Constraining dynamical dark energy from galaxy clustering with simulation-based priors

The effective field theory--based full-shape analysis with simulation-based priors (EFT-SBP) provides a unified framework that combines the strengths of perturbation theory and simulation-based modeling for galaxy clustering analyses. We apply EFT-SBP to the galaxy power spectrum and bispectrum measurements from the BOSS survey to test the recent preference for dynamical dark energy reported by the DESI Collaboration. While the combination of DESI baryon acoustic oscillation, Planck cosmic microwave background, and $\mathrm{Pantheon}+$ supernovae data favors a deviation from a cosmological constant, this preference is not supported when these data are combined with the BOSS EFT full-shape likelihood, which yields ${w}_{0}=\ensuremath{-}0.88\ifmmode\pm\else\textpm\fi{}0.11$ and ${w}_{a}=\ensuremath{-}0.34\ifmmode\pm\else\textpm\fi{}0.40$ at 95% CL for the dark energy equation of state parameters. Incorporating simulation-based priors in this analysis further tightens the constraints on dynamical dark energy by concentrating the posterior around the cosmological constant region and yielding ${w}_{0}=\ensuremath{-}0.91\ifmmode\pm\else\textpm\fi{}0.10$ and ${w}_{a}=\ensuremath{-}0.09\ifmmode\pm\else\textpm\fi{}0.35$ at 95% CL. The resulting figure of merit for the dark energy parameters improves by approximately 20% compared with the standard EFT full-shape analysis using conservative Gaussian priors. This improvement arises from modeling the EFT prior distribution with Gaussian mixture models, which accurately capture the non-Gaussian structure of simulation-based priors while preserving the ability to perform analytic marginalization over most EFT nuisance parameters. These results indicate that EFT-SBP provides an efficient and consistent framework for integrating simulation-based information into cosmological parameter inference and for testing extensions beyond $\mathrm{\ensuremath{\Lambda}}\mathrm{CDM}$.