Data consistency question regarding vtkhdf and OverlappingAMR

Support
1

Hi there,

I’m wondering, does the OverlappingAMR VTKHDF format need to be padded with zeros or duplicate values for the case where a level might be missing across time?

For instance, if I write point data/boxes etc for a timestamp that has 2 levels

then a timestamp with 1 level as the refined level was dropped

then a timestamp with 2 levels again as the refined level returned.

I get the feeling there can be a data inconsistency without padding for the case with a missing level.

I hope this makes sense, if not, please let me know.

I have a plan to test this to confirm, but it would be nice to get an answer for you folks too

Thanks

2

@Louis_Gombert @lgivord PTAL

3

hello @PhilipDeegan,

Indeed, when I added temporal support for overlapping AMR I didn’t take in account this use case where the number of level can change. A workaround for now is indeed to add some padding to 0.

tbh, I don’t think it would be complex to add some information in `/VTKHDF/Steps/LevelX/…` as attribute or (hdf) dataset to support it.

Do you think you can open an issue in VTK ? https://gitlab.kitware.com/vtk/vtk/-/issues