Extracting positive elements with indices in Python-CodePudding

I have a matrix, N. I would like to extract positive elements with their indices. The desired output is attached.

import numpy as np
from numpy import nan

N=np.array([[            nan,  9.65984145e-08,             nan,
                    nan,             nan,             nan,
                    nan],
       [-6.35107511e-09,             nan,  4.88443157e-09,
                    nan,  1.46664353e-09,             nan,
                    nan],
       [            nan, -1.38022162e-08,             nan,
         1.37404097e-08,             nan,  6.18065590e-11,
                    nan],
       [            nan,             nan, -7.82145993e-10,
                    nan,             nan,             nan,
         7.82145990e-10],
       [            nan, -3.98567193e-09,             nan,
                    nan,             nan,  3.98567195e-09,
                    nan],
       [            nan,             nan, -5.47197743e-11,
                    nan, -3.66918072e-09,             nan,
         3.72390048e-09],
       [            nan,             nan,             nan,
        -3.39767791e-09,             nan, -1.04008985e-09,
                    nan]])

The desired output is

(0,1) - 9.65984145e-08 
(1,2) - 4.88443157e-09 
(1,4) - 1.46664353e-09 and so on

CodePudding user response：

I think you are looking for np.where():

idx = np.where(N > 0)
>>> list(zip(*idx, N[idx]))
[(0, 1, 9.65984145e-08),
 (1, 2, 4.88443157e-09),
 (1, 4, 1.46664353e-09),
 (2, 3, 1.37404097e-08),
 (2, 5, 6.1806559e-11),
 (3, 6, 7.8214599e-10),
 (4, 5, 3.98567195e-09),
 (5, 6, 3.72390048e-09)]

CodePudding user response：

If we need them separately in two separate arrays, we can get values by Boolean mask N >= 0 and then by transposing np.where array as:

values = N[N >= 0]
# [9.65984145e-08 4.88443157e-09 1.46664353e-09 1.37404097e-08 6.18065590e-11 7.82145990e-10 3.98567195e-09 3.72390048e-09]

indices = np.array(np.where(N >= 0)).T
# [[0 1]
#  [1 2]
#  [1 4]
#  [2 3]
#  [2 5]
#  [3 6]
#  [4 5]
#  [5 6]]