Sparse QR Decomposition

This serial routine computes the QR decomposition of a matrix A in Remxn, m > = n, such that

P_r*QT*A*P_cT = | R | | 0 | where QT*Q = I in Remxm, R in Renxn upper triangular, and P_r in Remxm as well as P_c in Renxn are permutation matrices.

Note that this is a research code developed for the following paper. Thomas H. Robey and Deborah L. Sulsky, Row Ordering for Sparse QR Decomposition, SIAM J. Matrix Anal. and Appl., Vol. 15, No. 4, pp. 1208-1225, October 1994. The software is available in C and C++ under the GNU Library General Public License. To obtain the source code click on the desired version listed below.

Version 2.* has improved code for calculating fill that results in less memory and faster execution. Version 3.* reduces work in the minimum degree calculation resulting in faster execution.

Categories