[Numpy-discussion] Is this expected behavior?

[Christopher Barker]

Russel Howe wrote:
> Oh, of course.  I can reverse it myself.  Thanks, I did not think of that.

note that you may need to make sure that your arrays are in C-order.

-Chris


> Russel
> 
> Warren Weckesser wrote:
>> Russel Howe wrote:
>>> Since they are iterators, is it possible to check for the second 
>>> condition and reverse both of them so the behavior I expect happens or 
>>> does this  break something else?
>>>   
>> You may already know this, but just in case...
>>
>> In the second case, you can accomplish the shift by using reversed slices:
>>
>> a[:, -1:0:-1] = a[:, -2::-1]
>>
>>
>> Warren
>>
>>> Russel
>>> Robert Kern wrote:
>>>   
>>>> On Mon, Jan 18, 2010 at 13:41, Russel Howe <russel at appliedminds.com> wrote:
>>>>     
>>>>> 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.
>>>>>       
>>>> memmove() and memcpy() are not used for these operations (and in
>>>> general, they can't be). Rather, iterators are created and looped over
>>>> to do the assignments. Because you are not making copies on the
>>>> right-hand-side, you are modifying the RHS as the iterators assign to
>>>> the LHS.
>>>>
>>>>     
>>>>> 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)
>>>>>       
>>>> The first one works because the RHS pointer is always one step ahead
>>>> of the LHS pointer, thus it always reads pristine data.
>>>>
>>>>     
>>>>> 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)
>>>>>       
>>>> The second one fails to work as you expect because the RHS pointer is
>>>> always one step behind the LHS pointer, thus it always reads the data
>>>> that just got modified in the previous step. The data you expected it
>>>> to read has already been wiped out.
>>>>
>>>>     
>>> _______________________________________________
>>> NumPy-Discussion mailing list
>>> NumPy-Discussion at scipy.org
>>> http://mail.scipy.org/mailman/listinfo/numpy-discussion
>>>   
>> _______________________________________________
>> NumPy-Discussion mailing list
>> NumPy-Discussion at scipy.org
>> http://mail.scipy.org/mailman/listinfo/numpy-discussion
> 
> _______________________________________________
> NumPy-Discussion mailing list
> NumPy-Discussion at scipy.org
> http://mail.scipy.org/mailman/listinfo/numpy-discussion


-- 
Christopher Barker, Ph.D.
Oceanographer

Emergency Response Division
NOAA/NOS/OR&R            (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115       (206) 526-6317   main reception

Chris.Barker at noaa.gov