Natural Inflation: Consistency with Cosmic Microwave Background Observations of Planck and BICEP2

Katherine Freese, William H. Kinney

Published 2014-03-20, updated 2014-05-29Version 3

Natural inflation is a good fit to all cosmic microwave background (CMB) data and may be the correct description of an early inflationary expansion of the Universe. The large angular scale CMB polarization experiment BICEP2 has announced a major discovery, which can be explained as the gravitational wave signature of inflation, at a level that matches predictions by natural inflation models. The natural inflation (NI) potential is theoretically exceptionally well motivated in that it is naturally flat due to shift symmetries, and in the simplest version takes the form $V(\phi) = \Lambda^4 [1 \pm \cos(N\phi/f)]$. A tensor-to-scalar ratio $r>0.1 $ as seen by BICEP2 requires the height of any inflationary potential to be comparable to the scale of grand unification and the width to be comparable to the Planck scale. The Cosine Natural Inflation model agrees with all cosmic microwave background measurements as long as $f \geq m_{\rm Pl}$ (where $m_{\rm Pl} = 1.22 \times 10^{19}\ {\rm GeV}$) and $\Lambda \sim m_{\rm GUT} \sim 10^{16}\ {\rm GeV}$. This paper also discusses other variants of the natural inflation paradigm: we show that axion monodromy with potential $V\propto \phi^{2/3}$ is inconsistent with the BICEP2 limits at the 95\% confidence level, and low-scale inflation is strongly ruled out. Linear potentials $V \propto \phi$ are inconsistent with the BICEP2 limit at the 95\% confidence level, but are marginally consistent with a joint Planck/BICEP2 limit at 95\%. We discuss the pseudo-Nambu Goldstone model proposed by Kinney and Mahanthappa as a concrete realization of low-scale inflation. While the low-scale limit of the model is inconsistent with the data, the large-field limit of the model is marginally consistent with BICEP2. All of the models considered predict negligible running of the scalar spectral index, and would be ruled out by a detection of running.