I will give a brief report from the recent DOE petascale data workshop.

Both now and increasingly for the foreseeable future, scientists must address the challenges posed by petascale data sets. These sets may be produced by high-resolution simulations on massively parallel computers in complex applications, such as climate modeling and fusion calculations. They may also result from experiments and observational studies, such as those in cosmology and high-energy physics. Extracting scientific knowledge from these massive data sets has become both increasingly difficult and increasingly necessary as computer systems have grown larger and experimental devices more sophisticated. Mathematical techniques from several fields, including but not restricted to statistics, machine learning, image analysis, and pattern recognition, have long been used to analyze scientific data. However, many existing methods fail to provide adequate robustness, scalability, and combinatorial tractability when applied to petascale data sets.

The goal of this workshop is to engage mathematical scientists and applications researchers to define a research agenda for developing the next-generation mathematical techniques needed to meet the challenges posed by petascale data sets.