CaltechAUTHORS
  A Caltech Library Service

Nonlinear 3D cosmic web simulation with heavy-tailed generative adversarial networks

Feder, Richard M. and Berger, Philippe and Stein, George (2020) Nonlinear 3D cosmic web simulation with heavy-tailed generative adversarial networks. Physical Review D, 102 (10). Art. No. 103504. ISSN 2470-0010. https://resolver.caltech.edu/CaltechAUTHORS:20200720-100227202

[img] PDF - Published Version
See Usage Policy.

4Mb
[img] PDF - Submitted Version
See Usage Policy.

7Mb

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20200720-100227202

Abstract

Fast and accurate simulations of the nonlinear evolution of the cosmic density field are a major component of many cosmological analyses, but the computational time and storage required to run them can be exceedingly large. For this reason, we use generative adversarial networks (GANs) to learn a compressed representation of the 3D matter density field that is fast and easy to sample, and for the first time show that GANs are capable of generating samples at the level of accuracy of other conventional methods. Using subvolumes from a suite of GADGET-2 N-body simulations, we demonstrate that a deep-convolutional GAN can generate samples that capture both large- and small-scale features of the matter density field, as validated through a variety of n-point statistics. The use of a data scaling that preserves high-density features and a heavy-tailed latent space prior allow us to obtain state of the art results for fast 3D cosmic web generation. In particular, the mean power spectra from generated samples agree to within 5% up to k=3 and within 10% for k≤5 when compared with N-body simulations, and similar accuracy is obtained for a variety of bispectra. By modeling the latent space with a heavy-tailed prior rather than a standard Gaussian, we better capture sample variance in the high-density voxel PDF and reduce errors in power spectrum and bispectrum covariance on all scales. Furthermore, we show that a conditional GAN can smoothly interpolate between samples conditioned on redshift. Deep generative models, such as the ones described in this work, provide great promise as fast, low-memory, high-fidelity forward models of large-scale structure.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1103/PhysRevD.102.103504DOIArticle
https://arxiv.org/abs/2005.03050arXivDiscussion Paper
ORCID:
AuthorORCID
Feder, Richard M.0000-0002-9330-8738
Berger, Philippe0000-0003-3322-3510
Additional Information:© 2020 American Physical Society. Received 24 May 2020; accepted 23 September 2020; published 4 November 2020. R. M. F. is supported by the California Institute of Technology. P. B. was supported by Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. The authors would like to thank Tzu-Ching Chang, Olivier Doré, Michael Albergo, Jeremy Bernstein and Yun-Ting Cheng for useful discussions. The authors acknowledge the Texas Advanced Computing Center (TACC) at The University of Texas at Austin for providing GPU cluster resources that have contributed to the research results reported within this paper: Ref. [55]. The authors also acknowledge the use of the following software for visualization and analysis: h5py, matplotlib, nbodykit [46], numpy [56], powerbox [57], pylians, and pytorch.
Funders:
Funding AgencyGrant Number
CaltechUNSPECIFIED
NASA/JPL/CaltechUNSPECIFIED
Issue or Number:10
Record Number:CaltechAUTHORS:20200720-100227202
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20200720-100227202
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:104444
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:20 Jul 2020 17:19
Last Modified:04 Nov 2020 20:45

Repository Staff Only: item control page