The SVD Drivers

  The drivers for computing the singular value decomposition are of the form Xsvd where X denotes the precision and data type,

 

s single precision

d double precision.

Of course, the SVD is defined for complex matrices as well and it is a straightforward matter to convert the real arithmetic driver to a corresponding complex arithmetic driver.

These drivers may be easily modified to estimate the 2-norm condition number     by setting which = 'BE'. This will ask for a few of the smallest and a few of the largest singular values simultaneously. The condition number could then be estimated by taking the ratio of the largest and smallest singular values.

Since these drivers are simply special cases of dsdrv1, the parameter settings will not be described further. The only cautionary note is that the parameter which may be set to 'SA' if desired but this is not recommended if it is expected that ${\bf A}$ will be nearly rank deficient.