On the road to per cent accuracy IV: ReACT – computing the non-linear power spectrum beyond ΛCDM

Abstract

To effectively exploit large-scale structure surveys, we depend on accurate and reliable predictions of non-linear cosmological structure formation. Tools for efficient and comprehensive computational modelling are therefore essential to perform cosmological parameter inference analyses. We present the public software package ReACT, demonstrating its capability for the fast and accurate calculation of non-linear power spectra from non-standard physics. We showcase ReACT through a series of analyses on the DGP and f(R) gravity models, adopting LSST-like cosmic shear power spectra. Accurate non-linear modelling with ReACT has the potential to more than double LSST’s constraining power on the f(R) parameter, in contrast to an analysis that is limited to the quasi-linear regime. We find that ReACT is sufficiently robust for the inference of consistent constraints on theories beyond ΛCDM for current and ongoing surveys. With further improvement, particularly in terms of the accuracy of the non-linear ΛCDM power spectrum, ReACT can, in principle, meet the accuracy requirements for future surveys such as Euclid and LSST.

Hybrid Pℓ(k): general, unified, non-linear matter power spectrum in redshift space

“…P SPT is the 1-loop SPT prediction for the matter-matter power spectrum and we have dropped the superscript from P L , where it is assumed that it is always computed within the theory of dark energy or modified gravity under consideration. Finally, we note that we use the publicly available code ReACT to compute equation (2.11) [87]. All halo model quantities are computed assuming a Sheth-Tormen mass function [88,89], a power law virial concentration and a Navarro-Frenk-White (NFW) halo density profile [90].…”

Categorizing models using Self-Organizing Maps: an application to modified gravity theories probed by cosmic shear

“…For the purpose of this first study we assume a linear matter power spectrum. Indeed, while various approaches have been proposed to model non-linearities in modified gravity theories (see for example [Mead et al 2016;Bose et al 2020;Thomas 2020]), an alternative is to remove measurements of cosmic shear on non-linear scales for the cosmological analysis (as done in [Abbott et al 2019] for instance). In a similar way, we use the cosmic shear correlation functions in a range of θ roughly similar to the conservative scale cuts used for the constraints on deviations to GR in [Abbott et al 2019] 11 .…”

Seeking New Physics in Cosmology with Bayesian Neural Networks: Dark Energy and Modified Gravity

“…Rapidly generating accurate power spectra in the linear and nonlinear regime for both ΛCDM and non-ΛCDM models is therefore a crucial capability in order to obtain the necessary quality and quantity of training data. To this end we utilise the recently developed code ReACT [19] which calculates modified power spectra using the halo-model reaction method developed in Ref. [18].…”

“…[18] provided a method to predict the shape of the nonlinear matter power spectrum for a wide range of models which was subsequently implemented into a code called ReACT in Ref. [19]. Using this framework it is possible to generate a large dataset of mock matter power spectra for a broader class of extensions to ΛCDM with varying values of the model parameters.…”