Testing the Accuracy of Redshift Space Group Finding Algorithms

Using simulated redshift surveys generated from a high resolution N-body cosmological structure simulation, we study algorithms used to identify groups of galaxies in redshift space. Two algorithms are investigated; both are friends-of-friends schemes with variable linking lengths in the radial and transverse dimensions. The chief difference between the algorithms is in the redshift linking length. The algorithm proposed by Huchra \& Geller (1982) uses a generous linking length designed to find ``fingers of god'' while that of Nolthenius \& White (1987) uses a smaller linking length to minimize contamination by projection. We find that neither of the algorithms studied is intrinsically superior to the other; rather, the ideal algorithm as well as the ideal algorithm parameters depend on the purpose for which groups are to be studied. The Huchra/Geller algorithm misses few real groups, at the cost of including some spurious groups and members, while the Nolthenius/White algorithm misses high velocity dispersion groups and members but is less likely to include interlopers in its group assignments. In a companion paper we investigate the accuracy of virial mass estimates and clustering properties of groups identified using these algorithms.