Constraining dark matter decays with cosmic microwave background and weak lensing shear observations

Constraining dark matter decays with cosmic microwave background and weak lensing shear observations by Jozef Bucko et al. on Monday 28 November From observations of both low and high redshifts, it is well known that the bulk of dark matter (DM) has to be stable or, at least, very long-lived. However, the possibility that a small fraction of DM is unstable or that all of DM decays with a half-life time ($\tau$) significantly larger than the age of the universe is not ruled out. One-body decaying dark matter (DDM) consists of a minimal extension to the $\Lambda$CDM model. It causes a modification of the cosmic growth history as well as a suppression of the small-scale clustering signal, providing interesting consequences regarding the $S_8$-tension, the observed differences of the clustering amplitude between weak lensing (WL) and cosmic microwave background (CMB) observations. In this paper we investigate models where a fraction or all DM decays into radiation, focusing on the long-lived regime i.e. $\tau \gtrsim H_0^{-1}$ ( $H_0^{-1}$ being the Hubble time). We use WL data from the Kilo-Degree Survey (KiDS) and CMB data from Planck. First, we confirm that this DDM model cannot alleviate the $S_8$-tension. We then show that the most constraining power for DM decays does not come from the nonlinear weak lensing data but from CMB via the integrated Sachs-Wolfe effect. From the CMB data alone, we obtain constraints of $\tau \geq 288$ Gyr if all the DM is assumed to be unstable, and we show that a maximum fraction of $f=0.07$ is allowed to decay assuming the half-life time to be comparable to (or smaller than) one Hubble time. The constraints from the KiDS-1000 WL data are significantly weaker, being at $\tau \geq 60$ Gyr and $f<0.34$, respectively. Combining the CMB and WL data does not yield tighter constraints than the CMB alone, except for small half-life times where the maximum allowed fraction becomes $f=0.03$. All limits are provided at 95\% confidence level. arXiv: http://arxiv.org/abs/http://arxiv.org/abs/2211.14334v1