Importance-Reweighted Fock-Space Variational Monte Carlo
Abstract
Fock-space variational Monte Carlo (FS-VMC) evaluates variational quantities by sampling discrete many-body configurations. In molecular applications, concentrated Born distributions can hinder Markov-chain mixing, while Monte Carlo estimators can also exhibit large variance. We introduce importance-reweighted FS-VMC (IR-FS-VMC), which combines Metropolis--Hastings sampling, an evaluation kernel, and self-normalized importance sampling to estimate the Born-distribution energy, gradient, and stochastic reconfiguration (SR) matrix. Controlled comparisons on equilibrium H2O and Fe2S2 show that the acceptance rate and local-energy variance can respond differently to the sampling and estimation procedure. A single production protocol yields accurate variational energies for H2O dissociation, 36-site hydrogen lattices, and Fe2S2 and Fe4S4 active spaces without system-specific sampling parameters.
Related
- NucleoNet: A Nuclear-Aware Determinant Ansatz for Multi-Geometry Neural Variational Monte Carlo
- Kinetic Monte Carlo simulations of Ge–Sb–Te alloy crystallization in confined PCRAM cells
- Mechanistic Monte Carlo assessment of colocalized permeation enhancement and efflux modulation for oral peptide delivery
- Deformation-Aware Monte Carlo Assessment of a Bi2O3/Thermoplastic Polyurethane Shielding Surrogate Under Tensile Strain and Local Thickness Loss
- Experimental and Monte Carlo Insights into the Biofilm Control of Klebsiella pneumoniae Using Green-Synthesized Ag/Ni(OH)2/Clove Nanocomposites
BibTeX
@misc{che2026importancereweighted,
title = {{Importance-Reweighted Fock-Space Variational Monte Carlo}},
author = {Zheng Che},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2608.22925},
eprint = {2608.22925},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.22925},
}
RIS
TY - UNPB
TI - Importance-Reweighted Fock-Space Variational Monte Carlo
AU - Zheng Che
PY - 2026
DA - 2026/09/16
PB - arXiv
DO - 10.48550/arxiv.2608.22925
UR - https://arxiv.org/abs/2608.22925
AB - Fock-space variational Monte Carlo (FS-VMC) evaluates variational quantities by sampling discrete many-body configurations. In molecular applications, concentrated Born distributions can hinder Markov-chain mixing, while Monte Carlo estimators can also exhibit large variance. We introduce importance-reweighted FS-VMC (IR-FS-VMC), which combines Metropolis--Hastings sampling, an evaluation kernel, and self-normalized importance sampling to estimate the Born-distribution energy, gradient, and stochastic reconfiguration (SR) matrix. Controlled comparisons on equilibrium H2O and Fe2S2 show that the acceptance rate and local-energy variance can respond differently to the sampling and estimation procedure. A single production protocol yields accurate variational energies for H2O dissociation, 36-site hydrogen lattices, and Fe2S2 and Fe4S4 active spaces without system-specific sampling parameters.
ER -