The running-mass inflation model, which has strong motivation from particle physics, predicts density perturbations whose spectral index is strongly scale-dependent. For a large part of parameter space the spectrum rises sharply to short scales. In this paper we compute the production of primordial black holes, using both analytic and numerical calculation of the density perturbation spectra. Observational constraints from black hole production are shown to exclude a large region of otherwise permissible parameter space.

We describe a method of obtaining the inflationary potential from observations which does not use the slow-roll approximation. Rather, the microwave anisotropy spectrum is obtained directly from a parametrized potential numerically, with no approximation beyond linear perturbation theory. This permits unbiased estimation of the parameters describing the potential, as well as providing the full error covariance matrix. We illustrate the typical uncertainties obtained using the Fisher information matrix technique, studying the $\lambda \phi^4$ potential in detail as a concrete example.

Accurate estimation of cosmological parameters from microwave background anisotropies requires high-accuracy understanding of the cosmological model. Normally, a power-law spectrum of density perturbations is assumed, in which case the spectral index $n$ can be measured to around $\pm 0.004$ using microwave anisotropy satellites such as MAP and Planck. However, inflationary models generically predict that the spectral index $n$ of the density perturbation spectrum will be scale-dependent. We carry out a detailed investigation of the measurability of this scale dependence by Planck, including the influence of polarization on the parameter estimation. We also estimate the increase in the uncertainty in all other parameters if the scale dependence has to be included. This increase applies even if the scale dependence is too small to be measured unless it is assumed absent, but is shown to be a small effect. We study the implications for inflation models, beginning with a brief examination of the generic slow-roll inflation situation, and then move to a detailed examination of a recently-devised hybrid inflation model for which the scale dependence of $n$ may be observable.

If primordial scalar and tensor perturbation spectra can be inferred from observations of the cosmic background radiation and large-scale structure, then one might hope to reconstruct a unique single-field inflaton potential capable of generating the observed spectra. In this paper we examine conditions under which such a potential can be reliably reconstructed. For it to be possible at all, the spectra must be well fit by a Taylor series expansion. A complete reconstruction requires a statistically-significant tensor mode to be measured in the microwave background. We find that the observational uncertainties dominate the theoretical error from use of the slow-roll approximation, and conclude that the reconstruction procedure will never insidiously lead to an irrelevant potential.

Cosmic microwave anisotropy satellites promise extremely accurate measures of the amplitude of perturbations in the universe. We use a numerical code to test the accuracy of existing approximate expressions for the amplitude of perturbations produced by single-field inflation models. We find that the second-order Stewart-Lyth calculation gives extremely accurate results, typically better than one percent. We use our code to carry out an expansion about the general power-law inflation solution, providing a fitting function giving results of even higher accuracy.

We use a numerical code for accurate computation of the amplitude of linear density perturbations and gravitational waves generated by single-field inflation models to study the accuracy of existing analytic results based on the slow-roll approximation. We use our code to calculate the coefficient of an expansion about the exact analytic result for power-law inflation; this generates a fitting function which can be applied to all inflationary models to obtain extremely accurate results. In the appropriate limit our results confirm the Stewart--Lyth analytic second-order calculation, and we find that their results are very accurate for inflationary models favoured by current observational constraints.