Re: duplicate fminsearch

[Andy Adler]

I've put nmsmax as a subfunctioni inside fminsearch in the version in octave.

This would mean it is not available as a high level function. Mostly, fminsearch
just organizes the parameters and passes them to nmsmax.

Is that what you want?
--
Andy Adler <address@hidden> +1-613-520-2600x8785


On 30 August 2013 03:47, Olaf Till <address@hidden> wrote:
> On Thu, Aug 29, 2013 at 02:16:27PM -0400, Andy Adler wrote:
>> Hi Daniel,
>>
>> As far as I can tell, the fminsearch in octave is the one I sent to
>> Rik in July last year.
>> It was an effort at packaging and documenting (+ new options) to the version 
>> in
>> octave-forge.
>>
>> The old one can go, IMHO.
>> --
>> Andy Adler <address@hidden> +1-613-520-2600x8785
>>
>>
>> On 29 August 2013 13:31, Daniel J Sebald <address@hidden> wrote:
>> > On 08/29/2013 10:25 AM, John W. Eaton wrote:
>> >>
>> >> There are different fminsearch functions in Octave and the Octave Forge
>> >> optim package. Which one should be deleted?
>> >>
>> >> jwe
>> >
>> >
>> > Well, Andy should probably address that:
>> >
>> > changeset:   14895:e0525ecf156e
>> > parent:      14893:55d0f8d70fe9
>> > date:        Fri Jul 20 09:25:37 2012 -0700
>> > description:
>> > Add new function fminsearch.m
>> >
>> > The difference looks to be that fminsearch.m in Octave is a standalone
>> > script-code implementation whereas the fminsearch.m in optim package uses
>> > fmins.m which in turn uses nmsmax.m as well as a few others. fmins(),
>> > nmsmax() are not available in Octave.  Searching through optim package, it
>> > doesn't appear that a special library or anything is needed, it is just 
>> > that
>> > there are a number of extra files over which the work is distributed.
>> >
>> > Dan
>
> Yes, fminsearch in optim uses Etiennes option scheme. Now the optimset
> scheme prevails and it is probably better not to maintain two
> different schemes. So I'd say stick to the version in Octave. But
> first it would be necessary to provide nmsmax.m, called by
> fminsearch.m, in Octave, too.
>
> Olaf
>
> --
> public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net