skip to main content

DOE PAGESDOE PAGES

Title: Enabling large-scale viscoelastic calculations via neural network acceleration

One of the most significant challenges involved in efforts to understand the effects of repeated earthquake cycle activity is the computational costs of large-scale viscoelastic earthquake cycle models. Computationally intensive viscoelastic codes must be evaluated at thousands of times and locations, and as a result, studies tend to adopt a few fixed rheological structures and model geometries and examine the predicted time-dependent deformation over short (<10 years) time periods at a given depth after a large earthquake. Training a deep neural network to learn a computationally efficient representation of viscoelastic solutions, at any time, location, and for a large range of rheological structures, allows these calculations to be done quickly and reliably, with high spatial and temporal resolutions. We demonstrate that this machine learning approach accelerates viscoelastic calculations by more than 50,000%. Finally, this magnitude of acceleration will enable the modeling of geometrically complex faults over thousands of earthquake cycles across wider ranges of model parameters and at larger spatial and temporal scales than have been previously possible.
Authors:
ORCiD logo [1] ; ORCiD logo [1] ; ORCiD logo [1]
  1. Harvard Univ., Cambridge, MA (United States). Dept. of Earth and Planetary Sciences
Publication Date:
Grant/Contract Number:
FG02-97ER25308; EAR‐1033462; 6239; G12AC20038
Type:
Accepted Manuscript
Journal Name:
Geophysical Research Letters
Additional Journal Information:
Journal Volume: 44; Journal Issue: 6; Journal ID: ISSN 0094-8276
Publisher:
American Geophysical Union
Research Org:
Krell Inst., Ames, IA (United States)
Sponsoring Org:
USDOE National Nuclear Security Administration (NNSA); National Science Foundation (NSF); Southern California Earthquake Center; USGS
Country of Publication:
United States
Language:
English
Subject:
58 GEOSCIENCES
OSTI Identifier:
1465348
Alternate Identifier(s):
OSTI ID: 1402388

DeVries, Phoebe M. R., Thompson, T. Ben, and Meade, Brendan J.. Enabling large-scale viscoelastic calculations via neural network acceleration. United States: N. p., Web. doi:10.1002/2017GL072716.
DeVries, Phoebe M. R., Thompson, T. Ben, & Meade, Brendan J.. Enabling large-scale viscoelastic calculations via neural network acceleration. United States. doi:10.1002/2017GL072716.
DeVries, Phoebe M. R., Thompson, T. Ben, and Meade, Brendan J.. 2017. "Enabling large-scale viscoelastic calculations via neural network acceleration". United States. doi:10.1002/2017GL072716. https://www.osti.gov/servlets/purl/1465348.
@article{osti_1465348,
title = {Enabling large-scale viscoelastic calculations via neural network acceleration},
author = {DeVries, Phoebe M. R. and Thompson, T. Ben and Meade, Brendan J.},
abstractNote = {One of the most significant challenges involved in efforts to understand the effects of repeated earthquake cycle activity is the computational costs of large-scale viscoelastic earthquake cycle models. Computationally intensive viscoelastic codes must be evaluated at thousands of times and locations, and as a result, studies tend to adopt a few fixed rheological structures and model geometries and examine the predicted time-dependent deformation over short (<10 years) time periods at a given depth after a large earthquake. Training a deep neural network to learn a computationally efficient representation of viscoelastic solutions, at any time, location, and for a large range of rheological structures, allows these calculations to be done quickly and reliably, with high spatial and temporal resolutions. We demonstrate that this machine learning approach accelerates viscoelastic calculations by more than 50,000%. Finally, this magnitude of acceleration will enable the modeling of geometrically complex faults over thousands of earthquake cycles across wider ranges of model parameters and at larger spatial and temporal scales than have been previously possible.},
doi = {10.1002/2017GL072716},
journal = {Geophysical Research Letters},
number = 6,
volume = 44,
place = {United States},
year = {2017},
month = {3}
}