Menu
I have a 4 dimensional numpy array of shape (N, N, Q, Q). So given a row and column index (i, j), mat[i,j] is a QxQ matrix. I want to reshape this array to shape (N*Q, N*Q) such that
array([[[[ 0, 1],
[ 2, 3]],
[[ 4, 5],
[ 6, 7]]],
[[[ 8, 9],
[10, 11]],
[[12, 13],
[14, 15]]]])
goes to
array([[ 0., 1., 4., 5.],
[ 2., 3., 6., 7.],
[ 8., 9., 12., 13.],
[ 10., 11., 14., 15.]])
You can see that mat[0,0] goes to new_mat[0:2, 0:2]. Currently mat.reshape(N*Q, N*Q) takes mat[0,0] to new_mat[0:4, 0] (which is what I do not want). How can I use reshape or rollaxis or something similar to reshape this array? I eventually want to plot it with imshow, am currently stuck. I figure it's easy to do, I just haven't yet figured it out.
491
reshape() is an inbuilt function in python to reshape the array. We can reshape into any shape using reshape function. This function gives a new shape to the array.
Use numpy. reshape() to return a view of the original array. Now suppose you want to create a 2-dimensional copy of the 1-dimensional NumPy array then use the copy() function along with the reshape() function.
Reshaping Does Not Make a Copy of an Array: Instead, the original array and the reshaped array reference the same underlying data.
Nevermind, I figured it out. np.swapaxes(1, 2) was the missing piece I needed.
The answer is just to do mat.swapaxes(1, 2).reshape(N*Q, N*Q).
Feel foolish for posting without attempting to figure it out myself for too long, but I'll leave it up so others can benefit from it.
107
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With
