Making the most of data: Quantum Monte Carlo postanalysis revisited.
View / Open Files
Authors
Ichibha, Tom
Neufeld, Verena A
Hongo, Kenta
Maezono, Ryo
Thom, Alex JW
Publication Date
2022-04Journal Title
Phys Rev E
ISSN
2470-0045
Publisher
American Physical Society (APS)
Type
Article
This Version
VoR
Metadata
Show full item recordCitation
Ichibha, T., Neufeld, V. A., Hongo, K., Maezono, R., & Thom, A. J. (2022). Making the most of data: Quantum Monte Carlo postanalysis revisited.. Phys Rev E https://doi.org/10.1103/PhysRevE.105.045313
Abstract
In quantum Monte Carlo (QMC) methods, energy estimators are calculated as (functions of) statistical averages of quantities sampled during a calculation. Associated statistical errors of these averages are often estimated. This error estimation is not straightforward and there are several choices of the error estimation methods. We evaluate the performance of three methods (the Straatsma method, an autoregressive model, and a blocking analysis based on von Neumann's ratio test for randomness) for the energy time series given by three QMC methods [diffusion Monte Carlo, full configuration interaction Quantum Monte Carlo (FCIQMC), and coupled cluster Monte Carlo (CCMC)]. From these analyses, we describe a hybrid analysis method which provides reliable error estimates for a series of various lengths of FCIQMC and CCMC's time series. Equally important is the estimation of the appropriate start point of the equilibrated phase. We establish that a simple mean squared error rule method as described by White [K. P. White, Jr., Simulation 69(6), 323 (1997)10.1177/003754979706900601] can provide reasonable estimations.
Keywords
physics.comp-ph, physics.comp-ph
Sponsorship
Royal Society (uf110161)
Royal Society (UF160398)
Engineering and Physical Sciences Research Council (EP/L015552/1)
Identifiers
External DOI: https://doi.org/10.1103/PhysRevE.105.045313
This record's URL: https://www.repository.cam.ac.uk/handle/1810/335041
Statistics
Total file downloads (since January 2020). For more information on metrics see the
IRUS guide.