Solving the $H_{0}$ tension in $f(T)$ Gravity through Bayesian Machine Learning

TL;DR

This paper employs Bayesian Machine Learning and strong lensing time delay techniques to address the H0 tension in f(T) gravity, demonstrating the exponential model's effectiveness and robustness across various parameters and data sets.

Contribution

It introduces a novel application of Bayesian Machine Learning to f(T) gravity models, showing the exponential model can resolve the H0 tension effectively.

Findings

The exponential f(T) model can resolve the H0 tension efficiently.

The method's robustness varies with redshift ranges and lens/source parameters.

Constraints are validated using observational Hubble data.

Abstract

Bayesian Machine Learning~(BML) and strong lensing time delay~(SLTD) techniques are used in order to tackle the $H_{0}$ tension in $f(T)$ gravity. The power of BML relies on employing a model-based generative process which already plays an important role in different domains of cosmology and astrophysics, being the present work a further proof of this. Three viable $f(T)$ models are considered: a power law, an exponential, and a squared exponential model. The learned constraints and respective results indicate that the exponential model, $f(T)=\alpha T_{0}\left(1-e^{-p T / T_{0}}\right)$, has the capability to solve the $H_{0}$ tension quite efficiently. The forecasting power and robustness of the method are shown by considering different redshift ranges and parameters for the lenses and sources involved. The lesson learned is that these values can strongly affect our understanding of the $H_{0}$ tension, as it does happen in the case of the model considered. The resulting constraints of the learning method are eventually validated by using the observational Hubble data(OHD).