Philip Chmielowiec

> Thanks for the quick response! Making this change appears to result in a different set of bad data That's interesting, it looks like there's a mismatch somewhere with the...

```Python def plotit_raster(uxda: ux.UxDataArray) -> None: plotstart = time.time() fig, ax = plt.subplots(1, 1, figsize=(8,4), dpi=300, constrained_layout=True, subplot_kw=dict(projection=ccrs.PlateCarree())) ax.add_feature(cfeature.NaturalEarthFeature(category='cultural', scale='50m',edgecolor='k',facecolor='none',linewidth=0.2, name='admin_0_countries')) ax.add_feature(cfeature.NaturalEarthFeature(category='physical',color='k',facecolor='none', linewidth=0.5, scale='10m', name='coastline')) ax.set_global() raster = uxda.to_raster(ax=ax) img...

If you'd like to increase the resolution, we can set a `pixel_ratio` parameter too to sample more pixels. https://uxarray.readthedocs.io/en/latest/user-guide/mpl.html#controlling-the-resolution

pre-commit.ci autofix

> @philipc2 And I also thought carefully about optimizing the loop you mentioned: > > > Something like the following should work if we put it inside of the loop?...

@hongyuchen1030 Here's an optimized version of `get_cartesian_face_edge_nodes` using Numba. Here we re-use the `face_edge_connectivity` and `edge_node_connectivity` that can be computed internally and re-used. This runs in about `90ms` for me,...

> > ```python > > face_node_connectivity > > ``` > > Thanks for the information! This is actually what I had in mind for `get_cartesian_face_edge_nodes` using `face_edge_connectivity` and `edge_node_connectivity`. However,...

> > @hongyuchen1030 > > Here's an optimized version of `get_cartesian_face_edge_nodes` using Numba. Here we re-use the `face_edge_connectivity` and `edge_node_connectivity` that can be computed internally and re-used. > > This...

> > Looks like there is some small differences. We can move this optimizing into the work I'm putting together in #1195 and keep this PR smaller in scope. >...

I apologize if I've caused any confusion. > I understand that you’d prefer not to keep these variables in Grid._ds, but you previously mentioned that it’s okay to have them...