[FIXED] Creating custom output based on a matrix in Python

Issue

Problem Description

I have a NumPy matrix M generated as follows: M = np.arange(25).reshape(-1, 5) * 10. I'm trying to create custom output lists based on this matrix for different values of i. The desired output for various values of i is as follows:

• For i=5, the output should be [10, 20, 30, 40, 70, 80, 90, 130, 140, 190].
• For i=4, the output should be [10, 20, 30, 0, 70, 80, 0, 130, 0, 0].
• For i=3, the output should be [10, 20, 0, 0, 70, 0, 0, 0, 0, 0].
• For i=2, the output should be [10, 0, 0, 0, 0, 0, 0, 0, 0, 0].

I'm struggling to find an efficient way to achieve this. Can someone please help me with a Python code snippet that can generate these custom output lists based on the matrix M and the specified values of i? Thank you!

M.ravel()[np.flatnonzero(np.triu(M, k=1))] for i=5

Solution

Make the array:

In [134]: M=np.arange(25).reshape(5,5)*10

and the triu indices:

In [135]: idx = np.triu_indices(5,k=1);idx
Out[135]: (array([0, 0, 0, 0, 1, 1, 1, 2, 2, 3]), array([1, 2, 3, 4, 2, 3, 4, 3, 4, 4]))

for i=5:

In [136]: M[idx]
Out[136]: array([ 10,  20,  30,  40,  70,  80,  90, 130, 140, 190])

zero out a column, and repeat:

In [139]: M[:,-1:]=0; M
Out[139]: 
array([[  0,  10,  20,  30,   0],
       [ 50,  60,  70,  80,   0],
       [100, 110, 120, 130,   0],
       [150, 160, 170, 180,   0],
       [200, 210, 220, 230,   0]])

In [140]: M[idx]
Out[140]: array([ 10,  20,  30,   0,  70,  80,   0, 130,   0,   0])

and so on:

In [141]: M[:,-2:]=0; M[idx]
Out[141]: array([10, 20,  0,  0, 70,  0,  0,  0,  0,  0])

In [142]: M[:,-3:]=0; M[idx]
Out[142]: array([10,  0,  0,  0,  0,  0,  0,  0,  0,  0])

Answered By - hpaulj