[SciPy-user] hyp1f1 -- Confluent hypergeometric function (1F1)

[Nils Wagner]

Colin Gillespie wrote:
> Dear All,
>
> I'm using the confluent hypergeometric function and I'm running into 
> some problems.
>
>   
>>>> from scipy.special import hyp1f1
>>>> hyp1f1(1,1,10)
>>>>         
> 22026.465794806718 #Same as maple and Abramowitz and Stegun
>   
>>>> hyp1f1(3,4,-6)
>>>>         
> 0.027777777777777769   #Maple gives0.0260564..
>
> There seems to be some sort of rounding error. Can I request more digits (or is this a bug)?
>
> Thanks
>
> Colin
>
>
>   
It seems to be a bug. I found a Matlab implementation which reproduces
the Maple result.
http://ceta.mit.edu/comp_spec_func/mchgm.m

Nils