SciPost logo

SciPost Submission Page

PINNferring the Hubble Function with Uncertainties

by Lennart Röver, Björn Malte Schäfer, Tilman Plehn

Submission summary

Authors (as registered SciPost users): Tilman Plehn · Lennart Röver
Submission information
Preprint Link:  (pdf)
Date submitted: 2024-04-15 15:46
Submitted by: Röver, Lennart
Submitted to: SciPost Physics
Ontological classification
Academic field: Physics
  • Gravitation, Cosmology and Astroparticle Physics
  • High-Energy Physics - Phenomenology
Approaches: Computational, Phenomenological


The Hubble function characterizes a given Friedmann-Robertson-Walker spacetime and can be related to the densities of the cosmological fluids and their equations of state. We show how physics-informed neural networks (PINNs) emulate this dynamical system and provide fast predictions of the luminosity distance for a given choice of densities and equations of state, as needed for the analysis of supernova data. We use this emulator to perform a model-independent and parameter-free reconstruction of the Hubble function on the basis of supernova data. As part of this study, we develop and validate an uncertainty treatment for PINNs using a heteroscedastic loss and repulsive ensembles.

Current status:
In refereeing

Login to report or comment