The BICEP2 experiment has announced a signal for primordial gravity waves with tensor-to-scalar ratio r = 0.2 ^ { +0.07 } _ { -0.05 } [ 1 ] .
There are two ways to reconcile this result with the latest Planck experiment [ 2 ] .
One is by assuming that there is a considerable tilt of r , \mathcal { T } _ { r } , with a positive sign , \mathcal { T } _ { r } = d \ln r / d \ln k \gtrsim 0.57 ^ { +0.29 } _ { -0.27 } corresponding to a blue tilt for the tensor modes of order n _ { T } \simeq 0.53 ^ { +0.29 } _ { -0.27 } , assuming the Planck experiment best-fit value for tilt of scalar power spectrum n _ { S } .
The other possibility is to assume that there is a negative running in the scalar spectral index , dn _ { S } / d \ln k \simeq - 0.02 which pushes up the upper bound on r from 0.11 up to 0.26 in the Planck analysis assuming the existence of a tensor spectrum .
Simple slow-roll models fail to provide such large values for \mathcal { T } _ { r } or negative runnings in n _ { S } [ 1 ] .
In this note we show that a non-Bunch-Davies initial state for perturbations can provide a match between large field chaotic models ( like m ^ { 2 } \phi ^ { 2 } ) with the latest Planck result [ 3 ] and BICEP2 results by accommodating either the blue tilt of r or the negative large running of n _ { S } .