LightCurve objects store differently in numpy arrays following version 1.2

Problem description

LightCurve objects explode into objects for each row in time/flux when being put into numpy arrays in version 1.2 and above. Before they would just be stored as one LightCurve object. Could this be to do with the changes implemented in #532?

Is there a way to modify my code so that I can store a LightCurve object in a numpy array in the same way as before v1.2? Or, is this a bug / unforeseen impact relating to recent changes to Lightkurve?

Example

import numpy as np
import lightkurve as lk

lc = lk.LightCurve(np.linspace(0, 1, 100), np.random.normal(1.0, 0.01, 100))
print(np.array([lc]))

This outputs:

[[<lightkurve.lightcurve.LightCurve object at 0x102cab490>
 <lightkurve.lightcurve.LightCurve object at 0x102cab690>
 <lightkurve.lightcurve.LightCurve object at 0x102cab890>
 ...
 <lightkurve.lightcurve.LightCurve object at 0x102cb9850>
 <lightkurve.lightcurve.LightCurve object at 0x102cb98d0>
 <lightkurve.lightcurve.LightCurve object at 0x102cb9950>]]

(ellipses added in as output is 100 lines long).

Expected behavior

Before version 1.2, this would just output the following:

[<lightkurve.lightcurve.LightCurve object at 0x7fda3f0b6470>]

Environment

• platform: MacOS 10.15 and CentOS Linux 7
• lightkurve version: 1.2 and 1.3
• installation method: pip

[barentsen]

Ha! That’s interesting! The behavior appears to have changed because we added the LightCurve.__len__ method in #532, which triggers np.array to treat a LightCurve as an iterable.

A work-around is to do the following:

ar = np.empty(1, dtype=object)
ar[0] = lc

@alexlyttle Would you be willing to provide a bit more background as to why you are interested in being able to have numpy arrays contain Lightcurve objects?

[barentsen]

Thanks @alexlyttle! We appreciate your feedback!! 🙏