Mailman 3 Strange numpy.argmax behavior on object arrays in numpy 1.0. - NumPy-Discussion

Strange numpy.argmax behavior on object arrays in numpy 1.0.

Tom Denniston

29 Oct 2006 29 Oct '06

2:53 p.m.

I recently upgraded to numpy 1.0 from 1.0b5. I noticed that numpy.argmaxbehavior is very strange on object arrays. See below: (Pdb) numpy.__version__ '1.0' (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object)) 0 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=int)) 1 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object), axis=0) 0 I would expect the argmax to behave the same on the dtype=int and dtype=object examples but it doesn't. Am I missing some subtelty or is this just a bug? 1.0 is the most recent version, right? --Tom ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Numpy-discussion mailing list Numpy-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Attachments:

attachment.htm (text/html — 704 bytes)

Show replies by date

Charles R Harris

29 Oct 29 Oct

4:05 p.m.

New subject: Strange numpy.argmax behavior on object arrays in numpy 1.0.

On 10/29/06, Tom Denniston wrote:

...

I recently upgraded to numpy 1.0 from 1.0b5. I noticed that numpy.argmaxbehavior is very strange on object arrays. See below:

(Pdb) numpy.__version__ '1.0' (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object)) 0 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=int)) 1 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object), axis=0) 0

I would expect the argmax to behave the same on the dtype=int and dtype=object examples but it doesn't. Am I missing some subtelty or is this just a bug? 1.0 is the most recent version, right?

Suppose In [22]: array([1,[2,3]], dtype=object) Out[22]: array([1, [2, 3]], dtype=object) How would you compare the elements? In [27]: 2 < [0,0] Out[27]: True In [28]: [0,0] > 2 Out[28]: True Compares memory locations? In [28]: [2] < [0,0] Out[28]: False Lexical ordering? I don't know how python interprets these things. That said, I suspect your example should behave better, but it might give strange results sometimes anyway. Chuck ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Numpy-discussion mailing list Numpy-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Tom Denniston

5:10 p.m.

New subject: Strange numpy.argmax behavior on object arrays in numpy 1.0.

Oh. My mistake. I thought I had an array of 2 objects which were ints. I actually had an array of one list of 2 ints. It works properly if I construct the array properly. In [14]: a = numpy.empty((2), dtype=object) In [15]: a[0:]=[2,3] In [16]: a Out[16]: array([2, 3], dtype=object) In [17]: a.shape Out[17]: (2,) In [18]: a[0] Out[18]: 2 In [19]: a[1] Out[19]: 3 In [20]: numpy.argmax(a) Out[20]: 1 On 10/29/06, Charles R Harris wrote:

...

On 10/29/06, Tom Denniston wrote:

...
I recently upgraded to numpy 1.0 from 1.0b5. I noticed that numpy.argmax behavior is very strange on object arrays. See below:

(Pdb) numpy.__version__ '1.0' (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object)) 0 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=int)) 1 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object), axis=0) 0

I would expect the argmax to behave the same on the dtype=int and dtype=object examples but it doesn't. Am I missing some subtelty or is this just a bug? 1.0 is the most recent version, right?

Suppose

In [22]: array([1,[2,3]], dtype=object) Out[22]: array([1, [2, 3]], dtype=object)

How would you compare the elements?

In [27]: 2 < [0,0] Out[27]: True

In [28]: [0,0] > 2 Out[28]: True

Compares memory locations?

In [28]: [2] < [0,0] Out[28]: False

Lexical ordering?

I don't know how python interprets these things. That said, I suspect your example should behave better, but it might give strange results sometimes anyway.

Chuck

------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

_______________________________________________ Numpy-discussion mailing list Numpy-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/numpy-discussion

------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Numpy-discussion mailing list Numpy-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Charles R Harris

6:17 p.m.

New subject: Strange numpy.argmax behavior on object arrays in numpy 1.0.

On 10/29/06, Tom Denniston wrote:

...

Oh. My mistake. I thought I had an array of 2 objects which were ints. I actually had an array of one list of 2 ints. It works properly if I construct the array properly.

I think there is actually a bug here: In [61]: sort(array([3,2], dtype=object)) Out[61]: array([2, 3], dtype=object) In [62]: argmax(array([2,3], dtype=object)) Out[62]: 0 See, the sort works fine. I suspect argmax is using the wrong comparison function. I was just pointing out that sometimes it is hard to know what is going on with objects. Chuck ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Numpy-discussion mailing list Numpy-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Tom Denniston

30 Oct 30 Oct

8:32 a.m.

New subject: Strange numpy.argmax behavior on object arrays in numpy 1.0.

Yes I definately agree about the latter! On 10/29/06, Charles R Harris wrote:

...

On 10/29/06, Tom Denniston wrote:

...
Oh. My mistake. I thought I had an array of 2 objects which were ints. I actually had an array of one list of 2 ints. It works properly if I construct the array properly.

I think there is actually a bug here:

In [61]: sort(array([3,2], dtype=object)) Out[61]: array([2, 3], dtype=object)

In [62]: argmax(array([2,3], dtype=object)) Out[62]: 0

See, the sort works fine. I suspect argmax is using the wrong comparison function. I was just pointing out that sometimes it is hard to know what is going on with objects.

Chuck

------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

_______________________________________________ Numpy-discussion mailing list Numpy-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/numpy-discussion

Travis Oliphant

9:19 p.m.

New subject: Strange numpy.argmax behavior on object arrays in numpy 1.0.

Tom Denniston wrote:

...

I recently upgraded to numpy 1.0 from 1.0b5. I noticed that numpy.argmax behavior is very strange on object arrays. See below:

(Pdb) numpy.__version__ '1.0' (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object)) 0 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=int)) 1 (Pdb) numpy.argmax(numpy.array([2, 3], dtype=object), axis=0) 0

I would expect the argmax to behave the same on the dtype=int and dtype=object examples but it doesn't. Am I missing some subtelty or is this just a bug? 1.0 is the most recent version, right? This is a bug. I've fixed in in SVN. Thanks for the test.

-Travis ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642

6384

Age (days ago)

6386

Last active (days ago)

List overview

Download

5 comments

3 participants

participants (3)

Charles R Harris
Tom Denniston
Travis Oliphant

Strange numpy.argmax behavior on object arrays in numpy 1.0.

Tom Denniston

Charles R Harris

Tom Denniston

Charles R Harris

Tom Denniston

Travis Oliphant

tags

participants (3)