On Fri, May 11, 2012 at 9:53 PM, Levente Torok <torok...@gmail.com> wrote:
> On Fri, May 11, 2012 at 5:28 PM, Michael Creel <michael.cr...@uab.es> wrote:
>> On Fri, May 11, 2012 at 1:47 PM, Carnë Draug <carandraug+...@gmail.com>
>> wrote:
>>> On 11 May 2012 12:32, Levente Torok <torok...@gmail.com> wrote:
>>>> On Fri, May 11, 2012 at 1:22 PM, Carnë Draug <carandraug+...@gmail.com>
>>>> wrote:
>>>>> On 11 May 2012 07:59, Levente Torok <torok...@gmail.com> wrote:
>>>>>> I wanted to analyze gmm_exampe.m from econometrics package recently
>>>>>> installed but I found that it called a lot of deprecated functions.
>>>>>> I tracked them and replaced. Now it seems to be working.
>>>>>> I wish I could submit these modifications back to the package sources
>>>>>> but as I see I cannot find the way to patch it in the svn tree where I
>>>>>> found several versions of the corresponding file so I'd rather not to
>>>>>> touch it
>>>>>> instead I thought I share these modifications. I hope someone can put
>>>>>> it the package
>>>>>>
>>>>>> gmm_example.m:
>>>>>> replace call of poisson_rnd with poissrnd
>>>>>> str2mat call str2mat with char
>>>>>>
>>>>>> poisson_moments.m :
>>>>>> replace call of dmult.m with diag(A)*B
>>>>>>
>>>>>> scale_data.m:
>>>>>> replace call of dmult.m with diag(A)*B
>>>>>>
>>>>>> gmm_results.m
>>>>>> replace call of str2mat with char
>>>>>> chisquare_cdf with chi2cdf
>>>>>> normal_cdf with normcdf
>>>>>>
>>>>>> poisson_moments.m:
>>>>>> replace call of dmult.m with diag(A)*B
>>>>>>
>>>>>> unscale_parameters.m
>>>>>> replace call of nth with {1}{2}
>>>>>
>>>>> this is the wrong mailing list. Octave packages belong to octave-forge
>>>>> so bugs should be reported there.
>>>>>
>>>>> All the problems you mentioned have already been fixed on the
>>>>> development version (that's why you couldn't patch against the SVN
>>>>> tree) we just need someone to make a new release of the package.
>>>>>
>>>>
>>>> I see. Thanks. It was kind of a problematic stuff to me.
>>>> Wouldn't it be possible to keep forge and octave in sync?
>>>> I mean I would let someone install forge package unless it is
>>>> absolutely sure that it will function with the installed octave
>>>> version.
>>>
>>> Yes, it's possible and there's more who would like that to happen.
>>> Unfortunately, there's no one around with the means to put something
>>> like that in place.
>>>
>>> Carnë
>>
>> Just a note on these functions: all of the MPI capability depends on
>> MPITB, which no longer works with current Octave. At some point I plan
>> of adapting the functions to use openmpi_ext, but I don't know when
>> that will be possible. About OF package releases, I haven't kept up
>> with how that works.
>> Best,
>> Michael
>
> Hi Michael,
>
> Are talking about gmm / poisson / dmult functions or parcellfun ?
> I have made some experiment with parcellfun and it seems to speed up
> on multicore machines for this reason it is not clear what you are
> pointing at.
>
> Cheers,
>
> Lev
In gmm_estimate and mle_estimate, you'll see some code for using
MPITB. That allowed for use of multiple cores or multiple machines
(cluster), but it no longer works. It will be quite easy to make it
use openmpi_ext.
M.
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Octave-dev mailing list
Octave-dev@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/octave-dev
