Simulating Large-scale Structure

Abstract

After two decades of direct dynamical simulation of large-scale structure in the universe, it is safe to say the subject is now mature. Still, there are parts of the problem that are less well developed than others. In general, the collisionless dynamics of the dark matter component is better understood than the collisional gas dynamics of the baryonic component. In situations where the gas dynamics is relatively simple, such as the Lyman-$α$ forest and the intracluster medium in X-ray clusters, our ability to reproduce observational data has evolved rapidly, and the interpretive and predictive power of such experiments should now be taken seriously. A comparison of twelve gas dynamic codes to the problem of forming a single X-ray cluster shows that numerical inaccuracies are modest (typically below ten percent), leaving missing physics as the main source for large systematic differences between theory and observation. Galaxy formation, being more complex, is farther behind in its development, but simulations capable of resolving the morphological range of the Hubble sequence in cosmologically interesting volumes may be just around the corner.