I am trying to replace a row in a 2d numpy array.
array2d = np.arange(20).reshape(4,5)
for i in range(0, 4, 1):
array2d[i] = array2d[i] / np.sum(array2d[i])
but I'm getting all 0s:
[[0 0 0 0 0]
[0 0 0 0 0]
[0 0 0 0 0]
[0 0 0 0 0]]
The expected result is:
[[0 0.1 0.2 0.3 0.4]
[0.14285714 0.17142857 0.2 0.22857143 0.25714286]
[0.16666667 0.18333333 0.2 0.21666667 0.23333333]
[0.17647059 0.18823529 0.2 0.21176471 0.22352941]]
The reason you are getting 0's is because the array's dtype is int but the division returns floats in range 0 to 1 and because you modify the rows in-place they are converted to integers (i.e. to 0 in your example). So to fix it use array2d = np.arange(20, dtype=float).reshape(4,5).
But there is no need for the for-loop:
array2d = np.arange(20).reshape(4,5)
array2d = array2d / np.sum(array2d, axis=1, keepdims=True)
Note that here I didn't specify the dtype of the array to be float, but the resulting array's dtype is float because on the second line we created a new array instead of modifying the first array in-place.
https://numpy.org/doc/stable/user/basics.indexing.html#assigning-values-to-indexed-arrays
