brandon-rhodes
commented
3 years ago Thanks for the question! This is a general NumPy issue, about which I am hoping to write up some documentation on within the next month or two. We can look at the error using simple NumPy addition and leaving Skyfield out:
import numpy as np
x = np.array([0, 10, 20, 30])
y = np.arange(0, 10, 1)
print(x + y)The result:
ValueError: operands could not be broadcast together with shapes (4,) (10,) Because NumPy addition wants natively to do pairwise addition (and the same is true with multiplication and division and everything else), it runs into a problem when one array is a different length than the other.
The solution is to move the two lengths (here, 4 and 10) on to two different dimensions. NumPy is always willing to broadcast an operation when either of the lengths is 1. So here’s the solution:
import numpy as np
x = np.array([0, 10, 20, 30])
y = np.arange(0, 10, 1)
x = x[..., np.newaxis]
y = y[np.newaxis, ...]
print(x + y)By keeping the 4 in the first axis and moving the 10 to the second axis, there's no conflict, and we get out a grid of 40×10 results:
[[ 0 1 2 3 4 5 6 7 8 9]
[10 11 12 13 14 15 16 17 18 19]
[20 21 22 23 24 25 26 27 28 29]
[30 31 32 33 34 35 36 37 38 39]]My advice would be to try the same trick with the arrays you are passing in for the time and the longitude. Fingers crossed, there's a chance that Skyfield — which is careful pretty much to only pay attention to the first axis of an array, and ignore its deeper structure — will go ahead and compute n×m positions for you.
But maybe it won't and improvements will have to be made; try out the idea and see what happens!
My guess is that keeping time as the first dimension and pushing longitude to the second dimension will work best, but please try both ways and let us know what happens. Thanks!
randomwalkingg
I am new to Skyfield and have the following question.
I got the following error: ValueError: operands could not be broadcast together with remapped shapes [original->remapped]: (3,3,10)->(3,10,3) (3,4)->(4,3).
It is OK if do
g = sites.at(time[0]). Does it mean that I cannot propagate "sites" over a time interval? And is there a way to solve my challenge without looping?