coupling parameters

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
turuncu
Posts: 128
Joined: Tue Feb 01, 2005 8:21 pm
Location: Istanbul Technical University (ITU)
Contact:

coupling parameters

#1 Unread post by turuncu »

Hi,

I try to couple WRF and ROMS and i set the parameters but i got following error,

MOD_COUPLER - Number assigned processors: 001
not equal to spawned MPI nodes: 004

MOD_COUPLER - Number assigned processors: 001
not equal to spawned MPI nodes: 004

MOD_COUPLER - Number assigned processors: 001
not equal to spawned MPI nodes: 004

MOD_COUPLER - Number assigned processors: 001
not equal to spawned MPI nodes: 004

I submit the job using 4 proc and my coupling paramaters like that,

Code: Select all

        Nmodels = 2

! Coupled models order label used to process information arrays, [1:Nmodels]
! string values are expected. The order is arbitrary, the indices below are
! ordered from bottom to top grids.

     OrderLabel = ocean \
                  atmos

! Number of parallel threads assigned to each model in the coupled system,
! [1:Nmodels] KEYWORD entries are expected with the order label (OrderLabel)
! in parenthesis. The sum of all entries must be equal to the total number
! of processors.

Nthreads(ocean) = 2
Nthreads(atmos) = 2

! Time interval (seconds) between coupling of models, [1:Nmodels] real
! values are expected corresponding to the entries of a lower triangular
! matrix Aij for i>j.  For example, for three model components ordered
! as ocean, waves, and atmos, the time interval is read as:
!
!      Aij     1      2     3
!      i\j   ocean  waves atmos
!  1  ocean    -      -     -          v(1): ocean-waves coupling interval
!  2  waves   v(1)    -     -          v(2): ocean-atmos coupling interval
!  3  atmos   v(2)   v(3)   -          v(3): waves-atmos coupling interval
!
!  TimeInterval = v(1) v(2) v(3)

   TimeInterval = 0.0 120.0d0

! Coupled model standard input file name, [1:Nmodels] KEYWORD entries are
! expected with the label code in parenthesis.

 INPname(ocean) = med_ocean.in
 INPname(waves) = med_atmos.in

! Coupled models variables information file name.

        CPLname = coupling.dat
...
...
...
the med_ocean.in contains the ocean component parameters but med_atmos.in is empty for now. I expect that model must use 2 proc in ocean component and 2 proc for atmospheric component. I put some print statement into main coupling code but the atmos component PET list is empty. I did not find any example about WRF+ROMS coupling parameters so if anyone has such kind of example please let me know. The could be a problem in my configuration.

Thanks,

turuncu
Posts: 128
Joined: Tue Feb 01, 2005 8:21 pm
Location: Istanbul Technical University (ITU)
Contact:

Re: coupling parameters

#2 Unread post by turuncu »

I found a bug in ROMS rev 832 in ROMS/Utility/inp_par.F file which triggers the error. To solve it go line 736 and replace 'waves' with 'atmos' in if else statement. There is two waves definition out there.

The code works after fixing this file. Anyway, thanks :)

User avatar
arango
Site Admin
Posts: 1367
Joined: Wed Feb 26, 2003 4:41 pm
Location: DMCS, Rutgers University
Contact:

Re: coupling parameters

#3 Unread post by arango »

turuncu wrote:I found a bug in ROMS rev 832 in ROMS/Utility/inp_par.F file which triggers the error. To solve it go line 736 and replace 'waves' with 'atmos' in if else statement. There is two waves definition out there.
I assume that meant version 382 instead. I don't see the two waves definitions that you are talking about:

Code: Select all

            DO i=1,Nmodels
              IF (i.eq.Nval) THEN
                OrderLabel(i)=TRIM(ADJUSTL(Cval(Nval)))
                IF (INDEX(TRIM(OrderLabel(i)),'ocean').ne.0) THEN
                  Iocean=i
                ELSE IF (INDEX(TRIM(OrderLabel(i)),'waves').ne.0) THEN
                  Iwaves=i
                ELSE IF (INDEX(TRIM(OrderLabel(i)),'atmos').ne.0) THEN
                  Iatmos=i
                END IF
              END IF
            END DO
Your coupling.in looks good to me. However, the error that you are getting is about a mismatch between the number of nodes assigned to each couple model and the total number of spawned MPI nodes.

turuncu
Posts: 128
Joined: Tue Feb 01, 2005 8:21 pm
Location: Istanbul Technical University (ITU)
Contact:

Re: coupling parameters

#4 Unread post by turuncu »

sorry, the revison is 382 and i download it yesterday. I don't know why but may be it is my fault. So anyway, thank you look into this.

Post Reply