What is the best way to perform an anti-transpose in python?
Question:
Lets say I have an array
a = np.arange(16).reshape((4,4))
0 1 2 3
4 5 6 7
8 9 10 11
12 13 14 15
But I want
15 11 7 3
14 10 6 2
13 9 5 1
12 8 4 0
which is a flip across the secondary diagonal, or a kind of anti-transpose.
How can I do this in numpy?
Answers:
One could do one of the following:
rot90(a,2).T
rot90(flipud(a),1)
rot90(fliplr(a), -1)
or as hpaulj suggested in the comments (thanks hpaulj)
a[::-1,::-1].T
Here are the speed rankings as ratios of the slowest method after anti-transposing 1000 random 10000×10000 arrays.
- 63.5% –
a[::-1,::-1].T
- 85.6% –
rot90(a,2).T
- 97.8% –
rot90(flipud(a),1)
- 100% –
rot90(fliplr(a),-1)
Here’s another to throw into the mix.
a.ravel('F')[::-1].reshape(a.shape)
Try it in this manner,
np=np[::-1] #reverse the array
a = np.arange(16).reshape((4,4))
np.flip(a).T
From the np.flip documentation & @hpaulj ‘s comment:
flip(m) corresponds to m[::-1,::-1,...,::-1] with ::-1 at all positions.
Lets say I have an array
a = np.arange(16).reshape((4,4))
0 1 2 3
4 5 6 7
8 9 10 11
12 13 14 15
But I want
15 11 7 3
14 10 6 2
13 9 5 1
12 8 4 0
which is a flip across the secondary diagonal, or a kind of anti-transpose.
How can I do this in numpy?
One could do one of the following:
rot90(a,2).T
rot90(flipud(a),1)
rot90(fliplr(a), -1)
or as hpaulj suggested in the comments (thanks hpaulj)
a[::-1,::-1].T
Here are the speed rankings as ratios of the slowest method after anti-transposing 1000 random 10000×10000 arrays.
- 63.5% –
a[::-1,::-1].T - 85.6% –
rot90(a,2).T - 97.8% –
rot90(flipud(a),1) - 100% –
rot90(fliplr(a),-1)
Here’s another to throw into the mix.
a.ravel('F')[::-1].reshape(a.shape)
Try it in this manner,
np=np[::-1] #reverse the array
a = np.arange(16).reshape((4,4))
np.flip(a).T
From the np.flip documentation & @hpaulj ‘s comment:
flip(m) corresponds to m[::-1,::-1,...,::-1] with ::-1 at all positions.