How do I use a generator to initialize a numpy array?

Question

Code below:

def computerCost(x,y,theta):
    m = len(y)
    J = np.sum((np.dot(x,theta) - y)**2) /(2*m)
    return J

m = 100
x = np.linspace(-5,10,m)
y = np.linspace(1,100,m)
x, y = x.reshape(m,1), y.reshape(m,1)
theta_0 = np.linspace(-10,10,100)
theta_1 = np.linspace(-1,4,100)
X,Y = np.meshgrid(theta_0,theta_1)

###### Here I want to initialize a numpy array with generator.
J_vals = np.array(computerCost(x,y,np.array([a,b])) for a,b in zip(np.ravel(X), np.ravel(Y)) )

print('out:',J_vals)

Running this code in Python 3.5 gives:

out:<generator object <genexpr> at 0x0000028ACF28B258>

The console prints that J_vals is a generator. Is there some way to change the generator into a np.ndarrray?

Can you try to explain what you are trying to solve? Maybe you can vectorize the operation, removing the need for zip() and fromiter(). — Nils Werner
– Nils Werner, Commented Jan 11, 2018 at 8:49
Inside your function you are doing np.dot(x, theta) where x is an array of shape (100, 1) and theta is of shape (2,). This will not work. Can you try to explain the math you are trying to do here? — Nils Werner
– Nils Werner, Commented Jan 11, 2018 at 8:57
@NilsWerner You are right. I just want to use Least squares to draw a line model to fitting my data. And the computerCost function is to computer cost in one theta(matrix:[theta_0,theta_1]) — Rosand Liu
– Rosand Liu, Commented Jan 11, 2018 at 9:06
See my answer on how to do it without the for loop and the zip. — Nils Werner
– Nils Werner, Commented Jan 11, 2018 at 9:14

Praveen · Accepted Answer · 2018-01-11 08:45:24Z

6

You're looking for np.fromiter.

Here's a simpler example to demonstrate how it works:

>>> a = [1, 2, 3]
>>> b = [4, 5, 6]
>>> np.fromiter((i + j for (i, j) in zip(a, b)), np.float)
array([ 5.,  7.,  9.])

Note you have to supply the data type as the second argument, and that the generator expression must be parenthesized since it's not the sole argument.

When I tried this with your sample code, I got an error saying shapes are not aligned... I'm guessing it's an issue with the dot product.

edited Jan 11, 2018 at 8:45

answered Jan 11, 2018 at 8:27

Praveen

7,2723 gold badges47 silver badges65 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Rosand Liu Over a year ago

Thank you very much. You are right, I should change np.arrray([a,b]) in J_vals = np.array(computerCost(x,y,np.array([a,b])) for a,b in zip(np.ravel(X), np.ravel(Y))) to np.array([[a,b]]) since the shape problem in dot product operation.

Nils Werner · Accepted Answer · 2018-01-11 09:13:20Z

2

You can use NumPy broadcasting to vectorize your operation, avoiding the need for Python loops altogether:

def computerCost(x, y, theta):
    return np.sum((x * theta - y) ** 2, axis=(0, 1)) / (2 * len(y))

m = 100
x = np.linspace(-5,10,m)[:, None, None]
y = np.linspace(1,100,m)[:, None, None]

theta_0 = np.linspace(-10,10,100)
theta_1 = np.linspace(-1,4,100)
X, Y = np.meshgrid(theta_0,theta_1)

XY = np.stack((X.ravel(), Y.ravel()))[None, :, :]

computerCost(x, y, XY)
# array([ 7442.62878788,  7340.86628993,  7240.13955518, ...,  1322.02086831,
#         1320.72740104,  1320.46969697])

answered Jan 11, 2018 at 9:13

Nils Werner

37.2k7 gold badges85 silver badges108 bronze badges

2 Comments

Rosand Liu Over a year ago

I want to know why you add [:,None,Noe] in x = np.linspace(-5,10,m)[:, None, None], And why this operation can work well. @NilsWerner

Nils Werner Over a year ago

You need to arrange your axes right for NumPy broadcasting to work.

Collectives™ on Stack Overflow

How do I use a generator to initialize a numpy array?

2 Answers 2

1 Comment

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related