This issue is about adding a test to assess SAMRAI's possible bug that leads to missing overlap calculations for data lying on faces.
The idea would be to setup a 2D Patch Hierarchy with refinement boxes so that two L1 patches are separated by a distance such that the right (for example) border of the ghost box of the first lies exactly on the left border of the box (not ghost box) of the second patch.
initialize L0 with some data, then initialize L1 with some different data.
And then run a regridding schedule on L1. We should expect that all faces of both L1 patches are obtained from L0 refinement except those faces lying on the right border ghost box of the first L1 patch which should be copied from data on the second patch.
This issue is about adding a test to assess SAMRAI's possible bug that leads to missing overlap calculations for data lying on faces.
The idea would be to setup a 2D Patch Hierarchy with refinement boxes so that two L1 patches are separated by a distance such that the right (for example) border of the ghost box of the first lies exactly on the left border of the box (not ghost box) of the second patch.
initialize L0 with some data, then initialize L1 with some different data.
And then run a regridding schedule on L1. We should expect that all faces of both L1 patches are obtained from L0 refinement except those faces lying on the right border ghost box of the first L1 patch which should be copied from data on the second patch.