Re: ellipse

[Søren Hauberg]

fre, 30 01 2009 kl. 11:38 -0500, skrev John W. Eaton:
> BTW, is there a good reason that you switched the order of the N and
> SHIFT arguments from the original?  That change will break all the
> code that the original author (my former boss) wrote that uses this
> function.  I don't mind making the switch if there is a good reason,
> but it's not immediately obvious to me what that might be.

Honestly, I can't really remember what I thought at the moment...

That being said, I actually think it makes sense to put 'shift' before
'n'. I use this function to plot covariances of data, where the code
roughly look like this

  mu = mean (data);
  C = cov (data);
  ellipse (inv (C), 2, mu);

It is quite rare that I want to change the number of points on the
ellipse as 100 is most often a perfectly fine value. It is however very
common the my data has a non-zero mean value. So, I see myself using a
non-default value of 'shift' much more often than a non-default value of
'n'. Therefore I think 'shift' should appear before 'n'.

That being said, it's not an important issue for me, so I don't mind
swapping things back. Alternatively, we could add something like

  if (isvector (n) && isscalar (shift))
    tmp = n;
    n = shift;
    shift = tmp;
  endif

to allow both styles.

Soren