scatter plot is slow

[mktippett]

What happened?

scatter plot is slow when the dataset has large (length) coordinates even though those coordinates are not involved in the scatter plot.

What did you expect to happen?

scatter plot speed does not depend on coordinates that are not involved in the scatter plot, which was the case at some point in the past

Minimal Complete Verifiable Example

import numpy as np
import xarray as xr
from matplotlib import pyplot as plt
%config InlineBackend.figure_format = 'retina'
%matplotlib inline

# Define coordinates
month = np.arange(1, 13, dtype=np.int64)
L = np.arange(1, 13, dtype=np.int64)

# Create random values for the variables SP and SE
np.random.seed(0)  # For reproducibility
SP_values = np.random.rand(len(L), len(month))
SE_values = SP_values + np.random.rand(len(L), len(month))

# Create the dataset
ds = xr.Dataset(
    {
        "SP": (["L", "month"], SP_values),
        "SE": (["L", "month"], SE_values)
    },
    coords={
        "L": L,
        "month": month,
        "S": np.arange(250),
        "model": np.arange(7),
        "M": np.arange(30)
    }
)
# slow
ds.plot.scatter(x='SP', y='SE')

ds = xr.Dataset(
    {
        "SP": (["L", "month"], SP_values),
        "SE": (["L", "month"], SE_values)
    },
    coords={
        "L": L,
        "month": month
    }
)
# fast
ds.plot.scatter(x='SP', y='SE')

MVCE confirmation

• Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray.
• Complete example — the example is self-contained, including all data and the text of any traceback.
• Verifiable example — the example copy & pastes into an IPython prompt or Binder notebook, returning the result.
• New issue — a search of GitHub Issues suggests this is not a duplicate.
• Recent environment — the issue occurs with the latest version of xarray and its dependencies.

Relevant log output

No response

Anything else we need to know?

For me, slow = 25 seconds and fast = instantaneous

Environment

INSTALLED VERSIONS

commit: None
python: 3.11.9 | packaged by conda-forge | (main, Apr 19 2024, 18:45:13) [Clang 16.0.6 ]
python-bits: 64
OS: Darwin
OS-release: 23.5.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: ('en_US', 'UTF-8')
libhdf5: 1.14.3
libnetcdf: 4.9.2

xarray: 2024.6.0
pandas: 2.2.2
numpy: 1.26.4
scipy: 1.13.1
netCDF4: 1.6.5
pydap: installed
h5netcdf: 1.3.0
h5py: 3.11.0
zarr: 2.18.2
cftime: 1.6.4
nc_time_axis: 1.4.1
iris: None
bottleneck: 1.3.8
dask: 2024.6.0
distributed: 2024.6.0
matplotlib: 3.8.4
cartopy: 0.23.0
seaborn: 0.13.2
numbagg: 0.8.1
fsspec: 2024.6.0
cupy: None
pint: 0.24
sparse: 0.15.4
flox: 0.9.8
numpy_groupies: 0.11.1
setuptools: 70.0.0
pip: 24.0
conda: None
pytest: 8.2.2
mypy: None
IPython: 8.17.2
sphinx: None

[headtr1ck]

Thanks for the report and the example.
This is really crazy, the slow example takes 2min on my machine while the fast one is basically instant.

After digging a bit, the problem seems to be, that xr.plot.dataset_plot._temp_dataarray broadcasts everything against everything creating a super large array of shape (12, 12, 250, 7, 30)...

@Illviljan do you have any idea? Probably removing unessesary coords before the broadcast might help?