This looks like the difference between memmove and memcpy to me, but I
am not sure what the expected behavior of numpy should be. The first
shift behaves the way I expect, the second is surprising.
I know about numpy.roll. I was hoping for something faster, which this
would be if it worked.
In [1]: a = (np.random.random((10,10))*10).astype('u1')
In [2]: a
Out[2]:
array([[8, 0, 5, 4, 8, 2, 7, 8, 7, 6],
[6, 6, 3, 3, 9, 8, 0, 8, 9, 5],
[5, 0, 1, 1, 2, 5, 8, 2, 5, 3],
[9, 0, 0, 2, 8, 2, 0, 7, 7, 0],
[9, 8, 6, 9, 6, 3, 9, 4, 4, 5],
[2, 7, 6, 9, 3, 8, 9, 9, 6, 9],
[2, 8, 8, 4, 0, 3, 7, 6, 7, 6],
[2, 4, 9, 2, 4, 7, 3, 6, 7, 4],
[3, 2, 0, 7, 0, 7, 6, 6, 1, 6],
[2, 3, 8, 8, 9, 6, 7, 2, 5, 0]], dtype=uint8)
In [3]: a[:, :-1] = a[:, 1:]
In [4]: a
Out[4]:
array([[0, 5, 4, 8, 2, 7, 8, 7, 6, 6],
[6, 3, 3, 9, 8, 0, 8, 9, 5, 5],
[0, 1, 1, 2, 5, 8, 2, 5, 3, 3],
[0, 0, 2, 8, 2, 0, 7, 7, 0, 0],
[8, 6, 9, 6, 3, 9, 4, 4, 5, 5],
[7, 6, 9, 3, 8, 9, 9, 6, 9, 9],
[8, 8, 4, 0, 3, 7, 6, 7, 6, 6],
[4, 9, 2, 4, 7, 3, 6, 7, 4, 4],
[2, 0, 7, 0, 7, 6, 6, 1, 6, 6],
[3, 8, 8, 9, 6, 7, 2, 5, 0, 0]], dtype=uint8)
In [5]: a[:, 1:] = a[:, :-1]
In [6]: a
Out[6]:
array([[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[6, 6, 6, 6, 6, 6, 6, 6, 6, 6],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[8, 8, 8, 8, 8, 8, 8, 8, 8, 8],
[7, 7, 7, 7, 7, 7, 7, 7, 7, 7],
[8, 8, 8, 8, 8, 8, 8, 8, 8, 8],
[4, 4, 4, 4, 4, 4, 4, 4, 4, 4],
[2, 2, 2, 2, 2, 2, 2, 2, 2, 2],
[3, 3, 3, 3, 3, 3, 3, 3, 3, 3]], dtype=uint8)
In [7]: np.__version__
Out[7]: '1.3.0'
_______________________________________________
NumPy-Discussion mailing list
[email protected]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
