Question

I’m trying to change non-adjacent values of an array simultaneously, and it doesn’t seem to be working. Trying the example below, the code does not break, but the values do not change. I have run into the same problem using recarrays. Has anyone run into this before? Any thoughts? For example:

import numpy as np

simulants = np.zeros([10, 5])

simulants[:, 4] =np.array([5,25, 10, 5, 45, 30, 60, 1, 80, 80])

simulants[simulants[:,4]==5][:,0]=np.array([1, 2])

Answer 1

The order of indexing can matter. Try:

simulants[:,0][simulants[:,4]==5]=np.array([1,2])

The reason why order matters is because simulants[:,0] is a view, while
simulants[simulants[:,4]==5] is a (sub)copy of the array. (Modifying a view of an array modifies the underlying array. Modifying a copy does not.)

simulants[:,0] is a view because it uses basic slicing.
simulants[simulants[:,4]==5] is a copy because it uses a bool ndarray which triggers advanced indexing.

---

Edit: As @eryksun points out in the comments, the assignment could be better handled with a single index:

simulants[simulants[:,4]==5, 0] = np.array([1,2]