[IPython-User] numerical inaccuracy on osx (notebook)

epi massimodisasha@gmail....
Mon Jun 24 13:09:21 CDT 2013


Matthias,

i removed "all" the ipython installation i found on my system (system python, brew python)

then i reinstalled ipython with pip  (pip install -U ipython)

which ipython
/usr/local/share/python/ipython

ipython --version
0.13.2


testing in the notebook  the lines :

import numpy as np
url='http://www.cdc.noaa.gov/Correlation/amon.us.long.data'
data = np.genfromtxt(url, skip_header=1, skip_footer=4)

data[:, 1:][0]

array([ 0.22 ,  0.153,  0.224,  0.143,  0.196,  0.218,  0.232,  0.209,
        0.276,  0.126,  0.136,  0.23 ])

from the url :   http://www.cdc.noaa.gov/Correlation/amon.us.long.data

the first row is :

1856    0.220    0.153    0.225    0.144    0.196    0.219    0.232    0.209    0.276    0.126    0.137    0.231


I think the problem is in the notebook + bumpy/pylab  can be possible ?
because in the notebook the pandas example work just fine (same results between notebook and shell/qtconsole)
or something related to my locale / system profiles / ipython config ?



i tried to open the qtconsole from inside the notebook "  %qtconsole magic "
and i got the same problem.

attached a screenshot [1]
on the left the ipython shell and qt console running from the bash command line
on the right the notebook and qt console running from the qtconsole magic

[1] http://geofemengineering.it/logs/test_np.png


Il giorno 24/giu/2013, alle ore 13:09, Matthias BUSSONNIER <bussonniermatthias@gmail.com> ha scritto:

> 
> Le 24 juin 2013 à 19:00, epi a écrit :
> 
>> Hi Matthias,
> 
> Hi Massimo, 
> 
>> 
>> i do not have EPD/ANACONDA installed.
>> i'm  running homebrew python with ipython installed from source (GIT master)
> 
> Does the same happened if you start qtconsole from the notebook with the %qtconsole magic ?
> I'm really astonished about how qtconsole  and notebook could differ.
> -- 
> Matthias
> 
> _______________________________________________
> IPython-User mailing list
> IPython-User@scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-user



More information about the IPython-User mailing list