Improving Slow-Roll Estimates in Starobinsky Inflation Using an Analytic Hubble Parameter

Abstract

Abstract Inflationary observables are naturally formulated in terms of the Hubble-flow hierarchy. Because exact background solutions are often unavailable, the Hubble-flow parameters are commonly approximated using potential slow-roll (PSR) quantities constructed from derivatives of the inflationary potential. We revisit the accuracy of this replacement for Starobinsky inflation by mapping analytic Jordan-frame background variables to their Einstein-frame counterparts and deriving the Hubble-flow parameters directly. The leading analytic background is an exact solution of a controlled deformation of the Starobinsky action. Linearizing the pure Starobinsky equation about this background yields a closed-form perturbative correction. Using the same first-order spectral expressions throughout, the perturbatively corrected background tracks the numerical evolution of the first two Hubble-flow parameters more accurately than the next-to-leading-order (NLO) PSR approximation over the common domain in which the NLO expressions remain physical. Moreover, the NLO estimate of \(\widetilde{\epsilon}_1\) reaches a maximum value of approximately \(0.114\), subsequently decreases, and therefore cannot determine the end of inflation from \(\widetilde{\epsilon}_1=1\). For the Planck 2018 central value \(n_s=0.9649\), the perturbatively corrected background gives \(\widetilde{N}_*\simeq55.60\), compared with \(55.72\) for the converged numerical background and \(56.98\) for the conventional large-\(\widetilde{N}_*\) estimate. For the adopted ACT DR6 combined-data value \(n_s=0.974\), it gives \(\widetilde{N}_*\simeq75.43\), compared with the conventional value \(76.92\). The analytic construction therefore provides a self-consistent endpoint and a first-order estimate of the e-fold number that closely reproduces the converged numerical result.