DDMA's approach to data analysis integrates mathematical research with expertise in computer science. DDMA's work on the computing environment as it effects data analysis includes:

• Parallel Language Research. One of the great challenges we will face in effectively using the current and next generation computers is transitioning from traditional languages like Fortran to truly parallel languages that are designed to work on machines with muliticore processors and multiple hierarchies of memory and storage. Existing techniques and tools to utilize these machines effectively are not often associated with programmer productivity, and new language research attempts to remedy this. We propose a compiler infrastructure and tool project that can be used to interpret new parallel language features, and provide a sandbox for research and testing of new features to drive language standards efforts.
• Problem Solving Environments (PSE's). Problem solving environments (PSE's) for data analysis are intended to make underlying computer and software complexity invisible to the domain experts that use the tools. At their best, PSE's solve simple or complex problems, support rapid prototyping or detailed analysis, and can be used in introductory education or at the frontiers of science. They hide issues of parallelism, data management, coupling software components that are exposed to the user with traditional tools. PSE's form the bridge between computer scientists and domain experts. They are essential to producing repeatable, understandable results on complex datasets being analyzed on complex computer systems.
• Performance Analysis. It's easy to write bad parallel programs, and hard to write good ones. Performance analysis and tuning maximize utilization of computing resources, but will become harder to do on heterogeneous architectures and multiple hierarchies of memory and storage. Furthermore, analysis of performance data is still a largely qualitative endeavor, with very rudimentary analysis techniques applied to understanding the causes and solutions to performance problems. Leveraging the sophisticated data analysis teams developed by DDMA has proven to yield new and interesting insights into performance analysis problems.
• Numerical Methods Tensor multiplication is a ubiquitous task in many scientific and engineering applications. While there has been a great deal of research on efficient numerical methods for matrix multiplication on parallel machines, the general multiplication of arbitrary-rank tensors with an arbitrary number of contractions has seen less progress. We consider a new set of C++ classes that compute, store and multiply tensors on both serial and parallel platforms.