import numpy as np

# activation function 
def sigma(z):
    return np.where(z>0, z,0) # ReLU
    

if __name__ == '__main__':

    xs = np.array([0,1]) # input
    N = xs.shape[0]
    print('N=',N)
    w = np.ones((N,N))   # weights
    b = np.zeros((N))    # biases
    
    print('w.shape = ',w.shape)
    print('b.shape = ',b.shape)
    

    # manipulate weight and biases
    b[1]=-0.5
    w[0,1] = 0

    print('w=\n',w)
    print('b=\n',b)

    # propagate inputs xs through the perceptron,
    ys = sigma(np.dot(w,xs) + b) 

    for x,y in zip(xs,ys):
        print('x=',x,'y=',y)