Matrix with given numbers in random places in python/numpy

3
votes

I have an NxN matrix filled with zeros. Now I want to add to the matrix, say, n ones and m twos to random places. I.e. I want to create a matrix where there is some fixed amount of a given number at random places and possibly a fixed amount of some other given number in random places. How do I do this?

In Matlab I would do this by making a random permutation of the matrix indices with randperm() and then filling the n first indices given by randperm of the matrix with ones and m next with twos.

3
pythonrandommatrixnumpy

3 Answers

4
votes

You can use numpy.random.shuffle to randomly permute an array in-place.

>>> import numpy as np
>>> X = np.zeros(N * N)
>>> X[:n] = 1
>>> X[n:n+m] = 2
>>> np.random.shuffle(X)
>>> X = X.reshape((N, N))
3
votes

Would numpy.random.permutation be what you are looking for?

You can do something like this:

In [9]: a=numpy.zeros(100)

In [10]: p=numpy.random.permutation(100)

In [11]: a[p[:10]]=1

In [12]: a[p[10:20]]=2

In [13]: a.reshape(10,10)
Out[13]: 
array([[ 0.,  1.,  0.,  0.,  0.,  2.,  0.,  1.,  0.,  0.],
       [ 0.,  0.,  1.,  0.,  0.,  0.,  0.,  0.,  2.,  0.],
       [ 0.,  2.,  0.,  0.,  0.,  0.,  2.,  0.,  0.,  1.],
       [ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  2.,  0.,  2.,  1.,  1.,  0.],
       [ 0.,  0.,  0.,  0.,  1.,  0.,  2.,  0.,  0.,  0.],
       [ 0.,  2.,  0.,  2.,  0.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  0.,  0.,  0.,  1.,  0.,  0.],
       [ 1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  2.,  0.,  0.,  0.,  1.,  0.]])

Here we create a random permutation, then set the first 10 indices taken from the permutation in a to 1, then the next 10 indices to 2.

1
votes

To generate the indices of the elements for where to add ones and twos, what about this?

# assuming N, n and m exist.
In [1]: import random
In [3]: indices = [(m, n) for m in range(N) for n in range(N)]
In [4]: random_indices = random.sample(indices, n + m)
In [5]: ones = random_indices[:n]
In [6]: twos = random_indices[n:]

Corrected as commented by Petr Viktorin in order not to have overlapping indexes in ones and twos.

An alternate way to generate the indices:

In [7]: import itertools
In [8]: indices = list(itertools.product(range(N), range(N)))