OK, I think I’ve managed to track the problem down a bit further:
the sort() method is failing for arrays pickled on another machine!
That is, it’s definitely not sorting the array, but changing to a very strange order (neither the way it started nor sorted).
Again, the array seems to otherwise behave fine (indeed, it even satisfies all(a==a1) for a pair that behave differently in this circumstance).
On 4/5/06, Andrew Jaffe <a.h.jaffe@…287…> wrote:
I’ve encountered a strange problem: I’ve been running some python code
on both a linux box and OS X, both with python 2.4.1 and the latest
numpy and matplotlib from svn.
I have found that when I transfer pickled numpy arrays from one machine
to the other (in either direction), the resulting data looks all right
(i.e., it is a numpy array of the correct type with the correct values
at the correct indices), but it seems to produce the wrong result in (at
least) one circumstance: matplotlib.hist() gives the completely wrong
picture (and set of bins).
This can be ameliorated by running the array through
but this seems like a complete kludge (and is only needed when you do
the transfer between machines).
I’ve attached a minimal code that exhibits the problem: try
on one machine, transfer the output file to another machine, and run
on another, and you should see a very strange result (and it should be
fixed if you set asarray=True).
a = numpy.linspace(-3,3,num=100) if write: f1 = file("a.cpkl", 'w') cPickle.dump(a, f1) f1.close
f1 = open("a.cpkl", 'r') a1 = cPickle.load(f1) f1.close() pylab.subplot(1,2,1) h = pylab.hist(a) if asarray: a1 = numpy.asarray(a1, dtype=numpy.float64
pylab.subplot(1,2,2) h1 = pylab.hist(a1) return a, a1