I'm trying to generate negative binomial random numbers where the n parameter
is non integer (<1 in fact). This should be possible, as n only needs to be 
real. However, I get the following message:

ValueError: n <= 0

I assume this is because some rounding is going on behind the scenes.

The reason I am doing this is because I'm trying to use the alternative
parameterisation of the NB, with parameters k,m for which p=k/(k+m). I noticed
that this parameterisation does not exist over in the scipy distributions 
either. Has anyone else run into this problem? The latter parameterisation
is far more common in ecology, because it is an overdispersed Poisson.

Thanks,
cf

_______________________________________________
Numpy-discussion mailing list
[email protected]
http://projects.scipy.org/mailman/listinfo/numpy-discussion

Reply via email to