f01qdf
f01qdf
© Numerical Algorithms Group, 2002.
Purpose
T
F01QDF Operations with orthogonal matrices, compute QB or Q B after
factorization by F01QCF or F01QFF
Synopsis
[b,ifail] = f01qdf(a,zeta,b<,trans,wheret,ifail>)
Description
Q is assumed to be given by
T
Q=(Q Q ...Q ) ,
n n-1 1
Q being given in the form
k
(I 0 )
Q =(0 T )
k ( k)
where
T
T =I-u u ,
k k k
((zeta) )
( k)
u =(z ),
k ( k )
(zeta) is a scalar and z is an (m-k) element vector. z must be
k k k
supplied in the kth column of A in elements A(k+1,k),...,A(m,k)
and (zeta) must be supplied either in A(k,k) or in ZETA(k),
k
depending upon the parameter WHERET.
To obtain Q explicitly B may be set to I and pre-multiplied by Q.
T
This is more efficient than obtaining Q .
Parameters
f01qdf
Required Input Arguments:
a (:,:) real
zeta (:) real
b (:,:) real
Optional Input Arguments: <Default>
trans (1) string 'n'
wheret (1) string 's'
ifail integer -1
Output Arguments:
b (:,:) real
ifail integer