学术文献库

Constraints on a dark energy dominated Universe are obtained from an interplay between Bayesian Machine Learning and string Swampland criteria. The approach here differs from previous studies, since in the generative process Swampland criteria are used and, only later, the results of the fit are validated, by using observational data-sets. A generative process based Bayesian Learning approach is applied to two models and the results are validated by means of available $H(z)$ data. For the first model, a parametrization of the Hubble constant is considered and, for the second, a parametrization of the deceleration parameter. This study is motivated by a recent work, where constraints on string Swampland criteria have been obtained from a Gaussian Process and $H(z)$ data. However, the results obtained here are fully independent of the observational data and allow to estimate how the high-redshift behavior of the Universe will affect the low-redshift one. Moreover, both parameterizations in the generative process, for the Hubble and for the deceleration parameters, are independent of the dark energy model. The outcome, both data- and dark energy model-independent, may highlight, in the future, the borders of the Swampland for the low-redshift Universe and help to develop new string-theory motivated dark-energy models. The string Swampland criteria considered might be in tension with recent observations indicating that phantom dark energy cannot be in the Swampland. Finally, a spontaneous sign switch in the dark energy equation of state parameter is observed when the field traverses are in the $z\in[0,5]$ redshift range, a remarkable phenomenon requiring further analysis.