Accessing data stored in complex arrays

Question

I'm writing some fortran code where I'm having difficulty visualizing how the data in a hierarchy of arrays is stored, which is causing me troubles with how to manipulate a subset of this data. I've read-in some data stored in an unformatted binary PLOT3D file. The data format looks like the following:

1 int (nblocks): number of blocks (or zones) in a computational mesh
4 x nblocks ints: (ni(m), nj(m), nk(m), nvars(m)): number of i, j, and k points in each block along with a number of solution variables in each block.
loop over blocks (m)
ni(m) x nj(m) x nk(m)x nvars(m) floats (q(ni,nj,nk,nv,m)): solution variables at each i, j, and k points for each block.
end loop over blocks

program my_program

use iso_fortran_env

implicit none

character(*), parameter :: soln_file = 'my_file_name.q'

integer :: nblks, io_stat, imax, jmax, kmax, nv, m

integer, dimension (:), allocatable :: ni, nj, nk, nvars

real(real64), dimension (:,:,:,:), allocatable :: qq
real(real64), dimension (:,:,:,:,:), allocatable :: q

open(unit=10, form='unformatted', file=soln_file, status='old', iostat=io_stat)

if ( io_stat /= 0 ) then
    write(*,*) '*** Error reading solution file ', soln_file, ' ***'
    stop
end if

read(10) nblks

allocate( ni(nblks), nj(nblks), nk(nblks) )

read(10) ( ni(m), nj(m), nk(m), nvars(m), m = 1, nblks )

imax = maxval(ni)
jmax = maxval(nj)
kmax = maxval(nk)
nv = maxval(nvars)

allocate( q(imax,jmax,kmax,nv,nblks) )

do m = 1, nblks
    allocate( qq(ni(m),nj(m),nk(m),nvars(m)) )

    read(10) qq(ni(m),nj(m),nk(m),nvars(m))

    q(1:ni(m),1:nj(m),1:nk(m),1:nvars(m),m) = qq

    deallocate( qq )
end do

close(10)

deallocate( ni, nj, nk, nvars, q )

stop

end program my_program

In my case, I'm interested in extracting a subset, or just modifying the values in place, of a single solution variable at all of the points in every block. It seems like I should end up with the solution values for that variable at each i, j, and k point for every block in a 3 x nblocks array.

However, I get a seg fault at run time which indicates to me that I'm not sizing the array correctly.

I've updated my original comment with the snippet of relevant code from my program to illustrate how the array is being formed. The array q holds solution data at each i, j, and k point in a block for multiple blocks. Further, the solution data exists in any number of variables. I'm interested in how to extract a subset of the of the solution data for just a single variable. — user3394147
– user3394147, Commented Jun 22, 2019 at 14:37
Your code cannot be compiled. Prepare a real minimal reproducible example, test it and make sure it exhibits the right error. There is really nothing special about complex arrays, most likely you have some kind of error that can happen anywhere. To find it you must show the actual code. — Vladimir F Героям слава
– Vladimir F Героям слава, Commented Jun 22, 2019 at 14:41
Yes, but no-one can test it, because we do not have the required data files. Can't you really make a minimal example that does not require external data? — Vladimir F Героям слава
– Vladimir F Героям слава, Commented Jun 22, 2019 at 16:11
No, I can't provide the actual solution file, or a working sample of it. It's a a very large unformatted binary file, and I don't have a visual understanding of how the data within it is structured beyond the simple header which lists the number of blocks, and the i-, j-, and k-extents for each block. My question is more fundamental in how to extract a subset of a multidimensional array. In this specific example, I'm looking for an answer as to how to extract all of the solution data for 1 variable. — user3394147
– user3394147, Commented Jun 22, 2019 at 17:20

PTRK · Accepted Answer · 2019-06-24 13:05:01Z

1

In your block of code

do m = 1, nblks
    allocate( qq(ni(m),nj(m),nk(m),nvars(m)) )

    read(10) qq(ni(m),nj(m),nk(m),nvars(m))

    q(1:ni(m),1:nj(m),1:nk(m),1:nvars(m),m) = qq

    deallocate( qq )
end do

You're only reading 4 values pour qq, instead of ni(m) x nj(m) x nk(m) x nvars(m). Then you're trying to copy theses values into q wich is not conformant.

It seems to me that the correct loop should be

do m = 1, nblks
    allocate( qq(ni(m),nj(m),nk(m),nvars(m)) )

    do iv = 1 ,nvars(m)
      do ik = 1 ,nk(m)
        do ij = 1 ,nj(m)
          do ii = 1 ,ni(m)
             read(10) qq(ii,ij,ik,iv)
          end do
        end do
      end do
    end do

    q(1:ni(m),1:nj(m),1:nk(m),1:nvars(m),m) = qq

    deallocate( qq )
end do

I'm not sure this will solve your problem, but there's something fishy here.

answered Jun 24, 2019 at 13:05

PTRK

8991 gold badge5 silver badges19 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Accessing data stored in complex arrays

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related