Endre søk
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Learning the Universe: Learning to optimize cosmic initial conditions with non-differentiable structure formation models
Stockholms universitet, Naturvetenskapliga fakulteten, Fysikum.
Stockholms universitet, Naturvetenskapliga fakulteten, Fysikum. Stockholms universitet, Naturvetenskapliga fakulteten, Oskar Klein-centrum för kosmopartikelfysik (OKC). Kyoto University, Japan.ORCID-id: 0000-0002-5934-9018
Stockholms universitet, Naturvetenskapliga fakulteten, Fysikum. Stockholms universitet, Naturvetenskapliga fakulteten, Oskar Klein-centrum för kosmopartikelfysik (OKC).ORCID-id: 0000-0002-4677-5843
Rekke forfattare: 32025 (engelsk)Inngår i: Monthly notices of the Royal Astronomical Society, ISSN 0035-8711, E-ISSN 1365-2966, Vol. 542, nr 2, s. 1403-1422Artikkel i tidsskrift (Fagfellevurdert) Published
Abstract [en]

Making the most of next-generation galaxy clustering surveys requires overcoming challenges in complex, non-linear modelling to access the significant amount of information at smaller cosmological scales. Field-level inference has provided a unique opportunity beyond summary statistics to use all of the information of the galaxy distribution. However, addressing current challenges often necessitates numerical modelling that incorporates non-differentiable components, hindering the use of efficient gradient-based inference methods. In this paper, we introduce Learning the Universe by Learning to Optimize (LULO), a gradient-free framework for reconstructing the 3D cosmic initial conditions. Our approach advances deep learning to train an optimization algorithm capable of fitting state-of-the-art non-differentiable simulators to data at the field level. Importantly, the neural optimizer solely acts as a search engine in an iterative scheme, always maintaining full physics simulations in the loop, ensuring scalability and reliability. We demonstrate the method by accurately reconstructing initial conditions from halos identified in a dark matter-only N-body simulation with a spherical overdensity algorithm. The derived dark matter and halo overdensity fields exhibit cross-correlation with the ground truth into the non-linear regime Mpc. Additional cosmological tests reveal accurate recovery of the power spectra, bispectra, halo mass function, and velocities. With this work, we demonstrate a promising path forward to non-linear field-level inference surpassing the requirement of a differentiable physics model.

sted, utgiver, år, opplag, sider
2025. Vol. 542, nr 2, s. 1403-1422
Emneord [en]
software: machine learning, early Universe, large-scale structure of Universe
HSV kategori
Identifikatorer
URN: urn:nbn:se:su:diva-246672DOI: 10.1093/mnras/staf1289ISI: 001559618700001Scopus ID: 2-s2.0-105014738017OAI: oai:DiVA.org:su-246672DiVA, id: diva2:1996426
Tilgjengelig fra: 2025-09-09 Laget: 2025-09-09 Sist oppdatert: 2025-09-09bibliografisk kontrollert

Open Access i DiVA

Fulltekst mangler i DiVA

Andre lenker

Forlagets fulltekstScopus

Person

Doeser, LudvigAta, MetinJasche, Jens

Søk i DiVA

Av forfatter/redaktør
Doeser, LudvigAta, MetinJasche, Jens
Av organisasjonen
I samme tidsskrift
Monthly notices of the Royal Astronomical Society

Søk utenfor DiVA

GoogleGoogle Scholar

doi
urn-nbn

Altmetric

doi
urn-nbn
Totalt: 9 treff
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf