[Nek5000-users] Calling subroutine "hpts" for several files

[nek5000-users at lists.mcs.anl.gov]

Thanks!


/Johan

________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Tuesday, October 17, 2017 12:27:47 PM
To: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Calling subroutine "hpts" for several files


Hi Johan,


Yes -- you can have just one list on node 0 if you wish.


Make certain that the list length n=0 on on all other nodes (e.g.,


      if (nid.gt.0) n=0

      call the routine (..... , n)


The only reason to distribute the list is if it large and you don't

want to store it all on node 0.


The interpolation routine handles all data movement, as it should.


Best,

Paul


________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Tuesday, October 17, 2017 3:35:58 AM
To: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Calling subroutine "hpts" for several files


Thanks. After running the hemi example and looking into hemi.usr, I have some questions about how the interpolation works in parallell.

It seems as the interpolation is done by a call to the subroutine "intpts" in from the subroutine "interp_v":


"call intpts(wrk,3,pts,n,uvw,.true.,.true.,ihandle)"


This call interpolates velocities in wrk to positions in pts and outputs in uvw.

The positions "pts" are positions of Lagranian particles.


Each process has a local array "pts" and a sees a local piece of the velocity field, "wrk".

Is it important that each point of "pts" is within the local piece of "wrk" for a processor?


* If this is important; how can I see that this is considered within the code?


* If it is not important; Could "intpts" be called only from one process (say nid == 0) in order to simplify the code?


Also: Have I understood the hemi example correctly?


Best,


Johan

________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Monday, October 16, 2017 2:26:20 PM
To: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Calling subroutine "hpts" for several files


Dear Johan,


I would look at the hemi example.


There it shows how to interpolate a list of values

(interp_v is a routine inside hemi.usr).


Once you have a list, you can write it out yourself

(again, as shown in hemi.usr).


Make certain that your interrogation list is _not_ repeated

on every processor. (Otherwise you end up doing P times

more work.)


hth,

Paul


________________________________
From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
Sent: Monday, October 16, 2017 4:04:02 AM
To: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] Calling subroutine "hpts" for several files


Dear Neks,


I am post processing my data using the subroutine "hpts" to extract field values at specific points.

I want to loop over many files and extract the field values for each file.

Therefore I want to rename the file "hpts.out" that is created everytime I call hpts.

This code snippet is in userchk, and is looks like this:


      do i=1,100

        write(fname,'(I0.5)') i
        fname='cav0.f'//trim(fname)

        call load_fld(fname)

        call hpts

        call nekgsync
        if(nid==0)then
          call rename('hpts.out',trim(fname)//'_hpts')
        endif
        call nekgsync

      enddo

My basename is "cav", so fname is "cav0.f00XXX" for XXX = 1, ..., 100.
The code is intended to rename hpts.out for a file "cav0.f00XXX" into "cav0.f00XXX_hpts".

However, this does not work.
I get only one file; "cav0.f00001_hpts".
When I open this file it contains the data from "cav0.f000100 - the last file that was opened in the loop!

This makes me wonder if hpts.out is properly closed after beeing written to?
And also, how can I change my code to make it work as I indended?

Best,

Johan




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20171017/72b9a658/attachment-0001.html>