[nadeem_malik]

Hi, I am new here, and I am also new to paraview -- so my apologies in advance for what will (I am sure) seem like a naive question for seasoned workers in this field.

My question is quite simple. I have inherited a fortran DNS (Direct Numerical Simulation) code for vortex interaction studies. I want to output and store variables such as the velocity field U(x,y,z,t) into a file (fort.42 say) and then read it in to paraview for the usual plots and visualisations and movies, etc.

The problem is the very first step. I have very large output data even in binary -- at the moment it is about 4GB, but it will increase to terabytes in the future for larger systems. The data is written in the fortran program in a simple binary format, e.g. (assuming regular cubic grid).

f1=1.
nx=260
ny=260
nz=260
open(unit=3,file='fort.42',form='unformatted')
write(3) nx,ny,nz
write(3) f1,f1,f1,f1
write(3) (((f1,i=1,nx),j=1,ny),k=1,nz), &
(((U(i,j,k,1),i=1,nx),j=1,ny),k=1,nz), &
(((U(i,j,k,2),i=1,nx),j=1,ny),k=1,nz), &
(((U(i,j,k,3),i=1,nx),j=1,ny),k=1,nz), &
(((f1,i=1,nx),j=1,ny),k=1,nz)

I compile using gfortran.

When I attempt to read fort.42 in paraview, it returns an error - it cannot read the binary file.

I would greatly appreciate an explanation. If someone would be kind enough to suggest how the above code could be modified, or maybe write a simple code to convert the fort.42 in to paraview readable binary data, I would be very grateful.

I guess for non-uniform grid, you would be write out each grid coordinate explicitly as well?

Thanks
Nadeem

[FMDenaro]

I am quite sure that Paraview has documentations for the input data files...have read it?

[nadeem_malik]

Originally Posted by FMDenaro

I am quite sure that Paraview has documentations for the input data files...have read it?

Dear FMDenaro,

I have looked through the help menu, and a few websites offering some advice. But this simple looking problem has not been addressed directly - at least I cannot find it.

On the paraview help menu, it mentions reading all kinds of data files, from *.csv to RAW data, etc. The problem is how to get he data in to any one of these formats from inside a fortran code in the first place?

Searching through the web, I found an example for reading in a simple *.csv file containing a scalar field f(x,y,z) laid out on a regular cubic grid. I got it read in by paraview -- although now I want to know how to produce is isoscalar surfaces s(x,y,z)=f=C, where C can be any constant in the range of scalar values? But *.csv are ASCII files.

How do I generate an equivalent binary *csv file using fortran -- is there such a things as a binary *.csv file ?

One way, in principle, could be to write a short interface file in C or C++, which reads fortran binary (is this possible in c++ ?), and then print out the same data but in paraview readable binary ?

But whay can't I do this directly in fortran ?

Thanks
Nadeem

[FMDenaro]

I personally don't use paraview but tecplot, however have a look here

http://people.sc.fsu.edu/~jburkardt/...io/vtk_io.html

https://github.com/szaghi/VTKFortran

[naffrancois]

Hello,

Here is a piece of code I am using for unstructured data sets. The important bit here is the list of arguments set in the open statement. Then, for your special case of structured grids, the syntax may be simpler. Have a look to this pdf:

http://www.vtk.org/wp-content/upload...le-formats.pdf

Good luck

[sbaffini]

You can't expect paraview to know how your data is organized, less than ever if it is also in Fortran binary form, which typically (I don't know if this is standardized) uses 4 + 4 bytes at the beginning and end of records.

You should use the STREAM format as per naffrancois example.

But I suggest to open the file-format file also linked by naffrancois, to look for the specific format more suited to you (note that Structured Points or vtkImageData might be the most suitable ones, depending on the method you use, i.e. Finite Differences or Finite Volumes).

[nadeem_malik]

Thank you to naffrancois and to sbaffini and to FMDenaro for your various responses and suggestions.

I'll get back to you in case of problems arising -- but I suspect that the information supplied is enough. Thanks.

Best
Nadeem Malik