\lib\site-packages\skyfield\functions.py:77: RuntimeWarning

Hello Brandon,

Today I was working with the Skyfiled, and I’ve got the following warning. I believe you should handle the exception.

e = ephemeris[399]
obj = ephemeris[399]
coordinates = e.at(t).observe(obj).radec()

.\lib\site-packages\skyfield\functions.py:77: RuntimeWarning: invalid value encountered in double_scalars theta = arcsin(z / r)

You might be asking why I am calculating geocentric astrometric coordinates of earth !!

Actually, I do not, It was happened by mistake. Then I got curious and I checked geocentric astrometric position as below:

e = ephemeris[399]
obj = ephemeris[399]
positions = e.at(t).observe(obj).position.au

And the result was successful [0.0, 0.0, 0.0]. As logically, we can guess.

Finally, I guess you should handle the exception somewhere in your code to get off that warning in the previous code and get a logical result. (Maybe it is a division by zero exception).

Best,

[brandon-rhodes]

I guess you should handle the exception somewhere in your code to get off that warning

The message, it turns out, is a warning and not an exception. There are two big differences. First: the program keeps going, as can be confirmed with a print.

from skyfield.api import load
ts = load.timescale()
t = ts.now()
ephemeris = load('de421.bsp')
e = ephemeris[399]
obj = ephemeris[399]
coordinates = e.at(t).observe(obj).radec()
print(coordinates)

This results in:

/home/brandon/skyfield/skyfield/functions.py:77: RuntimeWarning: invalid value encountered in double_scalars
  theta = arcsin(z / r)
(<Angle 00h 00m 00.00s>, <Angle nan>, <Distance 0.0 au>)

Since the print() runs, the program survived and did not encounter an exception.

Second: because this is not an exception, it cannot be caught. If you will try putting a try…except around the statement yourself, I think you will be able to confirm that you cannot catch the warning; it is transparent to the Skyfield code.

There is probably a way to ask NumPy to promote warnings to exceptions, but in general libraries like Skyfield are not supposed to set global NumPy options for their own purposes; instead, end-user code is supposed to be in full control.

My understanding, then, is that there is nothing Skyfield can do that will improve the situation? Please run your own checks and see if I am correct. If I am wrong, then simply provide the sample code that you think Skyfield should run that handles this case differently, and we can try it out. Thanks!

[reza-ghazi]

Of course. Thank you. Until then.