How to add None for the element in matrix-CodePudding

I would like to generate a matrix Z_comb which consists of element z with size(len(m),len(m)). The value of z is from 0 up to (m-1).

For example:

m = [m[0],[m[1],[m[2]] = [4,2,1].

For m[0] = 2, then z[0] = [0,1] and len(m[0]) = 2

For m[1] = 4, then z[1] = [0,1,2,3] and len(m[1]) = 4

For m[2] = 1, then z[2] = [0] and len(m[2]) = 1

What is expected are: For m[0] = 2, then z[0] = [0,1, None, None] and len(m[0]) = 4

For m[1] = 4, then z[1] = [0,1,2,3] and len(m[1]) = 4

For m[2] = 1, then z[2] = [0, None, None, None] and len(m[2]) = 4

Here the code that I made:

    import numpy as np
    m = np.array([2,4,1])
    Z_comb = np.array([np.arange(0,m[0]),np.arange(0,m[1]),np.arange(0,m[2])],dtype=object)

The printed result is

Z_comb = array([array([0, 1]), array([0, 1, 2, 3]), array([0])], dtype=object)

Expected result:

Z_comb = [[0, 1,None,None], [0, 1, 2,3], [0, None, None, None]]

Can anyone tell me what to modify for the code, please? Thank you in advance.

CodePudding user response：

You can use a list comprehension to generate the matrix Z_comb, where for each element in m, you create a list of integers from 0 to m[i]-1, and then append None to the end of the list until the length of the list is 4. Here's an example of how you can do this:

m = [2, 4, 1]
Z_comb = [[i if i < m[j] else None for i in range(4)] for j in range(len(m))]

This will generate the expected result:

[[0, 1, None, None], [0, 1, 2, 3], [0, None, None, None]]

You can then turn Z_comb into a numpy array if you wish.

This solution does generate what you said the expected result is, although it doesn't have size (len(m), len(m)), which you also mention it should at the start.

CodePudding user response：

numpy solution

You can use broadcasting:

# input
m = np.array([2,4,1])

N = m.max()
# 4
a = np.arange(N) 
# array([0, 1, 2, 3])

Z_comb = np.where(m[:,None] >= a, a, np.nan)

NB. the output contains nan in place of None as None doesn't really make sense in a numpy array (this would force the object dtype and prevent vectorization).

Output Z_comb:

array([[ 0.,  1.,  2., nan],
       [ 0.,  1.,  2.,  3.],
       [ 0.,  1., nan, nan]])

pure python solution

If you want a pure python solution:

N = max(m)
Z_comb = [list(range(x)) [None]*(N-x) for x in m]

Output:

[[0, 1, None, None],
 [0, 1, 2, 3],
 [0, None, None, None]]

CodePudding user response：

df = pd.DataFrame(array).reshape(row,col) df.append(newrow)