[seqfan] Re: Question on 2F1 from Karol A. Penson

Richard J. Mathar mathar at mpia-hd.mpg.de
Wed Jul 10 12:02:17 CEST 2013


In answer to http://list.seqfan.eu/pipermail/seqfan/2013-July/011394.html:

With the standard integral formula
(see for example equation (9.68) in http://arxiv.org/abs/1207.5845 )
2F1(1,1/2+n; 3+n | -3)
= Gamma(3+n)/Gamma(1/2+n)/Gamma(3+n-1/2-n) *M
where
M = integral_{t=0..1} t^(1/2+n-1)*(1-t)^(3+n-1/2-n-1)/(1+3*t) dt
and for this value M Maple predicts

> t^(n-1/2)*(1-t)^(3/2)/(1+3*t) ;

> int(%,t=0..1) ;
      1/2 /    GAMMA(n - 1/2) hypergeom([1, -1 - n], [3/2 - n], -1/3)
3/4 Pi    |1/3 ------------------------------------------------------
          \                         GAMMA(2 + n)

           (-3 - n)   1/2          \
     + 32 3         Pi    sec(Pi n)|
                                   /

and this seems to lead to a terminating 2F1(1,-1-n; 3/2-n; -1/3)
for positive integer n and eventually Gamma-ratios.
Perhaps this is all wrong, but worth looking at.



More information about the SeqFan mailing list