Lukas Riedel

icon indicating a website link https://orcid.org/0000-0002-4667-3652

icon location Zürich, Switzerland icon location PostDoc @ ETH Zürich & MeteoSwiss Researching climate-related socio-economic impacts and risks

Results 24 comments of Lukas Riedel

Usage: Poor performance of NaN-aware xarray computations?

I found that dispatching to `sparse.nansum` instead of relying on `xarray.computation.nanops.nansum` gives a significant speedup, see https://github.com/pydata/xarray/issues/10922

Usage: Poor performance of NaN-aware xarray computations?

@hameerabbasi Are there any "guarantees" about the fill value *not* being present in the data? Meaning: Is there a way to be sure that the matrix was pruned and is...

`sum(..., skipna=True)` should dispatch to `sparse.nansum` for sparse arrays

I am not sure if I can follow, because the details on `__array_function__` and `__array_ufunc__` elude me 😬 But here's what happens on the xarray side of things for `sum()`:...

`sum(..., skipna=True)` should dispatch to `sparse.nansum` for sparse arrays

@Illviljan I tried to dig into this a bit more and noticed I made an embarrassing mistake: Instead of dispatching to `nanname` I dispatched to `name` in the above patch,...