This is the mail archive of the
gsl-discuss@sourceware.org
mailing list for the GSL project.
Re: Fwd: log probability density functions (PDFs)
- From: Sam Mason <s dot a dot mason at warwick dot ac dot uk>
- To: Patrick Alken <patrick dot alken at colorado dot edu>
- Cc: "gsl-discuss at sourceware dot org" <gsl-discuss at sourceware dot org>
- Date: Mon, 9 Jul 2012 22:04:54 +0100
- Subject: Re: Fwd: log probability density functions (PDFs)
- References: <CAPEJTf6T8vk+KgnoDBaZgJzjRUvJfLK+w-S-xwoZ_=f_wgNH3g@mail.gmail.com><CAPEJTf4jH_mmk_e5TiT1HxrF3fpLGozL2bikH4F4W_L+Rtpo2w@mail.gmail.com> <d75321b0fc2d4b7fb010d63b6252dd89@IDMEXCHT02.ads.warwick.ac.uk>
Hi Patrick,
I've put the following together:
http://www.samason.me.uk/hacking-the-gsl
Is that the sort of thing you'd consider linking to? Maybe with an
accompanying sentence of "log-probability density functions"?
Thanks,
Sam
On 9 July 2012 18:39, Patrick Alken <patrick.alken@colorado.edu> wrote:
> The best option would be to make a standalone extension (see the main
> gsl page for examples), so people can look at it and try it out. We can
> add the extension to the main web page and possibly add it to the main
> source code later if it looks useful.
>
> On 07/09/2012 11:29 AM, Sam Mason wrote:
>> Hi,
>>
>> I sent the following message to the help-gsl list a while ago, but
>> haven't received a reply...
>>
>> Is this sort of contribution useful to the project?
>>
>> Thanks,
>>
>> Sam
>>
>>
>> ---------- Forwarded message ----------
>> From: Sam Mason <s.a.mason@warwick.ac.uk>
>> Date: 15 June 2012 17:59
>> To: help-gsl@gnu.org
>>
>>
>> Hi,
>>
>> I sent a message to this a couple of months ago enquiring about the
>> availability of log-density functions. I've been using my own
>> versions of these since then, and thought it would be good to send
>> them back to the community.
>>
>> I've gone through most of the distributions and added in code
>> calculating their log-densities. I've copied the naming and layout
>> convention of the dirichlet distribution; which already had the
>> gsl_ran_dirichlet_pdf() and gsl_ran_dirichlet_lnpdf() functions. I've
>> not updated the docs yet, as I wanted to gage interest first.
>>
>> The code compiles and the tests run successfully on my computer. I've
>> also used the test code as a mini-benchmark, and the changed code
>> takes basically the same amount of time to execute (median = 1.0038
>> times GSL 1.15, 95%CI=1.0009-1.0064).
>>
>> Let me know what would make this more useful.
>>
>> Sam
>
>