Abstract:

I will present a new method for searching for resonance in the primordial power spectrum. These effects are predicted by a wide variety of models and come in two  flavors, one where the oscillations are log spaced and one where the oscillations are linearly spaced. The proposed method treats the oscillations as perturbations on top of the scale invariant power spectrum. This  decreases some of the computational demand in the search for resonance because it allows us to precompute the transfer functions and derivatives thereof. We show that we expect most of the degeneracy between late time and primordial parameters to reside in the distance to last scattering and the overall amplitude al low frequencies. The former will introduce a additional uncertainty in phase for log spaced oscillations, while for linear spaced oscillations it changes the effective frequency. Errors in the familiar LCDM parameters can be evade through expansion to higher orders. We test our code on simulated Planck-like data, and show we able to recover fiducial input oscillations with an amplitude a few times $\mathcal{O}(10^{-2})$ at the $95\%$ C.L. We vary all LCDM parameters + oscillatory parameters. We apply the code to WMAP 9 year data and confirm the existence of two possible resonant frequencies for log spaced oscillations and a large number of similar improvements for linear spaced oscillations. We then apply our code to the PLANCK data, and recover PLANCK results for log spaced oscillations at the low frequency end. We extend the search to include high frequencies and show that allowing to vary more parameters allows for larger improvements than found by the Planck team. We will discuss our findings in light of performed simulations at similar frequencies.