Slow-roll inflation and CMB anisotropy data

We emphasize that the estimation of cosmological parameters from CMB anisotropy data, such as the recent high resolution maps from BOOMERanG and MAXIMA-1, requires assumptions on the primordial spectra. The latter are predicted from inflation. The physically best motivated scenario is slow-roll inflation. However, very often the unphysical power-law inflation scenario is (implicitly) assumed in the CMB data analysis. We show that the predicted multipole moments differ significantly in both cases. We identify several misconceptions present in the literature (and in popular numerical codes), among others, generically inflation does neither predict $n_S-1=n_T$ nor that gravitational waves are negligible. We calculate the CMB multipole moments for various values of the slow-roll parameters and demonstrate that an important part of the space of parameters $(n_S,n_T)$ has been overlooked in the CMB data analysis so far.