HDLASQ

An improved DQDS algorithm. In this paper we present an improved differential quotient difference with stifts (DQDS) algorithm for computing all the singular values of a bidiagonal matrix to high relative accuracy. There are two key contributions: a novel deflation strategy that improves the convergence for badly scaled matrices, and some modifications to certain shift strategies that accelerate the convergence for most bidiagonal matrices. These techniques together ensure linear worst case complexity of the improved algorithm (denoted by V5). Our extensive numerical experiments indicate that V5 is typically 1·2×-4× faster than DLASQ (the LAPACK-3.4.0 implementation of DQDS) without any degradation in accuracy. On matrices for which DLASQ shows very slow convergence, V5 can be 3×-10× faster. We develop a hybrid algorithm (HDLASQ) by combining our improvements with the aggressive early deflation strategy (AggDef2 of Y. Nakatsukasa et al. [SIAM J. Matrix Anal. Appl. 33, No. 1, 22–51 (2012; Zbl 1248.65042)]). Numerical results show that HDLASQ is the fastest among these different versions.


References in zbMATH (referenced in 2 articles )

Showing results 1 to 2 of 2.
Sorted by year (citations)

  1. Li, Shengguo; Gu, Ming; Parlett, Beresford N.: An improved DQDS algorithm (2014)
  2. Andrisani, E.; Di Lena, G.: An improved dqds type algorithm (2000)