Re: Matrices and "type tags"

[Mike Miller]

On Thu, 9 Dec 1999, Thomas Hoffmann wrote:

> The state-of-the-art in Octave is, that we have complex and real
> matrix types recognized by the system and handled accordingly, but
> that we have to look for the other "special"  properties of matrices
> ourselves (e.g. one can (wrongly) give a non-symmetric matrix to
> chol()).

Thanks for pointing this out.  I never thought about what chol would do
with a square, asymmetric, positive-definite matrix.  It is the same in
both Octave and MATLAB:

octave:4> X=rand(3)+eye(3), e= eig(X)
X =

  1.49100  0.62694  0.13827
  0.13255  1.16509  0.54857     <---clearly asymmetric
  0.26631  0.76653  1.91000

e =

  1.36335
  2.41438                       <---clearly positive definite
  0.78837

octave:5> chol(X)'*chol(X) - X
ans =                           <---no error messages

  -0.00000   0.00000   0.00000
   0.49439   0.00000   0.00000  <---a reasonable solution
  -0.12805  -0.21796   0.00000


The lower triangle of the input matrix is ignored in computing the
Cholesky factorization.  So I'll be extra careful to have symmetric input!

Regards,

Mike



-----------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.che.wisc.edu/octave/octave.html
How to fund new projects:  http://www.che.wisc.edu/octave/funding.html
Subscription information:  http://www.che.wisc.edu/octave/archive.html
-----------------------------------------------------------------------