Keywords: scientific visualization, in situ, big data, large scale simulations
Modeling and simulation have become increasingly important in scientific investigations. As more complex phenomena are studied, simulation tools have provided an avenue for exploring scientific theories and their ability to explain observed data. In some cases, these simulations require extensive computation and can generate massive amounts of data. This presentation will explore statistical approaches that have been developed in conjunction with visualization of large scale simulation studies for efficient selection, analysis, and representation of data. Statistically informed visualization makes use of statistical concepts including sampling, monitoring, estimation, and change detection. In some cases, these methods can be implemented “in situ”, while a code is running. The combination of statistical methods with powerful computation and visualization tools can accelerate the process of scientific discovery.