Observational constraints on K-inflation models

We extend the ModeCode software of Mortonson, Peiris and Easther to enable numerical computation of perturbations in K-inflation models, where the scalar field no longer has a canonical kinetic term. Focussing on models where the kinetic and potential terms can be separated into a sum, we compute slow-roll predictions for various models and use these to verify the numerical code. A Markov chain Monte Carlo analysis is then used to impose constraints from WMAP7 data on the addition of a term quadratic in the kinetic energy to the Lagrangian of simple chaotic inflation models. For a quadratic potential, the data do not discriminate against addition of such a term, while for a quartic (\lambda \phi^4) potential inclusion of such a term is actually favoured. Overall, constraints on such a term from present data are found to be extremely weak.