|
|
Re: Bug in infinite sum
Posted:
Jul 22, 2012 4:33 AM
|
|
Am 21.07.2012 05:44, schrieb Dr. Wolfgang Hintze:
> Consider this nice sum
>
> s[x_] = Sum[Binomial[x, k], {k, 0, Infinity}]
> 2^x
>
> and that, (with the odd terms only)
>
> u[x_] = Sum[Binomial[x, 2*k + 1], {k, 0, Infinity}]
> 2^(-1 + x)
>
> But now the even terms only ... and the surprise:
> g[x_] = Sum[Binomial[x, 2*k], {k, 0, Infinity}]
>
> (Sqrt[Pi]*Gamma[1 + x]*GegenbauerC[x, 1/2 - x, 1])/
> (2^x*Gamma[1/2 + x])
>
> looks complicated, but let's see
> FullSimplify[g[x], x > 0]
>
> 2^x*Cos[Pi*x]
>
> much simpler, but definitely wrong (giving e.g. 0 for x=1/2)
> Of course g should be s - u = 2^x -1/2 2^s = 1/2 2^s.
>
> Best regards,
> Wolfgang
>
Hmm on version 8 (Windows 64-bit) I get:
Sum[Binomial[x, 2*k], {k, 0, Infinity}]
2^(-1+x)
and the same result for
Sum[Binomial[x, k], {k, 0, Infinity, 2}]
as it should be.
Peter
|
|
