Featuring the primordial power spectrum: new constraints on interrupted slow-roll from CMB and LRG data

TL;DR

This paper updates constraints on models with oscillatory features in the primordial power spectrum using recent CMB and galaxy data, finding improved fit and specific features at certain scales.

Contribution

It provides new constraints on interrupted slow-roll inflation models with oscillatory features using combined CMB, galaxy, and supernova data.

Findings

Features improve data fit by Delta(chi^2) ~ 7

Standard parameter uncertainties remain stable with features

Best fit features at scale k ~ 0.005 Mpc^-1

Abstract

Using the most recent data from the WMAP, ACT and SPT experiments, we update the constraints on models with oscillatory features in the primordial power spectrum of scalar perturbations. This kind of features can appear in models of inflation where slow-roll is interrupted, like multifield models. We also derive constraints for the case in which, in addition to cosmic microwave observations, we also consider the data on the spectrum of luminous red galaxies from the 7th SDSS catalog, and the SNIa Union Compilation 2 data. We have found that: (i) considering a model with features in the primordial power spectrum increases the agreement with data with the respect of the featureless "vanilla" LCDM model by Delta(chi^2) ~ 7; (ii) the uncertainty on the determination of the standard parameters is not degraded when features are included; (iii) the best fit for the features model locates the step in the primordial spectrum at a scale k ~ 0.005 Mpc^-1, corresponding to the scale where the outliers in the WMAP7 data at ell=22 and ell=40 are located.; (iv) a distinct, albeit less statistically significant peak is present in the likelihood at smaller scales, with a Delta(chi^2) ~ 3.5, whose presence might be related to the WMAP7 preference for a negative value of the running of the scalar spectral index parameter; (v) the inclusion of the LRG-7 data do not change significantly the best fit model, but allows to better constrain the amplitude of the oscillations.