Convergence of halo finders on subhalo properties in Aquarius simulation

Accuracy of halo-finding is increasingly important as many astrophysical applications are dependent on halofinders to identify structures and substructures. In this project I investigate the accuracy of halo finders in recovering subhalo properties such as mass, size, and other sources (methods) such as unbinding. The data for this project is sourced from a common post-processing analysis (common-analysis) project which sourced its data from a Milky Way sized halo in Aquarius simulation. I begin the project by showing the average mass scatter of 20% to exist in individual finder's own analysis (own-analysis) against the 10% scatter in the common-analysis. I then looked into sample subhaloes from the own-analysis and matched them with the common-analysis, in order demonstrate the mass scatter. I found missing substructures, center offsets and simulation artifacts which illustrated the possible cases for mass scatter. With VOBOZ finder, I showed that the mass scatter was due to different virial mass definitions. Similarly, with ADAPTAHOP and ROCKSTAR finders, I showed that the choice of including or excluding substructures lead to 10% scatter in the mass. Secondly, I compare the sizes of subhaloes across the finders. I demonstrated that multiple size definitions in AHF finder leads to two sets of sizes and therefore leads to scatter in the size. This would mean that, for all the finders with the exception of ROCKSTAR, sizes of subhaloes were not reliable as they were not derived from the mass. With ROCKSTAR, I demonstrated that the density remained same for all the subhaloes suggesting that size can be derived from mass. Users have to be cautious while reading sizes from halo catalogues, and sizes should not be trusted to be derived from the mass. I would recommend the users to generate the sizes themselves from the mass. Finally, I explore the spurious (unbound) fractions across the finders. With HBT finder, I illustrated that the spurious (unbound) subhaloes were distributed on the outskirts, that peaked beyond 200kpc from the halo center. With the cumulative mass functions for the configuration space finders H3D, AHF and SUBFIND, I showed that the former finders performed better unbinding than the latter finders. While H3D showed 50% of its haloes as spurious, AHF showed 80% and SUBFIND recorded 90% of its subhaloes as spurious. All these above mentioned configuration-space finders showed unbound fraction above 50% suggesting that an unbinding procedure is essential for configuration-space finders. With ROCKSTAR, a phase-space finder, I showed that only 5% of its subhaloes were spurious at a resolution of 20 particles indicating that the phase-space finders may be preferred to configuration space finders. As configuration-space finders lack velocity information, they have high fraction of spurious subhaloes, and hence running unbinding procedures is strongly recommended unless the application (such as Tidal relics, Streams, X-ray properties) wants to include the spurious particles. Finders such as ADAPTOHOP, H3D and H6D do not have a built in unbinding routine and therefore applications using them have to be cautious. In summary, I demonstrated that the choice of unbinding and the choice of halo finder does affect the final mass and size. The results from this study would provide the users a degree of caution and guide them in choosing the right halo finder that suits their application