Wrong output in take function #9293
Comments
|
You are working on the array in place in those two modes, while the raise mode works on a copy (I guess as a trick to be consistent on error, even though there are other corner cases where we can't guarantee this too). I think this issue is open elsewhere, don't know what the discussion result was there. If you don't want it to be in-place (which will cause this), you need to work on a copy anyway, so using the I suppose we could add an overlap guard here, or so, since we started doing that for ufuncs, and a bit other things now. |
ahaldane
added a commit
to ahaldane/numpy
that referenced
this issue
Mar 24, 2019
seberg
pushed a commit
that referenced
this issue
Mar 27, 2019
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I encountered an error when i use the numpy.take function.
test : 2D array (20000x128) with double values
P: integer vector 128 values in the range of [0...127] none are the same.
if i execute the following code everything works fine and as intended, no error.
np.take(test, P, axis=1, mode='raise', out=test)if i execute
np.take(test, P, axis=1, mode='wrap', out=test)my output should be the same as in the other code, since i do not have any out-of-bounds indices. However, the output is different. Values are scrambled and appear multiple times although they are all different.
if executed as
test = np.take(test, P, axis=1, mode='wrap')the output is correct.same for: mode = 'clip'
numpy version 1.12.1
python version 2.7.13
The text was updated successfully, but these errors were encountered: