Work

Data-driven reconstruction of the dark-energy equation of state with density-based symbolic regression

Year: 2026
Type: preprint
Abstract: Recent joint analyses of baryon acoustic oscillations from the Dark Energy Spectroscopic Instrument (DESI), cosmic microwave background, and type Ia supernova observations suggest a preference for dynamical dark energy within the Chevallier-Polarski-Linder (CPL) parametrization. In this work, we reconstruct the dark-energy equation of state from observations without prescribing its functional form, using symbolic regression of reconstructed logarithmic dark-energy density histories to obtain four two-parameter analytic models. Their coefficients and cosmological parameters are jointly constrained using the Planck+DESI+DESY5 data. For the model with the lowest χ², the reconstructed w(a) shows a smooth, CPL-like evolution, crossing from w < −1 to w > −1 near a ≃ 0.72 while avoiding future divergence. The other three models show similar low-redshift phantom crossings but allow additional local variations. Despite differences in the equations of state, all four reconstructed models show a maximum in the dark-energy density near the low-redshift phantom crossing and provide fits comparable to CPL, with similar inferred present-day expansion rates and matter densities. These results support the broader applicability of CPL as an effective description of late-time dark-energy evolution across different functional forms.
Language: English
Open Access status: green
OpenAlex ID: W7236086178
Cites:
Cited by:
Related to: 10