NSolve[(-(2/Cos[Pi*x]))*Sum[(-1)^i*(2*Pi*x)^(2*i)*Sum[(-1)^j/((2*Pi)^(2*j)*Zeta[2*j]*(2*i+1-2*j)!),{j,1,i}],{i,1,Infinity}]0,x](*Linecopiedforyourpost*)
NSolve
:This system cannot be solved with the methods available to NSolve.
Out[]=
NSolve-2Sec[πx]
∞
∑
i=1
i
(-1)
2i
(2π)
2i
x
i
∑
j=1
j
(-1)
-2j
(2π)
(1+2i-2j)!Zeta[2j]
0,x
SumConvergence
j
(-1)
-2j
(2π)
(1+2i-2j)!Zeta[2j]
,j(*SumConvergencetestdoesnotworkforthelastsum.Mycomputermaynothaveenoughresourcestoevaluatethetestortheexpressionmayhaveerrors*)​​SumConvergence[
i
(-1)
2i
(2π)
2i
x
,i](*Ihadtotestthefirstpartwithoutthenestedsumsincethenestedsumdidnotpasstheconvergencetest.*)
Out[]=
SumConvergence
j
(-1)
-2j
(2π)
(1+2i-2j)!Zeta[2j]
,j
Out[]=
Abs[x]<
1
2π
||2πx1||1+2πx0
(*SumConvergenceofthefirstsumshowsthatitonlyconvergeswhen-
1
2π
<x<
1
2π
*)
f[x_,n_]=(-(2/Cos[Pi*x]))*Sum[(-1)^i*(2*Pi*x)^(2*i)*Sum[(-1)^j/((2*Pi)^(2*j)*Zeta[2*j]*(2*i+1-2*j)!),{j,1,i}],{i,1,n}](*Mycomputerwillnotevaluatetheinfinitesum.Therefore,Ireplace"Infinity"withn.Then,Iplottedthefunctiontoshowwhathappensasngetsverylarge.*)​​Plot3Df[x,n],x,
-1
2π
,
1
2π
,{n,0,30},AxesLabel{"x","n","f(x)"}
Out[]=
-2Sec[πx]
n
∑
i=1
i
(-1)
2i
(2π)
2i
x
i
∑
j=1
j
(-1)
-2j
(2π)
(1+2i-2j)!Zeta[2j]
Out[]=
(*Thefunctionseemstoconvergefastasnbecomeslarger.Thisgraphisstillonlyanapproximation,andwithitImakethebestassumptionthattheonlytimethefullequationholdstrueiswhenx=0.*)