Implicit-shifted Symmetric QR Singular Value Decomposition of 3x3 Matrices

Theodore Gast , Chuyuan Fu, Chenfanfu Jiang, Joseph Teran
UCLA Mathematics Department Technical Report (CAM16-19)

Computing the Singular Value Decomposition (SVD) of 3x3 matrices is commonplace in 3D computational mechanics and computer graphics applications. We present a C++ implementation of implicit symmetric QR SVD with Wilkinson shift. The method is fast and robust in both float and double precisions. We also perform a benchmark test to study the performance compared to other popular algorithms.


Download Paper



Download Code v1.0 (Released 4/27/2016)



To cite:

Theodore Gast, Chuyuan Fu, Chenfanfu Jiang, and Joseph Teran, Implicit-shifted Symmetric QR Singular Value Decomposition of 3x3 Matrices. University of California, Los Angeles Mathematics Department Technical Report CAM16-19 (2016).

@techreport{qrsvd:2016,
title={Implicit-shifted Symmetric QR Singular Value Decomposition of 3x3 Matrices},
author={Gast, Theodore and Fu, Chuyuan and Jiang, Chenfanfu and Teran, Joseph},
year={2016},
institution={University of California Los Angeles}
}