Hello,
I use a FEM toolbox and create the output as hdf5 files (with an xdmf file, that I load in ParaView 5.6).
If I set the start time of the simulation not to 0 and make the time step size smaller than 10^-5 (I get the output after every timestep) I can’t open the files in ParaView - if I increase the time step size a little bit, it works perfectly.
The corrupt file throws the following error:
Warning: In /build/paraview/src/ParaView-v5.6.0/VTK/IO/Xdmf2/vtkXdmfReader.cxx, line 433
vtkXdmfReader (0x561e19e51420): Data type generated (vtkMultiBlockDataSet) does not match data type expected (vtkUnstructuredGrid). Reader may not produce valid data.
Since I really thought this is a strange error I tried to open the files in Visit, which can load even the files with the smaller time step size.
I have uploaded a small example file here:
Not working example: https://up.labstack.com/WAxsHpiG
Working example: https://up.labstack.com/Ee8ocOco
Can anybody help me with this issue?
Greetings,
Mathias
I took a look into your non working data.
It looks like you are defining multiple grid for a single timestep. I do not think this is intended.
If I am right, you just need to modify the value of the <Time> attribute, like in this file. solution.xdmf (3.0 KB)
If you really want to specify multiple grid for the same timestep, it looks like this is not supported by our reader yet. I’m not entirely sure this is a bug, has Xmf format is quite generalistic. Feel free to open an issue on our gitlab in this case. https://gitlab.kitware.com/vtk/vtk/issues
In any case, we should at least give a better analysis of this issue in the reader.
Hi Mathieu,
thanks for this really fast response! I think the mistake is then on my side, because I don’t have multiple grids. I have to look in my code what goes wrong there, I really didn’t think that this is caused on my side because it works most of the time (when not using small time steps). I will debug this further and report back here, unfortunately i also face some other problems at the moment so that I don’t have the time to investigate this issue right now, but it’s on my todo list.
Thanks again for your help and investigating my problem,
I know this is a bit late, but did you find a solution to this problem?
I seem to be having the exact same issue all of a sudden. I get the same error when opening an XDMF file where the timesteps are small, but if i just multiply all the times by 1000 I am able to open them in Paraview (with the time saved as integers). It seems that paraview wants to round all the times to the nearest integer for some reason, and is then unable to load files because there are multiple solutions at time t=0.
I am using the same mesh at every timestep, and this was working without problems until recently.
I have found the problem and solution. My python locale was set to a locale that uses the “,” symbol for decimals rather than “.”. My xdmf files thus had timestamps reading 0,004 instead of 0.004 so the time was interpreted as 0. This meant multiple solutions at t=0 and hence the error. I fixed the problem by changing my locale in python.