Researchers at the University of Maryland have developed the methodology and associated software for speeding up and reducing the amount of memory required for several tasks that occur repeatedly in many computations. These computations (solving the Laplace, Biharmonic, and Helmholtz equations) are important for areas such as solid modeling and graphics, data mining, acoustics, radar and antenna design, electrical circuit design and molecular modeling among others.
The general method used in these software packages, termed the Fast Multipole Method (FMM), is a method for speeding up the product of dense matrices with vectors. Where direct products of a matrix with a vector scales as the size of the matrix, the time taken for the new FMM software is proportional to the size of the vector and provides for significant gain in speed and decrease in memory. A University of Maryland start-up company will be based on this novel technology.
