Re: Arguments out?

[Paul Kienzle]

On Dec 11, 2003, at 6:05 PM, Vic Norton wrote:

> Yeah, I thought about that later, Thorsten. Thanks for pointing it out.
>
> So when you write something like
>
>    function   [U, S, V] = svd(A)
>
> and A comes in as an m x n, you might do the computations to get
>
>    U - orthogonal m x m
>    S - diagonal m x n
>    V - orthogonal n x n
>
> satisfying
>
>    A = U * S * V'
>
> Then you look at nargout.
>
> You will only accept nargout = 0, 1, or 3. If nargout = 0 or 1, you 
> want to return diag(S). If nargout = 3, you want to return the triple 
> [U, S, V]. Apparently your final lines should read something like this
>
>    if (nargout < 2)
>       U = diag(S);
>    elseif (nargout != 3)
>       error("you don't know what you are doing, dummy!");
>    endif
>
> This is what bothers me. U and diag(S) are completely different kinds 
> of objects. Writing U = diag(S) simply goes against my grain.

So write it using varargout:

function varargout = svd(A)
    ...
    switch nargout
      {0,1}, varargout = { diag(s) };
      3, varargout = { u, s, v };
      otherwise, usage("sigma=svd(A) or [u,s,v]=svd(A)");
   end
end

or equivalently using meaningless placeholders
for the returned values:

function [r1,r2,r3] = mysvd(A)
  ...
  if nargout<=1, r1=diag(s);
  else r1=u; r2=s; r3=v; end
end

Paul Kienzle
address@hidden



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

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