Non-parametric reconstruction of cosmological observables using Gaussian Processes Regression

TL;DR

This paper employs Gaussian process regression, a non-parametric machine learning technique, to reconstruct key cosmological observables, providing insights consistent with the standard cosmological model and estimating the Hubble constant.

Contribution

It introduces a novel application of Gaussian processes for non-parametric reconstruction of cosmological parameters, enhancing data analysis in cosmology.

Findings

Results align with the ΛCDM model.

Estimated Hubble constant at redshift zero: 68.798 ± 6.340 km/s/Mpc.

Reconstructed deceleration parameter and dark energy equation of state.

Abstract

The current accelerated expansion of the Universe remains ones of the most intriguing topics in modern cosmology, driving the search for innovative statistical techniques. Recent advancements in machine learning have significantly enhanced its application across various scientific fields, including physics, and particularly cosmology, where data analysis plays a crucial role in problem-solving. In this work, a non-parametric regression method with Gaussian processes is presented along with several applications to reconstruct some cosmological observables, such as the deceleration parameter and the dark energy equation of state, in order to contribute with some information that helps to clarify the behavior of the Universe. It was found that the results are consistent with $\Lambda$CDM and the predicted value of the Hubble parameter at redshift zero is $H_{0}=68.798\pm 6.340(1\sigma) \text{ km}\text{ s}^{-1}\text{ Mpc}^{-1}$.