CODEWITHSUNDEEP
pythonarraysnumpymatrix
Charlie Parker
Charlie Parker

Reputation: 5231

Dealing with N by 1 matrices in Numpy

Given a numpy array of size (n,) how do you transform it to a numpy array of size (n,1).

The reason is because I am trying to matrix multiply to numpy arrays of size (n,) and (,n) to get a (n,n) but when I do:

numpy.dot(a,b.T)

It says that you can't do it. I know as a fact that transposing a (n,) does nothing, so it would just be nice to change the (n,) and make them (n,1) and avoid this problem all together.

Upvotes: 4

Views: 2994

Answers (2)

ely
ely

Reputation: 77424

You can use None for dimensions that you want to be treated as degenerate.

a = np.asarray([1,2,3])
a[:]
a[:, None]

In [48]: a
Out[48]: array([1, 2, 3])

In [49]: a[:]
Out[49]: array([1, 2, 3])

In [50]: a[:, None]
Out[50]: 
array([[1],
       [2],
       [3]])

Upvotes: 4

CT Zhu
CT Zhu

Reputation: 54330

Use reshape (-1,1) to reshape (n,) to (n,1), see detail examples:

In [1]:

import numpy as np
A=np.random.random(10)
In [2]:

A.shape
Out[2]:
(10,)
In [3]:

A1=A.reshape(-1,1)
In [4]:

A1.shape
Out[4]:
(10, 1)
In [5]:

A.T
Out[5]:
array([ 0.6014423 ,  0.51400033,  0.95006413,  0.54321892,  0.2150995 ,
        0.09486603,  0.54560678,  0.58036358,  0.99914564,  0.09245124])
In [6]:

A1.T
Out[6]:
array([[ 0.6014423 ,  0.51400033,  0.95006413,  0.54321892,  0.2150995 ,
         0.09486603,  0.54560678,  0.58036358,  0.99914564,  0.09245124]])

Upvotes: 4

Related Questions