

In [17]:

    
import numpy as np

Matrix Indexing

Col Vec vs. Row Vec

Vectors are usually 1 dim.

To access an attribute/element of a vector as $\vec v_i$:

v[i]

When having a stacked vector in matrix, to access the $i$-th vector stacked as col vectors:

mat[:, i]

To access the $i$-th vector stacked as row vectors:

mat[i, :]
# or simply
mat[i]

Initialize an empty matrix with given dim using reshape.



In [19]:

    
dim = 3
X = np.array([]).reshape(dim, 0)
X









    Out[19]:





array([], shape=(3, 0), dtype=float64)

Append col-vec to matrix along col.



In [51]:

    
vec = [1, 2, 3]
Y = np.c_[X, vec]
Y









    Out[51]:





array([[ 1.],
       [ 2.],
       [ 3.]])



In [52]:

    
Y = np.c_[Y, vec]
Y









    Out[52]:





array([[ 1.,  1.],
       [ 2.,  2.],
       [ 3.,  3.]])

Append a matrix along col, differs in dim of axis=1

Convert vec to mat by expand_dims



In [53]:

    
mat = np.expand_dims(vec, axis=1)
mat









    Out[53]:





array([[1],
       [2],
       [3]])



In [54]:

    
np.hstack((Y, mat))









    Out[54]:





array([[ 1.,  1.,  1.],
       [ 2.,  2.,  2.],
       [ 3.,  3.,  3.]])



In [ ]: