Huite

There is still a potential inconsistency with the OverlapRegridder. Any overlap with a source face is sufficient to assign a value. The centroid would actually have to be located in...

We needed to compare grids in imod-python, what exactly is going wrong? There's a not a lot of info in the linked issue.

Looks like numpy_indexed might be significantly faster: ```python import numpy as np import numpy_indexed as npi nodes = np.arange(2_000_000) a = np.random.choice(nodes, 500_000) b = np.random.choice(nodes, 500_000) edges = np.column_stack((a,...

Not quite right yet...: ```python class UniqueRows: """ Numpy unique is extremely useful, but can be sped up by roughly a factor of two for our use, and some of...

It looks like the above is overkill, the most important part is generating a void dtype. Then np.unique is as fast as any other implementation. Might be worthwhile to use...

Some thoughts: * The centroid is a more "robust" coordinate, because it is guaranteed to be inside of a convex polygon. The circumcenter may be located outside. * However, it's...

Hi @saeed-moghimi-noaa The user guide part on selection should give relevant information on this: https://deltares.github.io/xugrid/examples/selection.html The `.ugrid.sel` method will return subsets with valid topologies. Note that this currently selects by...

While debugging this, I came across at least one bug with the creation of a dual grid (the centroidal voronoi tesselation) of only existing centroids (rather that also the projection...

Took rather a lot of digging, but I found the issue:  There's a weird edge in the edge_node_connectivity, which isn't associated with any face. ```python fig, ax = plt.subplots()...

Trying to solve this internally creates a lot of headache. I've added some checking for the edge_node_connectivity now (9a5bb600490c5807389da6f239ab510ad86c7bc8), it'll raise a ValueError stating which edges are the problem (in...