We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cond-mat.str-el

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Condensed Matter > Strongly Correlated Electrons

Title: Neural network wave functions and the sign problem

Abstract: Neural quantum states (NQS) are a promising approach to study many-body quantum physics. However, they face a major challenge when applied to lattice models: Convolutional networks struggle to converge to ground states with a nontrivial sign structure. We tackle this problem by proposing a neural network architecture with a simple, explicit, and interpretable phase ansatz, which can robustly represent such states and achieve state-of-the-art variational energies for both conventional and frustrated antiferromagnets. In the latter case, our approach uncovers low-energy states that exhibit the Marshall sign rule and are therefore inconsistent with the expected ground state. Such states are the likely cause of the obstruction for NQS-based variational Monte Carlo to access the true ground states of these systems. We discuss the implications of this observation and suggest potential strategies to overcome the problem.
Comments: 5+7 pages, 3+4 figures, with supplementary material
Subjects: Strongly Correlated Electrons (cond-mat.str-el); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG); Computational Physics (physics.comp-ph); Quantum Physics (quant-ph)
Cite as: arXiv:2002.04613 [cond-mat.str-el]
  (or arXiv:2002.04613v1 [cond-mat.str-el] for this version)

Submission history

From: Attila Szabó [view email]
[v1] Tue, 11 Feb 2020 19:00:01 GMT (613kb,D)
[v2] Fri, 21 Feb 2020 12:35:40 GMT (609kb,D)

Link back to: arXiv, form interface, contact.