[ncl-talk] qsort

[Buzan, Jonathan]

Hi Dennis,

I will work on coding up the suggested algorithm…
However, it might be easier to rephrase the question:

I have an array of HI(time), T(time), and RH(time).
I want to qsort(HI), then reorder T and RH to match qsort(HI).

Is there a way to do that?
Because I can simply use index numbers, and get around this issue.

-Jonathan




On Feb 25, 2019, at 5:31 PM, Dennis Shea <shea at ucar.edu<mailto:shea at ucar.edu>> wrote:

Jonathan:

For a sort variation, you might try dim_pqsort_n<https://www.ncl.ucar.edu/Document/Functions/Built-in/dim_pqsort_n.shtml>  which is from the fortran SLATEC <https://www.netlib.org/slatec/guide> library.

NCL's dim_pqsort_n should be able to handle any dimension size.

Under-the-hood:   To my knowledge, fortran's default integer type is 32-bits [ 4 bytes ] => 2,147,483,647
NCL's C-interface checks the size of the dimension being sorted. I it  exceeds the 32-bit size, NCL uses a special version that allows [64-bit; 8-byte; INTEGER*8] indexing.

Cheers
D


On Mon, Feb 25, 2019 at 8:17 AM Buzan, Jonathan <jbuzan at purdue.edu<mailto:jbuzan at purdue.edu>> wrote:
Hi Rick,

I am running into an issue where I have some repeat values.  Is there a way to add a precision level adjustment during the qsort algorithm?

-Jonathan




On Feb 25, 2019, at 16:06, Rick Brownrigg <brownrig at ucar.edu<mailto:brownrig at ucar.edu>> wrote:

Hi Jonathan -- yes, qsort() takes a double, or more generally a "numeric" type.

RB

On Mon, Feb 25, 2019 at 7:20 AM Buzan, Jonathan <jbuzan at purdue.edu<mailto:jbuzan at purdue.edu>> wrote:
Hello NCL-Talk,

Does qsort operate at double precision?
I am trying to resolve my stat_dispersion issue by isolating the problem...

-Jonathan


_______________________________________________
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
_______________________________________________
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
<dpsort.f>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20190226/0ea3bba0/attachment.html>