Multiplying data arrays of same dimensions fails

[e5k]

What is your issue?

Hi all,

I am trying to multiply two data arrays of same dimensions:

print(data1)
<xarray.DataArray 'var' (lat: 2160, lon: 4320)>
[9331200 values with dtype=int8]
Coordinates:
  * lon      (lon) float64 -180.0 -179.9 -179.8 -179.7 ... 179.8 179.9 180.0
  * lat      (lat) float64 89.96 89.88 89.79 89.71 ... -89.79 -89.88 -89.96

print(data2)
<xarray.DataArray 'var' (lat: 2160, lon: 4320)>
[9331200 values with dtype=float32]
Coordinates:
  * lon      (lon) float64 -180.0 -179.9 -179.8 -179.7 ... 179.8 179.9 180.0
  * lat      (lat) float64 89.96 89.88 89.79 89.71 ... -89.79 -89.87 -89.96

data1 * data2 returns this error:

ValueError: Cannot apply_along_axis when any iteration dimensions are 0

Note that following this thread, I made sure to have consistent dimensions and re-indexed both data arrays.

Since both arrays have different dtype, I have tried data1.astype(np.float64) * data2, but that returned the same error.

On the other hand, this returned an empty array:

data3 = data1.astype(np.float64) * data2.astype(np.float64)

print(data3)
<xarray.DataArray 'var' (lat: 0, lon: 0)>
array([], shape=(0, 0), dtype=float64)
Coordinates:
  * lon      (lon) float64 
  * lat      (lat) float64 

The only way I found to achieve this multiplication was to get the underlying np data:

data3 = data1.data * data2.data

Although this works for my need, I am still curious to understand why the pure xarray method fails. Can anyone inform me or point me towards a part of the documentation I might have missed?

Thanks loads

[welcome[bot]]

Thanks for opening your first issue here at xarray! Be sure to follow the issue template!
If you have an idea for a solution, we would really welcome a Pull Request with proposed changes.
See the Contributing Guide for more.
It may take us a while to respond here, but we really value your contribution. Contributors like you help make xarray better.
Thank you!

[andersy005]

@e5k,

a minor discrepancy in the coordinate values might exist, even though the dimensions and the coordinate labels are the same for both DataArray objects. this might happen due to a slight difference in the floating point precision for the latitude or longitude values in data1 and data2. you might want to inspect the coordinate values of both arrays and see if there is any such discrepancy.

import xarray as xr 

xr.testing.assert_equal(data1.lon, data2.lon)
xr.testing.assert_equal(data1.lat, data2.lat)

or

xr.testing.assert_allclose(data1.lon, data2.lon)
xr.testing.assert_allclose(data1.lat, data2.lat)

The only way I found to achieve this multiplication was to get the underlying np data:

data3 = data1.data * data2.data

Although this works for my need, I am still curious to understand why the pure xarray method fails. Can anyone inform me > or point me towards a part of the documentation I might have missed?

this worked because numpy is less strict about matching coordinates compared to xarray. it simply multiplies the arrays based on their shape without checking if the coordinates match (or performing any coordinate alignment)

[dcherian]

data1 * data2 returns this error:
ValueError: Cannot apply_along_axis when any iteration dimensions are 0

This one is a mystery to me. Can you create a minimal example?

[e5k]

@andersy005 - you were right! Sorry for not checking that, thanks for your help.

@dcherian - in case you are still interested: data1 data2

Thanks both!

[Illviljan]

ValueError: Cannot apply_along_axis when any iteration dimensions are 0

It is a very cryptic error, would been nice if it could've been made clearer.