[ncl-talk] could i try dim_kde function before ncl 6.5 released

Kevin Hallock hallock at ucar.edu
Fri Aug 4 16:43:13 MDT 2017


Hi Qi,

I just created a new “development snapshot” build of NCL available via the “conda” package manager containing the dim_kde function. Please see our conda documentation for installation information, if needed: http://ncl.ucar.edu/Download/conda.shtml <http://ncl.ucar.edu/Download/conda.shtml>

In order to install the “ncl dev” version, you will need to run a different “conda create” command than what is shown in our documentation. Please use the following commands to install and activate an “ncl_dev” environment with conda:
conda create -n ncl_dev -c ncar -c conda-forge ncl=dev gsl
source activate ncl_dev

Note that the “dim_kde” function is currently called “dim_kde_test” and is not officially supported; we will revert the name change at some point in the future once the function is deemed stable.

Please let us know if you experience any unexpected behavior with this function.

Thanks,
Kevin

> On Jul 27, 2017, at 4:18 PM, Mary Haley <haley at ucar.edu> wrote:
> 
> Qi,
> 
> For the record, we do have a dim_kde already implemented in our internal development version, but we haven't had a chance to heavily test it.
> 
> We are going to revisit this tomorrow to see what state it's in. If you are interested, we can provide you with a test NCL to try.
> 
> --Mary
> 
> 
> On Wed, Jul 26, 2017 at 11:32 AM, LI Qi <liqi123sh at qq.com <mailto:liqi123sh at qq.com>> wrote:
> Dear Dennis,
> 
> Thank you for your reply!
> Yesterday I just googled "ncl kernel density estimation" and that page popped up. At that time Im so happy to find NCL also has this function  (means I don't need to learn R).
> What's your said (hist_binwidth and invoking kde ) is exactly what I want.
> But if it's very bother to invoke the funciton or it's need much time and energy in the develepment, then just ignore me. I can learn R for the moment.
> 
> Best,
> Qi
> 
> 
> ------------------ Original ------------------
> From:  "Dennis Shea";<shea at ucar.edu <mailto:shea at ucar.edu>>;
> Date:  Thu, Jul 27, 2017 00:42 AM
> To:  "LI Qi"<liqi123sh at qq.com <mailto:liqi123sh at qq.com>>;
> Cc:  "ncl-talk"<ncl-talk at ucar.edu <mailto:ncl-talk at ucar.edu>>;
> Subject:  Re: [ncl-talk] could i try dim_kde function before ncl 6.5 released
> 
> Placing the 6.5.0 KDE function documentation onto the publicly available NCL URL was a mistake on my part. 
>           https://www.ncl.ucar.edu/Document/Functions/Built-in/dim_kde.shtml <https://www.ncl.ucar.edu/Document/Functions/Built-in/dim_kde.shtml>
> It was meant for NCL's internal draft documentation URL. I used
> 
> %> unw dim_kde.xml
> 
> instead of
> 
> %> unw -t dim_kde.xml
> 
> ----
> 
> If you look at:
>           http://www.ncl.ucar.edu/future_release.shtml <http://www.ncl.ucar.edu/future_release.shtml>
> there is no mention of the 'dim_kde' function. Hence, 'dim_dke' is not yet in the development NCL 6.5.0. There is no time estimate of when it will be included.
> 
> ---
> An NCL script that invokes the KDE fortran subroutine via a shared object is available. It returns the PDF (ie: "smooth curve"). I can make this available. Let ncl-talk know if you want this. 
> 
> There is also an as-yet undocumented NCL language 'hist_binwidth'  that estimates optimal bin-widths using Silverman's rule-of-thumb for normal and quantile densities; and, uses the 'dim_kde' function. An NCL script that invokes the shared object is available for this also.
> 
> Regards
> 
> On Tue, Jul 25, 2017 at 2:23 PM, LI Qi <liqi123sh at qq.com <mailto:liqi123sh at qq.com>> wrote:
> Dear NCL,
> 
> Is there any way I can try the gaussian kernel density estimation (KDE) (https://www.ncl.ucar.edu/Document/Functions/Built-in/dim_kde.shtml <https://www.ncl.ucar.edu/Document/Functions/Built-in/dim_kde.shtml>) before next version released?
> I have sample histograms and would like a smooth curve overlaid.
> 
> Thanks and best regards,
> Qi
> 
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu <mailto:ncl-talk at ucar.edu>
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
> 
> 
> 
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu <mailto:ncl-talk at ucar.edu>
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
> 
> 
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20170804/7d927f80/attachment.html 


More information about the ncl-talk mailing list