We study the cosmological constant ($\Lambda$) in the standard $\Lambda$CDM model by introducing the \textit{graduated dark energy} (gDE) characterised by a minimal dynamical deviation from the null inertial mass density of the $\Lambda$ in the form $\rho_{\rm inert}\propto \rho^{\lambda}<0$ with $\lambda<1$ being a ratio of two odd integers, for which its energy density $\rho$ dynamically takes negative values in the finite past. For large negative values of $\lambda$, it creates a phenomenological model described by a smooth function that approximately describes the $\Lambda$ spontaneously switching sign in the late universe to become positive today. We confront the model with the latest combined observational data sets of PLK+BAO+SN+$H$. It is striking that the data predict bimodal posterior probability distributions for the parameters of the model along with large negative $\lambda$ values; the new maximum significantly excludes the $\Lambda$, and the old maximum contains the $\Lambda$. The improvement in the goodness of fit for the $\Lambda$ reaches highly significant levels, $\Delta\chi_{\rm min}^2=6.4$ for the new maxima, while it remains at insignificant levels, $\Delta\chi_{\rm min}^2\lesssim0.02$, for the old maxima. We show that, in contrast to the old maxima, which do not distinguish from the $\Lambda$, the new maxima agree with the model-independent $H_0$ measurements, high-precision Ly-$\alpha$ data, and model-independent $Omh^2$ diagnostic estimates. Our results provide strong hints of a spontaneous sign switch in the cosmological constant and lead us to conjecture that the universe has transitioned from AdS vacua to dS vacua, at a redshift $z\approx 2.32$ and triggered the late-time acceleration, and suggests looking for such mechanisms in string theory constructions.

Bayesian statistics and MCMC (Markov Chain Monte Carlo) algorithms have found their place in the field of Cosmology. They have become important mathematical and numerical tools, especially in parameter estimation and model comparison. In this paper we review some of the fundamental concepts to understand Bayesian statistics, to then introduce the MCMC algorithms and samplers that allow us to perform the parameter inference procedure. We also provide a general description of the standard cosmological model, known as the $\Lambda$CDM model, along with several alternatives to it; and current datasets coming from astrophysical and cosmological observations. Finally, with the tools acquired we use a MCMC algorithm implemented in python -called SimpleMC- to test the cosmological models and find out the combination of parameters that best describes the universe.