Galmoss: A package for GPU-accelerated galaxy profile fitting

Mi Chen; Rafael S. de Souza; Quanfeng Xu; Shiyin Shen; Ana L. Chies-Santos; Renhao Ye; Marco A. Canossa-Gosteinski; Yanping Cong

doi:10.1016/j.ascom.2024.100825

Galmoss: A package for GPU-accelerated galaxy profile fitting

Mi Chen
, Rafael S. de Souza
, Quanfeng Xu
, Shiyin Shen
, Ana L. Chies-Santos
, Renhao Ye
, Marco A. Canossa-Gosteinski
, Yanping Cong

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

21 Downloads (Pure)

Abstract

We introduce galmoss, a python-based, torch-powered tool for two-dimensional fitting of galaxy profiles. By seamlessly enabling GPU parallelization, galmoss meets the high computational demands of large-scale galaxy surveys, placing galaxy profile fitting in the CSST/LSST-era. It incorporates widely used profiles such as the Sérsic, Exponential disk, Ferrer, King, Gaussian, and Moffat profiles, and allows for the easy integration of more complex models. Tested on 8289 galaxies from the Sloan Digital Sky Survey (SDSS) g-band with a single NVIDIA A100 GPU, galmoss completed classical Sérsic profile fitting in about 10 min. Benchmark tests show that galmoss achieves computational speeds that are 6 × faster than those of default implementations.

Original language	English
Article number	100825
Pages (from-to)	1-11
Number of pages	11
Journal	Astronomy and Computing
Volume	47
Early online date	16 Apr 2024
DOIs	https://doi.org/10.1016/j.ascom.2024.100825
Publication status	Published - 22 Apr 2024

Keywords

GPU computing
machine and deep learning
Extragalactic astronomy
Galaxies
Data analysis – methods
Statistical – GPU computing
General – methods

Access to Document

10.1016/j.ascom.2024.100825Licence: CC BY

Final_Version
© 2024 The Author(s). Published by Elsevier B.V. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY), https://creativecommons.org/licenses/by/4.0/
Final published version, 2.57 MBLicence: CC BY

Cite this

@article{6d1f843091854654b7cc78ee7e59a4f1,

title = "Galmoss: A package for GPU-accelerated galaxy profile fitting",

abstract = "We introduce galmoss, a python-based, torch-powered tool for two-dimensional fitting of galaxy profiles. By seamlessly enabling GPU parallelization, galmoss meets the high computational demands of large-scale galaxy surveys, placing galaxy profile fitting in the CSST/LSST-era. It incorporates widely used profiles such as the S{\'e}rsic, Exponential disk, Ferrer, King, Gaussian, and Moffat profiles, and allows for the easy integration of more complex models. Tested on 8289 galaxies from the Sloan Digital Sky Survey (SDSS) g-band with a single NVIDIA A100 GPU, galmoss completed classical S{\'e}rsic profile fitting in about 10 min. Benchmark tests show that galmoss achieves computational speeds that are 6 × faster than those of default implementations.",

keywords = "GPU computing, machine and deep learning, Extragalactic astronomy, Galaxies, Data analysis – methods, Statistical – GPU computing, General – methods",

author = "Mi Chen and Souza, \{Rafael S. de\} and Quanfeng Xu and Shiyin Shen and Chies-Santos, \{Ana L.\} and Renhao Ye and Canossa-Gosteinski, \{Marco A.\} and Yanping Cong",

note = "{\textcopyright} 2024 The Author(s). Published by Elsevier B.V. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY), https://creativecommons.org/licenses/by/4.0/ ",

year = "2024",

month = apr,

day = "22",

doi = "10.1016/j.ascom.2024.100825",

language = "English",

volume = "47",

pages = "1--11",

journal = "Astronomy and Computing",

issn = "2213-1337",

publisher = "Elsevier",

}

TY - JOUR

T1 - Galmoss: A package for GPU-accelerated galaxy profile fitting

AU - Chen, Mi

AU - Souza, Rafael S. de

AU - Xu, Quanfeng

AU - Shen, Shiyin

AU - Chies-Santos, Ana L.

AU - Ye, Renhao

AU - Canossa-Gosteinski, Marco A.

AU - Cong, Yanping

N1 - © 2024 The Author(s). Published by Elsevier B.V. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY), https://creativecommons.org/licenses/by/4.0/

PY - 2024/4/22

Y1 - 2024/4/22

N2 - We introduce galmoss, a python-based, torch-powered tool for two-dimensional fitting of galaxy profiles. By seamlessly enabling GPU parallelization, galmoss meets the high computational demands of large-scale galaxy surveys, placing galaxy profile fitting in the CSST/LSST-era. It incorporates widely used profiles such as the Sérsic, Exponential disk, Ferrer, King, Gaussian, and Moffat profiles, and allows for the easy integration of more complex models. Tested on 8289 galaxies from the Sloan Digital Sky Survey (SDSS) g-band with a single NVIDIA A100 GPU, galmoss completed classical Sérsic profile fitting in about 10 min. Benchmark tests show that galmoss achieves computational speeds that are 6 × faster than those of default implementations.

AB - We introduce galmoss, a python-based, torch-powered tool for two-dimensional fitting of galaxy profiles. By seamlessly enabling GPU parallelization, galmoss meets the high computational demands of large-scale galaxy surveys, placing galaxy profile fitting in the CSST/LSST-era. It incorporates widely used profiles such as the Sérsic, Exponential disk, Ferrer, King, Gaussian, and Moffat profiles, and allows for the easy integration of more complex models. Tested on 8289 galaxies from the Sloan Digital Sky Survey (SDSS) g-band with a single NVIDIA A100 GPU, galmoss completed classical Sérsic profile fitting in about 10 min. Benchmark tests show that galmoss achieves computational speeds that are 6 × faster than those of default implementations.

KW - GPU computing

KW - machine and deep learning

KW - Extragalactic astronomy

KW - Galaxies

KW - Data analysis – methods

KW - Statistical – GPU computing

KW - General – methods

UR - https://www.scopus.com/pages/publications/85190785726

U2 - 10.1016/j.ascom.2024.100825

DO - 10.1016/j.ascom.2024.100825

M3 - Article

SN - 2213-1337

VL - 47

SP - 1

EP - 11

JO - Astronomy and Computing

JF - Astronomy and Computing

M1 - 100825

ER -

Galmoss: A package for GPU-accelerated galaxy profile fitting

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this