CODEWITHSUNDEEP
pythonarraysnumpymultidimensional-arraynumpy-ndarray
MrSir
MrSir

Reputation: 596

Numpy - create a multidimensional array of a copy of an array

I want to generate a an array of ordered numbers and then multiply it into another array :

[ [0,1,2,3,4,5] [0,1,2,3,4,5] [0,1,2,3,4,5] ... [0,1,2,3,4,5] ]

I can generate the first [0,1,2,3,4,5] with nums = np.arange(0, 6) but then if I multiply by a number inside a list it just increases the values = [nums* 3] = [0,3,6,9,12,15]. How can I do this ?

Upvotes: 2

Views: 2685

Answers (3)

kmario23
kmario23

Reputation: 61325

BTW, why not simply use np.array() as in:

In [147]: nums = np.arange(6)

In [148]: nums
Out[148]: array([0, 1, 2, 3, 4, 5])

In [149]: [nums] * 5
Out[149]: 
[array([0, 1, 2, 3, 4, 5]),
 array([0, 1, 2, 3, 4, 5]),
 array([0, 1, 2, 3, 4, 5]),
 array([0, 1, 2, 3, 4, 5]),
 array([0, 1, 2, 3, 4, 5])]

In [150]: np.array([nums] * 5)
Out[150]: 
array([[0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5]])

In one-line:

In [151]: np.array([np.arange(6)] * 5)
Out[151]: 
array([[0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5]])

Upvotes: 3

Haleemur Ali
Haleemur Ali

Reputation: 28233

you can't multiply a numpy array with a scalar and expect the same behaviour as multiplying a python list (or string) with a scalar.

for numpy, the multiplication operator will broadcast the multiplication over all the array elements:

i.e. 

np.array([1,2,3]) * 2 == np.array([1*2, 2*2, 3*2) == np.array([2,4,6])

instead, you can use a list comprehension

np.array([np.arange(6) for _ in range(4)])
array([[0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5]])

or generate a list of lists through multiplication and then convert to numpy array & reshape:

np.array([list(range(6))*4]).reshape(4,6)
array([[0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5]])

or, generate an array of shape (1,6) and repeat along the first axix:

np.repeat(np.arange(6).reshape(1,6), repeats=4, axis=0)
# produces the same output as the example outputs above.

Upvotes: 1

sacuL
sacuL

Reputation: 51335

Using numpy methods (numpy.repeat and numpy.expand_dims):

np.repeat(np.expand_dims(np.arange(0,6), axis=0), repeats=5, axis=0)

array([[0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5],
       [0, 1, 2, 3, 4, 5]])

Or, more simply: 

np.repeat([np.arange(0,6)],repeats=5, axis=0)

The first method is useful if you were trying to expand a pre-existing one dimensional array. If you are trying to create your array from the start, the second method is more straightforward.

Upvotes: 7

Related Questions