This project will benefit both the research astronomy community and the general public in the US by providing access to the Millennium Database, the most important and widely used simulation worldwide, which has so far been a fully European effort, published and maintained in Germany, and largely inaccessible elsewhere. Once established, other simulation datasets will later be made available, providing a resource of tremendous value, capable of making a meaningful difference to US scientists. The virtual numerical laboratory that will be created represents technical solutions widely applicable to many areas of scientific and educational endeavor.
Cosmological computer simulations with large dynamic range that model statistically representative parts of the observable universe are an essential component of the quest to understand the structure of the universe, and in particular how galaxies form, how they are distributed in space, and how environment influences their evolution. The data sets produced by such simulations have started to exceed in size and richness the results of the largest observational surveys, straining the computational and storage resources that support the required complex algorithms. This is especially true because much of the analysis will be performed by scientists who were not involved in the original work and may be unfamiliar with the properties of the simulation outputs. This project will allow access to the original simulations, and simultaneously extend and complete the creation of synthetic observations that can be directly compared to telescope data. The system will allow experts and non-experts to investigate the virtual universes created by simulations and match them to observation, by forming a virtual observatory with access to everything from raw outputs, through idealized intermediate products, to realistic model observations. Users can investigate such aspects of gravitational collapse as scaling relations and distinct clustering patterns, and study systematic biases and information loss. The objective is to enable everyone to perform the sort of scientific experiments on simulation data that can currently only be done by specialists.
The broadly usable technical solutions that will be created by this study will tie raw Petabyte size simulation output files to structured relational databases, with direct applications as diverse as turbulence and genomics. Forward modelling to the observational domain will benefit any area where theory and simulation must be interpreted by non-specialists. The project will establish the Johns Hopkins group as a core US partner in the European Virgo Consortium, and further establish a public simulation database effort in the US.
