Ph.D., Stanford University
Scientific computation, parallel processing
Luk came to RPI in 1992 after fourteen years at Cornell University. His research centers on the development of fast and stable matrix algorithms for scientific and engineering computation. He studies the implementation of these techniques for both general-purpose parallel architectures and special-purpose VLSI devices. He is also interested in fault tolerance, particularly with regards to systolic computing.
Since 1987 Luk has chaired a SPIE Conference on Advanced Signal Processing Algorithms, Architectures, and Implementations, held every July in San Diego. A chief feature of this international meeting is its effort to bring together researchers from universities, national laboratories, and industries.
Currently, Luk has two sponsored research projects. One concerns fast image reconstruction based on the regularized pseudoinverse of the imaging operator. The other deals with computational issues in first-principles for solids. Both projects emphasize the development of new numerical algorithms as well as their efficient implementation on massively parallel processors.