Self-calibration of weak lensing cosmic shear biases

TL;DR

This paper tackles the calibration of residual biases in weak-lensing cosmic shear measurements by proposing a self-calibration method that jointly infers multiplicative and additive biases from empirical ellipticity distributions, without relying on external simulations or cosmology. It develops a Bayesian forward-modeling framework that maps the intrinsic ellipticity distribution to the observed one via the bias transformation $\hat{\epsilon}=(1+m)\epsilon+c+\epsilon_{\text{n}}$, and marginalizes over hyper-parameters of the intrinsic distribution to break degeneracies. The authors demonstrate, with simulated data, that $m$ and $c$ can be recovered with high accuracy even in the presence of noise, provided a large enough sample and a flexible intrinsic-distribution model; biases degrade gracefully with realistic measurement noise. The work suggests broad applicability to current and future surveys, offering a principled, cosmology-agnostic approach to calibrate shear biases and potentially reduce dependence on large-scale simulations.

Abstract

In order to reach the required performance of Stage-III and IV weak lensing surveys, cosmic shear measurements have to rely on external simulations to calibrate residual biases. Over the years, several techniques have been developed to mitigate the impact of residual biases prior to calibration, including the inference of shear responses on images to correct multiplicative biases, and the empirical correction of additive biases. We introduce a novel methodology that generalises upon the state-of-the-art approaches by inferring multiplicative and additive biases jointly from parameterised distributions of measured ellipticities, crucially without relying on external simulations and independently from cosmology. Shear biases are marginalised over the unknown hyper-parameters in the modelling, hence mitigating the impact of degeneracies. We apply the technique to a representative problem and show the performance of the estimation, even in the presence of noise. The method has a high potential for applicability to the calibration of weak lensing cosmic shear in current and future lensing surveys.

Figures (2)

• Figure 1: Distribution of true ellipticity and observed ellipticity after the effect of shear biases $m=0.05$ and $c=-0.01$.
• Figure 2: Joint MCMC samples of the inferred bias parameters ($m$ and $c$) and distribution hyper-parameters ($\sigma$, $\beta$, and $\kappa$). The red lines/crosses represent the true values and the blue ones represent the recovered mean values marginalised over nuisance parameters. Despite the degeneracies, the marginalised $m$ and $c$ are constrained within one standard deviation (see the two distributions at the top left).