Improved Ways to Compare Simulations to Data

Abstract

Theoretical models for structure formation with Gaussian initial fluctuations have been worked out in considerable detail and compared with observations on various scales. It is on nonlinear scales $\lsim 10 \ h^{-1}\ {\rm Mpc}$ that the greatest differences exist between $Ω=1$ models that have been normalized to agree on the largest scales with the COBE data; here especially there is a need for better statistical tests which are simultaneously {\it robust}, {\it discriminatory}, and {\it interpretable}. The era at which galaxy and cluster formation occurs is also a critical test of some models. Needs for the future include faster and cleverer codes, better control of cosmic variance in simulations, better understanding of processes leading to galaxy formation, better ways of comparing observational data with models, and better access to observational and simulation data.