About Us

My name is Sanjay Velamparambil.

I founded the Fast Algorithms Consultants(*) in January 2009 to offer technical consulting in the design, implementation and optimization of fast and efficient numerical algorithms.

Expertise

• Fast Multipole Method (FMM) and its multilevel variants for Laplace, Helmholtz, and Maxwell equations
• Low rank approximation/compression algorithms for dense matrices
• Fast integral equation methods in electromagnetic theory; esp. method of moments accelerated using FMM or recursive QR/SVD approaches
• Pre-corrected fast Fourier transform (FFT) methods for integral equations with convolution kernels
• Krylov subspace methods for the iterative solution of matrix equations, such as GMRES and QMR.
• Preconditioners for Krylov subspace methods, such as Algebraic Multigrid, Sparse Approximate Inverse and Incomplete LU decomposition
• Parallel processing and parallel algorithms. Extensive experience parallelization using MPI and threads

Experience

• Director of Research (Matrix Algorithms), Acceleware Corporation, January 2009-present
• Senior Research Scientist, Acceleware Corporation, September 2007-December 2008
• Research & Development Engineer, Ansoft Corporation, March 2002 - August 2007
• Research Scientist/Post-doc, University of Illinois at Urbana-Champaign, May 1997 - February 2002
• Post-doctoral Fellow, University of Victoria, British Columbia, April 2000 - July 2000
• Senior Engineer, Computervision R & D (India) Pvt. Ltd, Poona, India, March 1996 -April 1997
• Member Technical Staff, Center for Development of Advanced Computing (CDAC), Bangalore, October 1995 - March 1996

Education

• PhD in Electrical Engineering, Indian Institute of Science, Bangalore, India (1997)
• Master of Engineering in Electrical Engineering, Indian Institute of Science, Bangalore (1991)

Publications

• 13 Refereed journal papers, 2 book chapters and 18 conference publications

Notable Achievements

• University of Illinois(2001): Faster Algorithm Brings New Breakthroughs in Electromagnetic Simulations
• University of Illinois(1999): Fast and Efficient Algorithms: New Technique Reduces Computing Time and Handles Nearly 10 million unknowns

(*) Currently in the process of being incorporated into an LLC.