James Varah

Professor of Computer Science

B.Sc., UBC (1963); M.Sc., Stanford (1965); Ph.D., Stanford (1967); NRC Postdoctoral Fellow (1967-68); Assistant Professor, CalTech (1969-71); Killam Senior Fellow (1980-81); Visiting Professor, University of Waterloo (1982-83); Head of Computer Science, UBC (1984-87), Director of CICSR (1986-96).

Numerical analysis and computation, particularly computational linear algebra and graphical data fitting.

Research Interests

My research interests range over a broad spectrum of problems in numerical analysis and scientific computation. One basic theme is conditioning: given the mathematical formulation of a problem and assumptions about the inherent errors in the data, determine what computational accuracy can be expected in the results. Another basic theme is the algorithm: how to produce effective, efficient computational schemes for solving particular problems.

These and other themes are being explored in a variety of subject areas--in particular, I should mention numerical linear algebra problems, including iterative and direct methods for the solution of sparse linear systems, eigenvalue computations, applications of the singular value decomposition, and linear and nonlinear least squares problems. Another specific area is data fitting in two and three dimensions--the design of fast algorithms for generating smooth data fits, with applications in graphics and robotics.

Selected Publications