These computer models will help astronomers interpret what they'll be seeing over the next few years. Within a few months, after instruments have been calibrated, a five- to seven-year project called the Sloan Digital Sky Survey will begin exhaustively mapping one-quarter of the night sky. During the project's lifespan, a telescope in New Mexico will capture images of 100 million stars and 100 million distant galaxies. A second instrument called a spectrograph will break down the light from each star and galaxy into individual colors, which give information about distance and velocity.

Comparing numerical results with the images and spectrographic data from Sloan will enable astronomers to zoom in on several cosmological numbers: the density of mass in the universe, the fraction of matter that is unseen or "dark," and a hypothetical force called the cosmological constant.

This is "quack equals duck" reasoning. If researchers can tweak their models so that their simulations produce something that looks like the real universe as mapped by Sloan, then the parameters in the simulations are probably close approximations of those in the real universe. Having reliable models will, in turn, help answer fundamental questions about the universe such as whether it will expand into nothingness or collapse upon itself.

Thousands of subhalos can be seen in this simulation of a cluster of galaxies by researchers at the University of Washington. The color scale represents a factor of 3,000 in density, with yellow as the most dense and blue the least dense. Although the simulation has only 700,000 particles, its effective resolution is that of 3 billion particles. It is one of many being produced by the researchers as they refine their model of how the universe evolved.

Creating a universe, even a virtual one, is neither quick nor easy. Indeed, a simulation of the universe with enough detail to provide meaningful comparison with the Sloan data is far beyond the capabilities of even the fastest and biggest supercomputers. "There are some odd puzzles in how clusters look in the real universe that have not been replicated in any simulation," Lake says.

Fortunately, researchers do not need to simulate the whole universe in exquisite detail, just a small chunk of it.

Quinn and Lake--as well as graduate students Jeff Gardner and Joachim Stadel--take an approach similar to how hurricane forecasters produce their models. Since it's impossible to produce a detailed storm model for the entire globe, hurricane simulations focus on the areas where the hurricanes form and blow. However, hurricane researchers increase accuracy by also including coarser, less detailed models of global climate patterns, such as El Niño, that influence the number and strength of hurricanes.

This movie shows the evolution of structure in a low-density universe.