for i in range(n_iterations):
    # select a training example at random
    t = data[:, np.random.randint(0, n)].reshape(np.array([m, 1]))
    
    # find its Best Matching Unit
    bmu, bmu_idx = find_bmu(t, net, m)
    
    # decay the SOM parameters
    r = decay_radius(init_radius, i, time_constant)
    l = decay_learning_rate(init_learning_rate, i, n_iterations)
    
    # update weight vector to move closer to input
    # and move its neighbours in 2-D vector space closer
    
    for x in range(net.shape[0]):
        for y in range(net.shape[1]):
            w = net[x, y, :].reshape(m, 1)
            w_dist = np.sum((np.array([x, y]) - bmu_idx) ** 2)
            w_dist = np.sqrt(w_dist)
            
            if w_dist <= r:
                # calculate the degree of influence (based on the 2-D distance)
                influence = calculate_influence(w_dist, r)
                
                # new w = old w + (learning rate * influence * delta)
                # where delta = input vector (t) - old w
                new_w = w + (l * influence * (t - w))
                net[x, y, :] = new_w.reshape(1, 3)