philippeller / dama Goto Github PK

• when q is a scalar it excepts
• sometimes it returns nan, suspect when two samples have the same position (?)

Something to think about how to do that

KDE method gives nonsense results for density=False

-> Fix: compute sum over KDE grid + sum over events within boundaries, then reweight

Also, check if n-d bandwidth is correctly re-scaled for densities.

when weights are passed, then return_len cannot be calculated in binneddata

when np.nans are present then the bands are not plotted correctly (for grids with edges at least)

When using a grid array to create a mask and then apply that to the same gridarray, this works every second time only...

I.e. g[g > 0] for instance masks correctly. When executed again places where it was masked become zero. Next time it works again, and so on