Self-Healing Diffusion Monte Carlo applied to a simple fermionic model: A critical assessment of the method

Abstract

We investigate the Self-Healing Diffusion Monte Carlo (SHDMC) method using a one-dimensional model with periodic boundary conditions. An inversion symmetry is introduced to mimic the antisymmetry property of fermionic wave functions, with the bosonic and fermionic sectors being modeled by the even and odd eigenstates, respectively. As in realistic fermionic systems, the nodal structure is only partially constrained by symmetry, making this model a non-trivial testbed for nodal optimization algorithms such as SHDMC. We show that the nodal evolution under SHDMC iterations can be cast into a dynamical system exhibiting both attractive and repulsive fixed points. When applying SHDMC to this model in the absence of statistical noise, the fixed-node energy is found to increase during iterations and the node converges to an incorrect value. This occurs for any finite basis, even when the exact node is representable in it; the attractive fixed point approaches the exact node only as the basis becomes complete, while the convergence toward it becomes increasingly slow. We further show that this problem can be largely cured by modifying the nodal update criterion to give more importance to the nodal region. Our results suggest that, in the general case, an efficient nodal optimization may require both an improved nodal update criterion and a localized basis set allowing local distortions of the trial wave function, as is the case for this model.

The critical slowing down in training diffusion models
Luca Maria Del Bono, Giulio Biroli, Patrick Charbonneau +1
arXiv18 Sep 2026PreprintarXiv licenceFull textComputational physics Condensed matterarXiv
Computable fermionic non-Gaussianity from the covariance matrix
Poetri Sonya Tarabunga, B. Jobst, Raúl Morral-Yepes +4
arXiv (Cornell University)2 Jul 2026PreprintLicence not statedFull text19 citedAtomic, molecular & optical physicsOpenAlex
Lagrangian-based model of the e⁺e^- → V V̅ process applied to the e⁺e^- → D^∗+ D^∗− data
Peter Lichard
arXiv29 Sep 2026PreprintCC BY 4.0Full textHigh-energy & particle physicsarXiv
Improved classical optimization of fermionic Hamiltonians beyond Gaussian states
Ojas Parekh
arXiv9 Oct 2026PreprintarXiv licenceFull textQuantum physics & quantum informationarXiv
Quadratic Hamiltonian approach to heat transport in fermionic systems
Ilari Mäkinen, Ivan M Khaymovich, Jukka Pekola
Physical Review B29 Sep 2026Peer-reviewedCC BY 4.0Full text1 citedQuantum physics & quantum informationHAL

Cite

BibTeX RIS

BibTeX

@misc{caffarel2026selfhealing,
  title = {{Self-Healing Diffusion Monte Carlo applied to a simple fermionic model: A critical assessment of the method}},
  author = {Michel Caffarel and Manon Pinar and Anthony Scemama},
  howpublished = {HAL},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2609.01301},
  eprint = {2609.01301},
  archivePrefix = {arXiv},
  url = {https://hal.science/hal-05734618v2},
  note = {CC BY 4.0},
}

RIS

TY  - UNPB
TI  - Self-Healing Diffusion Monte Carlo applied to a simple fermionic model: A critical assessment of the method
AU  - Michel Caffarel
AU  - Manon Pinar
AU  - Anthony Scemama
PY  - 2026
DA  - 2026/10/02
PB  - HAL
DO  - 10.48550/arxiv.2609.01301
UR  - https://hal.science/hal-05734618v2
AB  - We investigate the Self-Healing Diffusion Monte Carlo (SHDMC) method using a one-dimensional model with periodic boundary conditions. An inversion symmetry is introduced to mimic the antisymmetry property of fermionic wave functions, with the bosonic and fermionic sectors being modeled by the even and odd eigenstates, respectively. As in realistic fermionic systems, the nodal structure is only partially constrained by symmetry, making this model a non-trivial testbed for nodal optimization algorithms such as SHDMC. We show that the nodal evolution under SHDMC iterations can be cast into a dynamical system exhibiting both attractive and repulsive fixed points. When applying SHDMC to this model in the absence of statistical noise, the fixed-node energy is found to increase during iterations and the node converges to an incorrect value. This occurs for any finite basis, even when the exact node is representable in it; the attractive fixed point approaches the exact node only as the basis becomes complete, while the convergence toward it becomes increasingly slow. We further show that this problem can be largely cured by modifying the nodal update criterion to give more importance to the nodal region. Our results suggest that, in the general case, an efficient nodal optimization may require both an improved nodal update criterion and a localized basis set allowing local distortions of the trial wave function, as is the case for this model.
ER  -

Abstract

Related

Cite