In 1997, Deep Blue, a specially modified IBM SP/6000 computer, defeated world chess champion Gary Kasparov. Mathematicians and computer scientists at IBM Research asked the question: "What next?" The continuing exponential growth in computer capabilities meant that
computers would continue to strengthen their game at a rate much faster than possible for any human.

Deep Blue morphed into Deep Computing: the application of very large amounts of computer power to very large amounts of data to enable better decision making.

New Tools for Biology

We are developing very fast pattern matching and pattern discovery methods that can be applied to the molecular sequences found in computational biology, using deep computing techniques. Protein folding is being attacked using a variety of methods. (See Blue Gene on p. 3, and
Computational Biology on p. 14.) We are developing special-purpose auxiliary processors (MD-Grape) to vastly speed up the computations needed in molecular dynamics.

Combinatorial Optimization

Long, standing techniques for optimization using linear (and nonlinear) programming models are being rapidly extended to capabilities previously unimaginable. The circuits of high performance computer chips are being optimized for power/speed. Rather than just computing an optimized
crew-pairing schedule for airlines once every month, work today looks at the problem of coupling precise weather prediction capability with an approach for crew-schedule repairing.

Weather Prediction

Deep computing brings the capability to model weather phenomena on a much finer scale, in both space and time. The result makes possible much more precise, localized forecasts over a 12 to 24 hour period.

Deep Blue for Stock Market?

We created a project called Deep Green. It takes live feeds of tic-by-tic trades on major exchanges, applies deep computing, combined with the insight gained by working with a leading mutual fund company. The result: software able to detect major market shifts faster than can be done by most investors.

Better Coffee

The ideal cup of coffee would come from flavorful beans that had the flavor molecules locked in, until the moment of brewing. We are using quantum chemistry models to help discover the molecular design that might be able to help this. The world of materials science is changed by
Deep Computing's ability to model chemical properties at the atomic level, allowing a comprehensive search for better materials in a wide variety of fields.

Please contact Paridhi Verma to obtain copies of the Computer Science Brochure