier(s):
 OSTI ID: 1365431
 Grant/Contract Number:
 AC0500OR22725
 Resource Type:
 Journal Article: Accepted Manuscript
 Journal Name:
 Journal of Chemical Physics
 Additional Journal Information:
 Journal Volume: 146; Journal Issue: 24; Journal ID: ISSN 00219606
 Publisher:
 American Institute of Physics (AIP)
 Country of Publication:
 United States
 Language:
 English
 Subject:
 36 MATERIALS SCIENCE; 72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS
Citation Formats
Kent, Paul R., and Krogel, Jaron T. Magnitude of pseudopotential localization errors in fixed node diffusion quantum Monte Carlo. United States: N. p., 2017.
Web. doi:10.1063/1.4986951.
Kent, Paul R., & Krogel, Jaron T. Magnitude of pseudopotential localization errors in fixed node diffusion quantum Monte Carlo. United States. doi:10.1063/1.4986951.
Kent, Paul R., and Krogel, Jaron T. Thu .
"Magnitude of pseudopotential localization errors in fixed node diffusion quantum Monte Carlo". United States.
doi:10.1063/1.4986951.
@article{osti_1376370,
title = {Magnitude of pseudopotential localization errors in fixed node diffusion quantum Monte Carlo},
author = {Kent, Paul R. and Krogel, Jaron T.},
abstractNote = {Growth in computational resources has lead to the application of real space diffusion quantum Monte Carlo to increasingly heavy elements. Although generally assumed to be small, we find that when using standard techniques, the pseudopotential localization error can be large, on the order of an electron volt for an isolated cerium atom. We formally show that the localization error can be reduced to zero with improvements to the Jastrow factor alone, and we define a metric of Jastrow sensitivity that may be useful in the design of pseudopotentials. We employ an extrapolation scheme to extract the bare fixed node energy and estimate the localization error in both the locality approximation and the Tmoves schemes for the Ce atom in charge states 3+/4+. The locality approximation exhibits the lowest Jastrow sensitivity and generally smaller localization errors than Tmoves although the locality approximation energy approaches the localization free limit from above/below for the 3+/4+ charge state. We find that energy minimized Jastrow factors including threebody electronelectronion terms are the most effective at reducing the localization error for both the locality approximation and Tmoves for the case of the Ce atom. Less complex or variance minimized Jastrows are generally less effective. Finally, our results suggest that further improvements to Jastrow factors and trial wavefunction forms may be needed to reduce localization errors to chemical accuracy when medium core pseudopotentials are applied to heavy elements such as Ce.},
doi = {10.1063/1.4986951},
journal = {Journal of Chemical Physics},
number = 24,
volume = 146,
place = {United States},
year = {Thu Jun 22 00:00:00 EDT 2017},
month = {Thu Jun 22 00:00:00 EDT 2017}
}
Web of Science

Benchmarking the pseudopotential and fixednode approximations in diffusion Monte Carlo calculations of molecules and solids
We performed diffusion Monte Carlo (DMC) calculations of the spectroscopic properties of a large set of molecules, assessing the effect of different approximations. In systems containing elements with large atomic numbers, we show that the errors associated with the use of nonlocal meanfieldbased pseudopotentials in DMC calculations can be significant and may surpass the fixednode error. In conclusion, we suggest practical guidelines for reducing these pseudopotential errors, which allow us to obtain DMCcomputed spectroscopic parameters of molecules and equation of state properties of solids in excellent agreement with experiment.Cited by 11 
Benchmarking the pseudopotential and fixednode approximations in diffusion Monte Carlo calculations of molecules and solids
We performed diffusion Monte Carlo (DMC) calculations of the spectroscopic properties of a large set of molecules, assessing the effect of different approximations. In systems containing elements with large atomic numbers, we show that the errors associated with the use of nonlocal meanfieldbased pseudopotentials in DMC calculations can be significant and may surpass the fixednode error. In conclusion, we suggest practical guidelines for reducing these pseudopotential errors, which allow us to obtain DMCcomputed spectroscopic parameters of molecules and equation of state properties of solids in excellent agreement with experiment.Cited by 11 
Communication: Fixednode errors in quantum Monte Carlo: Interplay of electron density and node nonlinearities
We elucidate the origin of large differences (twofold or more) in the fixednode errors between the first vs secondrow systems for singleconfiguration trial wave functions in quantum Monte Carlo calculations. This significant difference in the valence fixednode biases is studied across a set of atoms, molecules, and also Si, C solid crystals. We show that the key features which affect the fixednode errors are the differences in electron density and the degree of node nonlinearity. The findings reveal how the accuracy of the quantum Monte Carlo varies across a variety of systems, provide new perspectives on the origins of themore »