Speaker
Description
In the standard $\Lambda$CDM model of cosmology, the equation-of-state for dark energy is $w = p_{DE}/\rho_{DE} = -1$, which implies a constant energy density throughout the evolution of the Universe. However, recent observations of cosmic expansion and gravitational lensing increasingly favour a dynamic, time-varying dark energy. In the literature, the Chevallier–Polarski–Linder (CPL) parametrisation is widely used to describe this behaviour; it is a first-order Taylor expansion centred on the present-day scale factor ($a = 1$). The CPL model is a linear approximation, and its applicability is limited far from the present epoch: at high redshifts, it can cross into the "phantom" regime ($w < -1$), a region that is physically problematic for scalar-field models, while in the distant future, it diverges asymptotically. A few alternative dynamic dark energy models have been investigated in the last couple of decades; however, these often address only one of these issues, focusing either on avoiding future divergence or on preventing a transition into the phantom regime in the past at high redshifts.
The objective of my project is to analyse the evolution of various dynamic dark energy parametrisations in a theoretical framework. I examined quantities describing cosmic expansion and structure formation, such as the matter density parameter, the density of dark energy, and the linear growth factor and rate, and, based on their asymptotic behaviour, formulated physical and phenomenological conditions for comparing different parametrisations. Deriving the result explicitly from the relativistic continuity equation, I showed that a reversal in the monotonically decreasing trend of the matter density parameter, implying a return to matter dominance, occurs precisely when the equation-of-state parameter transitions from the negative to the positive regime, regardless of the specific parametrisation chosen. This allows for the establishment of a phenomenological upper bound, helping to avoid an otherwise difficult-to-justify return to matter dominance and the loss of the negative pressure associated with dark energy. By adhering to this criterion, along with the lower bound on the equation-of-state parameter ($w \in [-1, 0]$) characteristic of scalar-field quintessence models, I employed a symbolic regression (machine learning) approach to identify a parametrisation that satisfies these boundary conditions within a specific region of the parameter space.