24 Sep
2018
24 Sep
'18
1:04 p.m.
Dr. E, Mathematica thinks you need x real: In[429]:= %424[[1, 1]] /. x -> I \[Pi] Out[429]= (2^(-(1/2) - 3 n) Cos[(1 + 2 n) \[Pi]] (2 n)!)/((1 + 2 n) (n!)^2) Similarly for t in your ln 2 formula. For π, best convergence is 1 bit/term at x=0, rapidly decaying toward 0 bits/term away from x=0. —rwg On Mon, Sep 24, 2018 at 10:45 AM françois mendzina essomba2 < m_essob@yahoo.fr> wrote:
Hello,
For all x,
sum((2*n)!/(2^(3*n+1/2))/n!^2/(2*n+1)*(1/cosh(2*x)^(1/2))^(2*n+1)*cosh((2*n+1)*x),n = 0 .. infinity) = 1/4*Pi
FME...