The observational constraints on the primordial power spectrum have tightened considerably with the release of the first year analysis of the WMAP observations , especially when combined with the results from other CMB experiments and galaxy redshift surveys .
These observations allow us to constrain the physics of cosmological inflation :

1 . The data show that the Hubble distance is almost constant during inflation .
While observable modes cross the Hubble scale , it changes by less than 3 \% during one e-folding : \dot { d } _ { H } < 0.032 at 2 \sigma .
The distance scale of inflation itself remains poorly constrained : 1.2 \times 10 ^ { -28 } \mbox { cm } < d _ { H } < 1 \mbox { cm } .

2 . We present a new classification of single-field inflationary scenarios ( including scenarios beyond slow-roll inflation ) , based on physical criteria , namely the behaviour of the kinetic and total energy densities of the inflaton field .
The current data show no preference for any of the scenarios .

3 . For the first time the slow-roll assumption could be dropped from the data analysis and replaced by the more general assumption that the Hubble scale is ( almost ) constant during the observable part of inflation .
We present simple analytic expressions for the scalar and tensor power spectra for this very general class of inflation models and test their accuracy .