how to remove these loops in python for sigmoid function

Question

def sigmoid(a):
    g = a
    for i in range(a.shape[0]):
        for j in range(a.shape[1]):
            if a[i][j] >= 0:
                z = np.exp(-a[i][j])
                g[i][j] = 1 / (1 + z)
            else:
                z = np.exp(a[i][j])
                g[i][j] = 1 / (1 + z)
    return g

how can I improve this code? these loops are taking too much time. I tried the following code

def sigmoid(a):
    if a > 0:
       z = np.exp(-a)
       return 1/(1+z)
    else:
       z = np.exp(a)
       return 1/(1+z)

But its not working for a 2D array or even 1D array, gives error on if statement.

if a > 0: ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Your second function seems to be working fine for me with a single input. Are you trying to pass an array as an argument? — Ishwar Venugopal, Feb 26 '21 at 11:00
For second `def sigmoid(a)` you;re giving array and comparing it with single number. You should try: `a[i] > 0` ... — yakhyo, Feb 26 '21 at 11:00

score 3 · Accepted Answer · answered Feb 26 '21 at 11:04

3

abs(a) gives the absolute value of each element in a numpy array, so you can simply use that:

def sigmoid(a):
    z = np.exp(-abs(a))
    return 1/(1+z)

answered Feb 26 '21 at 11:04

interjay

107,303
21
270
254

I suggest you accept this answer in that case. – rpoleski Feb 26 '21 at 12:15

how to remove these loops in python for sigmoid function

1 Answers1