Speaker
Description
We evaluate the capability of the J-PAS survey to constrain the primordial power spectrum using a non-parametric Bayesian reconstruction approach. Simulated spectra are generated with localized oscillatory features—motivated by non-standard inflationary scenarios—and analyzed over the range $k \in [0.02,0.2] \text{ h } \mathrm{Mpc}^{-1}$, where J-PAS offers optimal sensitivity and non-linearities remain subdominant. The primordial spectrum is reconstructed through linear interpolation across $N$ knots in the $\log{(k, P_{\mathcal{R}})}$ plane, jointly sampled with cosmological parameters ${H_0, \Omega_b h^2, \Omega_c h^2}$ using PolyChord. Feature detection is quantified through both the Bayes factor and a hypothesis test. We explore the recovery of injected features under various J-PAS configurations—including redshift binning, tracer type, survey area, and filter strategy—and find that amplitudes as low as 2% can be detected when combining multiple tracers and redshift bins.
Ongoing work applies this reconstruction framework to real large-scale structure data from BOSS and eBOSS, providing the first data-driven validation of the method and setting competitive constraints on potential primordial features.
