Improved analytical modeling of the non-linear power spectrum in modified gravity cosmologies

Abstract

Reliable analytical modeling of the non-linear power spectrum (PS) of matter perturbations is among the chief pre-requisites for cosmological analyses from the largest sky surveys. This is especially true for the models that extend the standard general-relativity paradigm by adding the fifth force, where numerical simulations can be prohibitively expensive. Here we present a method for building accurate PS models for two modified gravity (MG) variants: namely the Hu-Sawicki $f(R)$, and the normal branch of the Dvali-Gabadadze-Porrati (nDGP) braneworld. We start by modifying the standard halo model (HM) with respect to the baseline Lambda-Cold-Dark-Matter ($\hat{I}›$CDM) scenario, by using the HM components with specific MG extensions. We find that our $P(k{)}_{\text{HM}}$ retains 5% accuracy only up to mildly non-linear scales ($k\lesssim 0.3$ $h/\,\mbox{Mpc}$) when compared to PS from numerical simulations. At the same time, our HM prescription much more accurately captures the ratio $\hat{I}\yen (k)=P(k{)}_{\text{MG}}/P(k{)}_{\hat{I}›\text{CDM}}$ up to non-linear scales. We show that using HM-derived $\hat{I}\yen (k)$ together with a viable non-linear $\hat{I}›$CDM $P(k)$ prescription (such as HALOFIT), we render a much better and more accurate PS predictions in MG. The new approach yields considerably improved performance, with modeled $P(k{)}_{\text{MG}}$ being now accurate to within 5% all the way to non-linear scales of $k\lesssim 2.5-3$ $h/\,\mbox{Mpc}$. The magnitude of deviations from GR as fostered by these MG models is typically $\mathcal{O}(10\%)$ in these regimes. Therefore reaching 5% PS modeling is enough for forecasting constraints on modern-era cosmological observables.