CaltechAUTHORS
  A Caltech Library Service

Learning phase transitions from dynamics

van Nieuwenburg, Evert and Bairey, Eyal and Refael, Gil (2018) Learning phase transitions from dynamics. Physical Review B, 98 (6). Art. No. 060301. ISSN 2469-9950. http://resolver.caltech.edu/CaltechAUTHORS:20180723-093013664

[img] PDF - Published Version
See Usage Policy.

627Kb
[img] PDF - Submitted Version
See Usage Policy.

1363Kb
[img] PDF (contains analysis of the training of the network. We perform network calibration, check for generalization, check the dependence of the critical disorder on the training regimes and discuss the imbalance as a comparison. We also consider the learned...) - Supplemental Material
See Usage Policy.

677Kb

Use this Persistent URL to link to this item: http://resolver.caltech.edu/CaltechAUTHORS:20180723-093013664

Abstract

We propose the use of recurrent neural networks for classifying phases of matter based on the dynamics of experimentally accessible observables. We demonstrate this approach by training recurrent networks on the magnetization traces of two distinct models of one-dimensional disordered and interacting spin chains. The obtained phase diagram for a well-studied model of the many-body localization transition shows excellent agreement with previously known results obtained from time-independent entanglement spectra. For a periodically driven model featuring an inherently dynamical time-crystalline phase, the phase diagram that our network traces coincides with an order parameter for its expected phases.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1103/PhysRevB.98.060301DOIArticle
https://arxiv.org/abs/1712.00450arXivDiscussion Paper
ORCID:
AuthorORCID
van Nieuwenburg, Evert0000-0003-0323-0031
Additional Information:© 2018 American Physical Society. Received 23 December 2017; revised manuscript received 10 April 2018; published 9 August 2018. E.v.N. gratefully acknowledges financial support from the Swiss National Science Foundation through Grant No. P2EZP2-172185. E.v.N. also acknowledges fruitful discussions with Manuel Endres. E.B. is grateful to Netanel Lindner for his support and acknowledges financial support from the European Research Council (ERC) under the European Union Horizon 2020 Research and Innovation Programme (Grant Agreement No. 639172). G.R. is grateful to the the NSF for funding through Grant No. DMR-1040435 as well as the Packard Foundation. We are grateful for support from the IQIM, an NSF physics frontier center funded in part by the Moore Foundation. The authors used the tensorflow [69] backend for keras [70]. E.v.N. and E.B. contributed equally to this work.
Group:IQIM, Institute for Quantum Information and Matter
Funders:
Funding AgencyGrant Number
Swiss National Science Foundation (SNSF)P2EZP2-172185
European Research Council (ERC)639172
NSFDMR-1040435
David and Lucile Packard FoundationUNSPECIFIED
Institute for Quantum Information and Matter (IQIM)UNSPECIFIED
Gordon and Betty Moore FoundationUNSPECIFIED
Record Number:CaltechAUTHORS:20180723-093013664
Persistent URL:http://resolver.caltech.edu/CaltechAUTHORS:20180723-093013664
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:88118
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:23 Jul 2018 16:39
Last Modified:09 Aug 2018 16:07

Repository Staff Only: item control page