Levin, Hyperint



Barton Willis wrote:

>
>   (%i126) hyper_int(1/sqrt(x*(x+1)*(x-1)),x);
>   (%o126)
> (2*hypergeometric([1/4,1/2],[5/4],x^2)*x*sqrt(1-x^2))/sqrt(x^3-x)
> 
> This might also be correct, but I don't know which recursions to use to
> show it.
> Evidence that %o126 might be correct:
> 
>   (%i127) taylor(diff(%,x) - 1/sqrt(x*(x+1)*(x-1)),x,0,128);
>   (%o127)/T/ 0+...
> 

It may be that this elliptic integral is particular because 0 is in the 
middle of -1, 1, which is invariant by the homographies that preserve 
infinity. And indeed:

(%i2) hyper_int(1/sqrt(x*(x+2)*(x-1)),x);
(%o2)                                false

while 
(%i4) inverse_jacobi_int(1/sqrt(x*(x+1)*(x-1)),x);
(%o4) 
                             %i sqrt(x - 1)
  -2 %i inverse_jacobi_sn(--------------, 2) sqrt(x - 1) sqrt(x) sqrt(x + 1)
                                sqrt(2)
   -------------------------------------------------------------------------
                                          3
                                    sqrt(x  - x)


-- 
Michel Talon