[Numpy-discussion] behaviour of np.loadtxt

[Derek Homeier]

On Wed, 14 Sep 2011, Davide wrote:

> Dear list,
> 
> I'm encountering a problem with np.loadtxt.
> 
> Suppose i have a file containing three columns of data (and 10 rows), like:
> 
> 0.001 0.003 0.005
> 0.001 0.003 0.006
> 0.002 0.004 0.002
> 0.004 0.002 0.007
> 0.001 0.003 0.006
> 0.002 0.004 0.002
> 0.004 0.002 0.007
> 0.001 0.003 0.006
> 0.002 0.004 0.002
> 0.004 0.002 0.007
> 
> If i give:
>
>  len ( np.loadtxt( filename, unpack=True ) )
> 
> i get 3, since i have a three rows array. That's fine.
> 
> If instead the file has a single column, i.e.,
> 
> 0.004
> 0.002
> 0.007
> 0.004
> 0.002
> 0.007
> 0.004
> 0.002
> 0.007
> 0.004
> 
> the command
>
>  len ( np.loadtxt( filename, unpack=True ) )
> 
> returns 10, where i would expect it to return 1, to be consistent with
> the behaviour when there are multiple columns.
> 
> Is there a reason for why it is not like that?

Yes, the default for loadtxt is to automatically "squeeze" the output to the 
minimum no. of dimensions required, thus it returns a 1D-array in the latter 
case. If you are using numpy 1.6 or later, you can ensure to get a consistent 
shape by passing the "ndmin=2" option.

Cheers,
 			Derek
--
----------------------------------------------------------------
Derek Homeier          Centre de Recherche Astrophysique de Lyon
ENS Lyon                                      46, Allée d'Italie
69364 Lyon Cedex 07, France                  +33 1133 47272-8894
----------------------------------------------------------------