Learning a phase diagram from dynamics


van Nieuwenburg, E. (2018). Learning a phase diagram from dynamics. Perimeter Institute. https://pirsa.org/18040132


van Nieuwenburg, Evert. Learning a phase diagram from dynamics. Perimeter Institute, Apr. 23, 2018, https://pirsa.org/18040132


          @misc{ pirsa_18040132,
            doi = {10.48660/18040132},
            url = {https://pirsa.org/18040132},
            author = {van Nieuwenburg, Evert},
            keywords = {Condensed Matter},
            language = {en},
            title = {Learning a phase diagram from dynamics},
            publisher = {Perimeter Institute},
            year = {2018},
            month = {apr},
            note = {PIRSA:18040132 see, \url{https://pirsa.org}}

Evert van Nieuwenburg California Institute of Technology


Time series data contains useful information on the phase of a system. Here we propose the use of recurrent neural networks (LSTM) to learn and extract such information in order to classify phases and locate phase boundaries. We demonstrate this on a many-body localized model, and attempt to interpret the learned behavior by looking at individual LSTM cells. We also discuss the validity of the learned model and investigate its limits.