How to append an array to an empty matrix from for loop by using python numpy or list?-CodePudding

I would like to run this code in order to append an array from each iteration to an empty matrix.

    import numpy as np
    Q = np.array([[10,20,3]])
    storage_Q = np.empty((6,3), dtype = object)
    TotalK = 6 

    for k in range(TotalK):
    Q = Q   [[2,0,1]]
    storage_Q = np.append(Q,Q, axis = 0)

    storage_Q

    I got a result:

    array([[22, 20,  9],
          [22, 20,  9]])

####

What I expect the output is:
####
    storage_Q_ = [[12,20,4], 
                 [14,20,5],
                 [16,20,6], 
                 [18,20,7],
                 [20,20,8],
                 [22,20,9]]

Can anyone suggest me how to revise the code, please? Thank you in advance :)

CodePudding user response：

#input
Q = np.array([[10,20,3]])
increment = np.array([[2,0,1]])
TotalK = 6 

increment_matrix = increment*np.arange(1, TotalK 1).reshape(TotalK,1)
storage_Q = increment_matrix Q

#output
storage_Q

array([[12, 20,  4],
       [14, 20,  5],
       [16, 20,  6],
       [18, 20,  7],
       [20, 20,  8],
       [22, 20,  9]])

Let me know if you understand the procedure or if you have any doubt.

CodePudding user response：

Look at your code step by step:

In [281]:    Q = np.array([[10,20,3]])
     ...:    storage_Q = np.empty((6,3), dtype = object)
     ...:    TotalK = 6

In [282]: storage_Q
Out[282]: 
array([[None, None, None],
       [None, None, None],
       [None, None, None],
       [None, None, None],
       [None, None, None],
       [None, None, None]], dtype=object)

Now do on loop step:

In [283]: k=0    
In [284]: Q = Q   [[2,0,1]]
     ...: storage_Q = np.append(Q,Q, axis = 0)    
In [285]: Q
Out[285]: array([[12, 20,  4]])
In [286]: storage_Q
Out[286]: 
array([[12, 20,  4],
       [12, 20,  4]])

You changed Q, and then joined it to itself with np.append, and assigned that to storage_Q. You threw away the initial definition (the object dtype).

Next loop will be more of the same, a new Q value, and new storage_Q.

Replace that with a list append, which you should understand better (?)

In [287]: Q = np.array([[10,20,3]])
     ...: storage_Q = []
In [288]: for k in range(6):
     ...:     Q = Q np.array([[2,0,1]])
     ...:     storage_Q.append(Q)
     ...:     

In [289]: storage_Q
Out[289]: 
[array([[12, 20,  4]]),
 array([[14, 20,  5]]),
 array([[16, 20,  6]]),
 array([[18, 20,  7]]),
 array([[20, 20,  8]]),
 array([[22, 20,  9]])]

And combine them:

In [290]: np.vstack(storage_Q)
Out[290]: 
array([[12, 20,  4],
       [14, 20,  5],
       [16, 20,  6],
       [18, 20,  7],
       [20, 20,  8],
       [22, 20,  9]])

Yes, it is possible to do this with a whole-array method, but I suspect you need more practice with basic Python loops - and debugging.

Here's a loop using the (6,3) initial array:

In [298]: Q = np.array([10,20,3])
     ...: storage_Q = np.zeros((6,3),int)  # np.empty ok here
     ...: for i in range(6):
     ...:     Q = Q [2,0,1]
     ...:     storage_Q[i,:] = Q   # assign to a row, not replace
     ...:     

In [299]: storage_Q
Out[299]: 
array([[12, 20,  4],
       [14, 20,  5],
       [16, 20,  6],
       [18, 20,  7],
       [20, 20,  8],
       [22, 20,  9]])