Closed jcgraciosa closed 4 months ago
I assumed that the petsc writer and h5 reader would match up automatically. You can try the fix I just checked in and see if it works more generally (it seems to fix this code).
Thanks for the fix! I thought of fixing in the write side, but doing it in the read side makes sense. I've tested the fix and it works. Closing the issue.
Describe the bug When reading a discontinuous mesh variable(in this case, dP1 or dP2) using the
read_timestep()
function, the dimensions of the coordinates and data obtained from the h5 file are incorrect.What is the bug? When the
write()
function indiscretisation.py
is executed (line 1833-1894), it seems that the dimensionality of the discontinuous mesh variable is tied to that of the number of mesh elements (Number of 2-cells per rank
). This causes and error to be thrown duringread_timestep
.What version code? This issue is found in the development and uw_constants branches
Steps to reproduce: The following can be run the reproduce the issue:
This is a portion of the read_timestep function that shows that the data stored in the h5 files now have incorrect dimensions.