Speaker
Description
$f(R)$ gravity remains a well-motivated extension of $\Lambda$CDM, but the constraints on it depend on how faithfully the modified field equations are solved and on the data used. In earlier work (Hough, Abebe & Ferreira 2020, Eur. Phys. J. C 80, 787), we tested several $f(R)$ models, including the Starobinsky and Hu–Sawicki models, against 359 JLA Type Ia supernovae, using a cosmographic closure for the deceleration and jerk parameters. We now revisit these two models with Kosmulator 2.0, the new release of our open-source Python inference framework. Both models are written in terms of a single deviation parameter $b$ at fixed index $n$, with $\Lambda$ fixed by $E(0)=1$, so that $\Lambda$CDM is recovered as $b \to 0$. The modified Friedmann equation is integrated directly, so the deceleration and jerk parameters follow from the solution, and the linear perturbations are solved in a modified version of [hi_class]. We confront the models with the Pantheon+, DES-SN5YR and Union3 supernova compilations in turn, cosmic chronometers, DESI DR2 baryon acoustic oscillations, Big Bang nucleosynthesis deuterium data, and CMB temperature, polarisation and lensing data from Planck 2018, ACT DR6 and SPT-3G. Kosmulator samples the posteriors with emcee and zeus under an autocorrelation-based convergence criterion, and compares each model with $\Lambda$CDM through $\Delta\chi^2$ and its Gaussian-equivalent significance, and through the AIC, AICc, BIC, DIC and WAIC. We this talk we will present preliminary constraints on $b$, $H_0$ and $\Omega_m$, with and without the CMB.