[SciPy-user] weird behaviour in scipy.random seed

Giovanni Samaey Giovanni.Samaey@cs.kuleuven...
Tue Apr 24 07:35:10 CDT 2007


Hi,

I am seeing the following weird behaviour when seeding the random number 
generator in scipy.random.
The basic idea is that it gives an error when I take the seed out of a 
scipy array of integers, but works fine
when I cast them to standard python integers.  (See code below.)

Should I file a ticket for this?  Does anyone have the same?

Giovanni

from scipy import random as R
a=R.RandomState(seed=0)  # this is OK
a=R.RandomState(seed=0.)  # this gives an error, which is OK
ValueError: object of too small depth for desired array

Now I try the following

x = scipy.arange(500)
x[0].dtype             # this says 'int64'
a=R.RandomState(seed=x[0])  # this again give an error, which is not OK
ValueError: object of too small depth for desired array

Casting to int, resolves the problem:

a=R.RandomState(seed=int(x[0]))  # this does not give an error.



More information about the SciPy-user mailing list