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

Download:

Current browse context:

cond-mat.mtrl-sci

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 > Materials Science

Title: Structural phase transition of two-dimensional monolayer SnTe from artificial neural network

Abstract: As machine learning becomes increasingly important in engineering and science, it is inevitable that machine learning techniques will be applied to the investigation of materials, and in particular the structural phase transitions common in ferroelectric materials. Here, we build and train an artificial neural network to accurately predict the energy change associated with atom displacements and use the trained artificial neural network in Monte-Carlo simulations on ferroelectric materials to investigate their phase transitions. We apply this approach to two-dimensional monolayer SnTe and show that it can indeed be used to simulate the phase transitions and predict the transition temperature. The artificial neural network, when viewed as a universal mathematical structure, can be readily transferred to the investigation of other ferroelectric materials when training data generated with ab initio methods are available.
Comments: 5 pages, 5 figures
Subjects: Materials Science (cond-mat.mtrl-sci); Computational Physics (physics.comp-ph)
Cite as: arXiv:2012.11137 [cond-mat.mtrl-sci]
  (or arXiv:2012.11137v2 [cond-mat.mtrl-sci] for this version)

Submission history

From: Dawei Wang [view email]
[v1] Mon, 21 Dec 2020 06:33:53 GMT (430kb,D)
[v2] Mon, 29 Mar 2021 03:43:05 GMT (446kb,D)

Link back to: arXiv, form interface, contact.