A hierarchy of normalizing flows for modelling the galaxy-halo relationship

Christopher C. Lovell; Sultan Hassan; Daniel Anglés-Alcázar; Greg Bryan; G. Fabbian; Shy Genel; ChangHoon Hahn; Kartheik Iyer; James Kwon; Natalí de Santi; Francisco Villaescusa-Navarro

doi:10.48550/arXiv.2307.06967

A hierarchy of normalizing flows for modelling the galaxy-halo relationship

Christopher C. Lovell, Sultan Hassan, Daniel Anglés-Alcázar, Greg Bryan, G. Fabbian, Shy Genel, ChangHoon Hahn, Kartheik Iyer, James Kwon, Natalí de Santi, Francisco Villaescusa-Navarro

Institute of Cosmology & Gravitation

Research output: Contribution to conference › Paper › peer-review

108 Downloads (Pure)

Abstract

Using a large sample of galaxies taken from the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project, a suite of hydrodynamic simulations varying both cosmological and astrophysical parameters, we train a normalizing flow (NF) to map the probability of various galaxy and halo properties conditioned on astrophysical and cosmological parameters. By leveraging the learnt conditional relationships we can explore a wide range of interesting questions, whilst enabling simple marginalisation over nuisance parameters. We demonstrate how the model can be used as a generative model for arbitrary values of our conditional parameters; we generate halo masses and matched galaxy properties, and produce realisations of the halo mass function as well as a number of galaxy scaling relations and distribution functions. The model represents a unique and flexible approach to modelling the galaxy-halo relationship.

Original language	English
Number of pages	8
DOIs	https://doi.org/10.48550/arXiv.2307.06967
Publication status	Published - 13 Jun 2023
Event	Machine Learning for Astrophysics Workshop at the Fortieth International Conference on Machine Learning (ICML 2023) - Hawaii Duration: 29 Jul 2023 → 29 Jul 2023 https://ml4astro.github.io/icml2023/#rationale

Conference

Conference	Machine Learning for Astrophysics Workshop at the Fortieth International Conference on Machine Learning (ICML 2023)
Abbreviated title	ICML 2023
Period	29/07/23 → 29/07/23
Internet address	https://ml4astro.github.io/icml2023/#rationale

Keywords

astro-ph.GA

Access to Document

10.48550/arXiv.2307.06967

2307.06967v1Accepted author manuscript (Post-print), 0.99 MBLicence: CC BY

Cite this

Lovell, C. C., Hassan, S., Anglés-Alcázar, D., Bryan, G., Fabbian, G., Genel, S., Hahn, C., Iyer, K., Kwon, J., Santi, N. D., & Villaescusa-Navarro, F. (2023). A hierarchy of normalizing flows for modelling the galaxy-halo relationship. Paper presented at Machine Learning for Astrophysics Workshop at the Fortieth International Conference on Machine Learning (ICML 2023). https://doi.org/10.48550/arXiv.2307.06967

@conference{3be1f88c099f4308b3a0aa519d9d9f83,

title = "A hierarchy of normalizing flows for modelling the galaxy-halo relationship",

abstract = " Using a large sample of galaxies taken from the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project, a suite of hydrodynamic simulations varying both cosmological and astrophysical parameters, we train a normalizing flow (NF) to map the probability of various galaxy and halo properties conditioned on astrophysical and cosmological parameters. By leveraging the learnt conditional relationships we can explore a wide range of interesting questions, whilst enabling simple marginalisation over nuisance parameters. We demonstrate how the model can be used as a generative model for arbitrary values of our conditional parameters; we generate halo masses and matched galaxy properties, and produce realisations of the halo mass function as well as a number of galaxy scaling relations and distribution functions. The model represents a unique and flexible approach to modelling the galaxy-halo relationship. ",

keywords = "astro-ph.GA",

author = "Lovell, \{Christopher C.\} and Sultan Hassan and Daniel Angl{\'e}s-Alc{\'a}zar and Greg Bryan and G. Fabbian and Shy Genel and ChangHoon Hahn and Kartheik Iyer and James Kwon and Santi, \{Natal{\'i} de\} and Francisco Villaescusa-Navarro",

note = "8 pages, 2 figures, accepted for ICML 2023 Workshop on Machine Learning for Astrophysics; Machine Learning for Astrophysics Workshop at the Fortieth International Conference on Machine Learning (ICML 2023), ICML 2023 ; Conference date: 29-07-2023 Through 29-07-2023",

year = "2023",

month = jun,

day = "13",

doi = "10.48550/arXiv.2307.06967",

language = "English",

url = "https://ml4astro.github.io/icml2023/\#rationale",

}

Lovell, CC, Hassan, S, Anglés-Alcázar, D, Bryan, G, Fabbian, G, Genel, S, Hahn, C, Iyer, K, Kwon, J, Santi, ND & Villaescusa-Navarro, F 2023, 'A hierarchy of normalizing flows for modelling the galaxy-halo relationship', Paper presented at Machine Learning for Astrophysics Workshop at the Fortieth International Conference on Machine Learning (ICML 2023), 29/07/23 - 29/07/23. https://doi.org/10.48550/arXiv.2307.06967

TY - CONF

T1 - A hierarchy of normalizing flows for modelling the galaxy-halo relationship

AU - Lovell, Christopher C.

AU - Hassan, Sultan

AU - Anglés-Alcázar, Daniel

AU - Bryan, Greg

AU - Fabbian, G.

AU - Genel, Shy

AU - Hahn, ChangHoon

AU - Iyer, Kartheik

AU - Kwon, James

AU - Santi, Natalí de

AU - Villaescusa-Navarro, Francisco

N1 - 8 pages, 2 figures, accepted for ICML 2023 Workshop on Machine Learning for Astrophysics

PY - 2023/6/13

Y1 - 2023/6/13

N2 - Using a large sample of galaxies taken from the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project, a suite of hydrodynamic simulations varying both cosmological and astrophysical parameters, we train a normalizing flow (NF) to map the probability of various galaxy and halo properties conditioned on astrophysical and cosmological parameters. By leveraging the learnt conditional relationships we can explore a wide range of interesting questions, whilst enabling simple marginalisation over nuisance parameters. We demonstrate how the model can be used as a generative model for arbitrary values of our conditional parameters; we generate halo masses and matched galaxy properties, and produce realisations of the halo mass function as well as a number of galaxy scaling relations and distribution functions. The model represents a unique and flexible approach to modelling the galaxy-halo relationship.

AB - Using a large sample of galaxies taken from the Cosmology and Astrophysics with MachinE Learning Simulations (CAMELS) project, a suite of hydrodynamic simulations varying both cosmological and astrophysical parameters, we train a normalizing flow (NF) to map the probability of various galaxy and halo properties conditioned on astrophysical and cosmological parameters. By leveraging the learnt conditional relationships we can explore a wide range of interesting questions, whilst enabling simple marginalisation over nuisance parameters. We demonstrate how the model can be used as a generative model for arbitrary values of our conditional parameters; we generate halo masses and matched galaxy properties, and produce realisations of the halo mass function as well as a number of galaxy scaling relations and distribution functions. The model represents a unique and flexible approach to modelling the galaxy-halo relationship.

KW - astro-ph.GA

UR - https://icml.cc/

U2 - 10.48550/arXiv.2307.06967

DO - 10.48550/arXiv.2307.06967

M3 - Paper

T2 - Machine Learning for Astrophysics Workshop at the Fortieth International Conference on Machine Learning (ICML 2023)

Y2 - 29 July 2023 through 29 July 2023

ER -

A hierarchy of normalizing flows for modelling the galaxy-halo relationship

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this