On Fri, Nov 25, 2011 at 08:38:18PM -0500, Nir Krakauer wrote:
> This is an implementation of Powell's conjugate gradient descent
> method that might go in the minimization section of optim.
>
> Best,
>
> Nir
I cannot say much to the algorithm, probably there should be no problem
adding this function (have you checked if there is already something
similar in 'optim'?).
But a few remarks to some coding aspects.
- It is now obsolete (and should
not be done in new code) to have an argument ('args' in your case) with
extra arguments for the user function. These extra arguments can be passed
using anonymous functions, e.g.
optimizer (@ (x) user_function (x, extra_argument, ...), ...)
and need not be cared for by the optimizer. Only the argument which is
optimized is to be cared for
- For the optimization functions of core Octave, 'optimset' is now used
for handling optimization options. In the optim package, some functions
(which were programmed before Octaves 'optimset' was available)
have different mechanisms of option handling. But this makes things more
complicated. In new code as yours, we should stick to some standard of
option handling. And as things stand now, the standard is Octaves 'optimset'.
So I think you should re-consider the handling of the 'control' argument.
Note that if there is a _good_ reason (which possibly should be discussed
on the list) new optimset options can be defined.
Olaf
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure
contains a definitive record of customers, application performance,
security threats, fraudulent activity, and more. Splunk takes this
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________
Octave-dev mailing list
Octave-dev@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/octave-dev
