Closed keewis closed 3 months ago
dask
-awareness and subdomain convolution seem like bigger tasks, so I'll merge this and revisit these two items in new PRs.
The tests are not quite sufficient to verify that it is, indeed, a gaussian kernel we got (right now it's simply checking that we got a normalized kernel). I'm still trying to find a way to verify that without having to compare with a pre-computed kernel.
Building on top of the neighbours search and the distance calculation, adds a function to compute gaussian kernels from cell ids.
Intended features:
healpy
's smoothing function)dask
to allow computing the kernel without running out of memory