Re: Octave project involvement

[David Bateman]

John W. Eaton wrote:
> On  7-Sep-2007, David Bateman wrote:
> 
> | so the bsxfun function is slightly slower than the repmat, but doesn't
> | use the additional memory. The reason bsxfun is slower than it needs to
> | be is that it can make no assumption that the matrix type returned from
> | the feval will be the same at each call,
> 
> Even for the built-in functions listed in Matlab docs for bsxfun?  Are
> you thinking of cases where a complex value can be generated from real
> arguments, for example?  At least in those cases, I think it doesn't
> matter for Matlab because a complex variable is just a double value
> with a complex part.  But it does matter for Octave since real and
> complex are separate types.
> 
> jwe

Well I didn't special case the functions listed in the bsxfun, so I
can't be sure. I don't think the cost of the slow-up is significant for
the cases where bscfun make sense (ie where memory is extremely tight).

D.