In[]:=
deploy
Mon 5 Jul 2021 16:00:07
https://math.stackexchange.com/questions/4180259/behavior-of-inverse-of-fs-frac1n-sum-i-1n-left1-frac1i-right

Normalize so that f(n,0)=1

In[]:=
f[s_,n_]:=
1
HarmonicNumber[n]
Sum
s
1-
1
i
1
i
,{i,1,n};​​Plot[f[100,n],{n,1,100},PlotLabel"f(100,n)",AxesLabel{"n","f"}]
Out[]=
In[]:=
eps=
-6
10
;​​ClearAll[ff];​​doit[n_,x0_]:=s/.FindRoot[f[s,n]eps,{s,x0},WorkingPrecisionLog[2,n]+$MachinePrecision];​​ff[1]=1;​​ff[n_Integer]:=ff[n]=doit[n,ff[n-1]];data=Table[{n,ff[n]},{n,2,100}];​​ListLogLogPlot[data,PlotLabelg[N@eps,n],AxesLabel{"n","g"}]
Out[]=

Katsurda result check

original formula

In[]:=
katsurda[x_]:=NSum
j
(-x)
j!
Zeta[j],{j,2,Infinity};​​katsurdaApprox[x_]:=x(Log[x]+2EulerGamma-1)-Zeta[0];
In[]:=
plot1=Plot[katsurda[x],{x,0,30}];​​plot2=DiscretePlot[katsurdaApprox[x],{x,0,30,2}];​​Show[plot1,plot2]
Out[]=
5
10
15
20
25
30
20
40
60
80
100
120

derivative formula

In[]:=
dkatsurda[x_]:=NSum
-
j
(-x)
j!
Zeta[j+1],{j,1,Infinity};​​dkatsurdaApprox[x_]:=Log[x]+2EulerGamma;​​plot1=Plot[dkatsurda[x],{x,0,30}];​​plot2=DiscretePlot[dkatsurdaApprox[x],{x,0,30,2}];​​Show[plot1,plot2]
Out[]=
5
10
15
20
25
30
2
4
6