From thermogravity to thermocosmology with a running non-conservation scale

Abstract

Thermogravity offers a novel approach to cosmic acceleration, deriving trace-free gravitational equations and matter non-conservation from thermodynamic processes in spacetime. We extend this framework towards thermocosmology by introducing a running non-conservation scale $L_2(H)$, while restricting non-conservation to cold dark matter. We investigate two prescriptions, namely a Hubble-proportional rate and an inverse-Hubble running model, deriving their cosmological dynamics and effective geometric dark-energy description. The first admits an exact analytical solution and is related to quadratic running-vacuum cosmologies, while the second approaches standard matter dilution at early times and allows reversals in the direction of energy exchange. Interestingly, the reconstructed geometric dark energy can exhibit quintessence-like and phantom-like behavior, including smooth crossings of the phantom divide, without introducing a phantom field or superacceleration. A unified dynamical-system analysis reveals distinct matter-era critical configurations and a continuous family of de Sitter solutions that are transversely attracting under suitable conditions. Confronting both models with Pantheon+ Type Ia supernovae, cosmic chronometers, and DESI DR2 baryon acoustic oscillations, we find a mild preference for positive present-day matter non-conservation. The Akaike information criterion slightly favors the thermogravity scenarios, whereas the Bayesian information criterion favors the simpler $Λ$CDM model. Thus, running thermocosmology provides a viable framework for nontrivial effective dark-energy dynamics.