Repository navigation
Dataset.broadcast_like(other) should broadcast against like variables in other #6549
Description
Is your feature request related to a problem?
I am a bit puzzled about how xarrays is broadcasting Datasets.
It seems to always add all dimensions to all variables.
Is this what you want in general?
See this example:
import xarray as xr
da = xr.DataArray([[1, 2, 3]], dims=("x", "y"))
# <xarray.DataArray (x: 1, y: 3)>
# array([[1, 2, 3]])
ds = xr.Dataset({"a": ("x", [1]), "b": ("z", [2, 3])})
# <xarray.Dataset>
# Dimensions: (x: 1, z: 2)
# Dimensions without coordinates: x, z
# Data variables:
# a (x) int32 1
# b (z) int32 2 3
ds.broadcast_like(da)
# returns:
# <xarray.Dataset>
# Dimensions: (x: 1, y: 3, z: 2)
# Dimensions without coordinates: x, y, z
# Data variables:
# a (x, y, z) int32 1 1 1 1 1 1
# b (x, y, z) int32 2 3 2 3 2 3
# I think it should return:
# <xarray.Dataset>
# Dimensions: (x: 1, y: 3, z: 2)
# Dimensions without coordinates: x, y, z
# Data variables:
# a (x, y) int32 1 1 1 # notice here without "z" dim
# b (x, y, z) int32 2 3 2 3 2 3Describe the solution you'd like
I would like broadcasting to behave the same way as e.g. a simple addition.
In the upper example da + ds produces the dimensions that I want.
Describe alternatives you've considered
ds + xr.zeros_like(da) this works, but seems more like a "dirty hack".
Additional context
Maybe one can add an option to broadcasting that controls this behavior?
Activity
keewis commentedon Apr 30, 2022
see also #6304 which covers
xr.broadcastheadtr1ck commentedon Apr 30, 2022
I tried adding a
joininput toDataset.broadcast_likeand passing it toalign, but that did not work (at least forjoin="inner"). Still got the same result...headtr1ck commentedon May 1, 2022
related to #6227
[-]Improved Dataset broadcasting[/-][+]Add a broadcasting mode that inserts size-1 labeled dimensions[/+][-]Add a broadcasting mode that inserts size-1 labeled dimensions[/-][+]Add a broadcasting mode that inserts size-1 unlabeled dimension[/+][-]Add a broadcasting mode that inserts size-1 unlabeled dimension[/-][+]Improved Dataset.broadcasting[/+]dcherian commentedon Feb 6, 2025
I keep misunderstanding this issue so typing this out to make sure I got it right.
Writing out dimension names in square brackets
IIUC the request is to avoid broadcasting the variables in
dsagainst each other, and to only broadcast each variable againstdaseparately. Did I get it right?[-]Improved Dataset.broadcasting[/-][+]`Dataset.broadcast_like(other)` should broadcast against like variables in other[/+]dcherian commentedon Feb 6, 2025
@mjwillson posted a nice summary in #10031 :
alvarosg commentedon Feb 6, 2025
In the light of #10031, and broadcasting a dataset to itself not behaving as a no-op, should we label this as "bug" rather than "enhancement"?
headtr1ck commentedon Oct 27, 2025
I found a use-case for broadcasting a dataset like requested: When trying to change values in-place, this only works if the replacement values are already broadcasted.
An example can be found here: https://github.com/pydata/xarray/blob/main/xarray/computation/computation.py#L842 where instead of broadcasting I had to resort to adding a zero-valued dataset.