New submission from Bryan <[email protected]>:

To reproduce:

1) Start pypy (results are usually more nondeterministic with ipython rather 
than straight 
pypy, but it doesn't matter which)
2) Run the following:
>>>> import numpy as np
>>>> np.array([]).reshape(0,2).sum(0)

Here are some of the results I've gotten:
array([  2.68156159e+154,  -2.31584192e+077])
array([ 0.,  0.]) # which is correct
array([  1.72723371e-77,   1.72723371e-77])
array([  1.28822975e-231,   1.28822975e-231])

and so forth.

It looks like some kind of overflow problem. With the regular pypy repl, you 
usually get long 
sequences of the same answer, though restarting the repl consistently produces 
different 
answers.

Note that Python + main numpy produces the correct result of array([ 0., 0.]) 
consistently.

I apologize if this is the wrong place for this bug report. Let me know if 
there's somewhere 
more appropriate to submit it.

I'm using the pypy fork of numpy 1.8.0.dev-6b8df83 and pypy 2.2.1 on OS X 10.9. 
Let me know if 
you'd like any other information.

----------
messages: 6394
nosy: pypy-issue, qiemem
priority: bug
status: unread
title: numpy: summing over an empty, mutli-dimensional array produces 
incorrect, nondeterministic results

________________________________________
PyPy bug tracker <[email protected]>
<https://bugs.pypy.org/issue1650>
________________________________________
_______________________________________________
pypy-issue mailing list
[email protected]
https://mail.python.org/mailman/listinfo/pypy-issue

Reply via email to