Constraining modified gravity with weak lensing peaks

TL;DR

This paper demonstrates that weak-lensing peak statistics can effectively distinguish between general relativity and modified gravity models like $f(R)$ and nDGP, providing promising constraints for future stage-IV surveys.

Contribution

It introduces a method combining peak statistics and Gaussian Process emulation to forecast constraints on modified gravity models from weak-lensing data.

Findings

Peak abundance is sensitive to modified gravity strength.

Peak two-point correlation function probes screening mechanisms.

Forecasts show potential to distinguish models at 2-sigma level.

Abstract

It is well established that maximizing the information extracted from upcoming and ongoing stage-IV weak-lensing surveys requires higher-order summary statistics that complement the standard two-point statistics. In this work, we focus on weak-lensing peak statistics to test two popular modified gravity models, $f(R)$ and nDGP, using the FORGE and BRIDGE weak-lensing simulations, respectively. From these simulations we measure the peak statistics as a function of both cosmological and modified gravity parameters simultaneously. Our findings indicate that the peak abundance is sensitive to the strength of modified gravity, while the peak two-point correlation function is sensitive to the nature of the screening mechanism in a modified gravity model. We combine these simulated statistics with a Gaussian Process Regression emulator and a Gaussian likelihood to generate stage-IV forecast posterior distributions for the modified gravity models. We demonstrate that, assuming small scales can be correctly modelled, peak statistics can be used to distinguish GR from $f(R)$ and nDGP models at the two-sigma level with a stage-IV survey area of $300 \, \rm{deg}^2$ and $1000 \, \rm{deg}^2$, respectively. Finally, we show that peak statistics can constrain $\log_{10}\left(|f_{R0}|\right) = -6$ to 2\% precision, and $\log_{10}(H_0 r_c) = 0.5$ to 25\% precision.