Pairwise velocities of the large-scale structure encode information about the growth of structure. They can be observed indirectly through redshift-space distortions and the kinetic Sunyaev Zeldovich effect. In the late-time Universe, pairwise velocities have developed into broad non-Gaussian distributions due to non-linearity, but the information explicitly useful for cosmology is their means – the streaming velocities; the scatters on the means are often treated as nuisance to be marginalized over. This reduces the constraining power of our observations. We show that this does not need to be the case. We apply the density-split (DS) technique to analyze the pairwise velocity distribution (PVD) around different density environments.
The next part of my talk will focus on the modelling aspect of DS. A major challenge with DS is that the modelling is necessary to start from real space, whereas in observations, the density field is in redshift space. Significant efforts have been made to bridge this gap between model and observations. Though these methods work in principle, they are relatively cumbersome to operate numerically. We show that the problem can be solved by splitting the density differently. Instead of volume-weighting, as done for counts-in-cells, we switch to using galaxy-number weighting to define the density for each galaxy. We show that by doing this, the modelling framework for the conventional 2-point correlation function works equally for DS-clustering in redshift space.