Loading [MathJax]/jax/output/CommonHTML/jax.js

Wednesday, December 28, 2022

Changing Flint Miller Summation

 Consider a circle of perimeter,

2πr=10

a unit circle that has been scaled by 

r=102π.  

A scaling factor of 2π10 has to be applied to the results afterwards because of this scaling.

This way, as a integer n run along the circumference, integers overlap with a previous integer (the last digit being the same).  All integer falls into 10 points on the circumference.

The scaling factor is necessary because on a unit circle, one unit length along the circumference extends one radian (rad) at the center.  It is not true here.  Scaling the unit circle to radius 102π, to accommodate 10 points on the circumference, simplifies the analysis. 

Consider, 

(2π10)2n=1csc2(n.2π10)n3=π225n=1csc2(nπ5)n3

where (2π10)2 scales (csc2n) back to its length on a unit circle.

n=1csc2(nπ5)n3=m=010n=11(10m+n)3csc2(nπ5)

because when n=10nπ5=2π and the next 10 addends scaled by 1n3 repeats on the circle in the same position.  We extract the first 10 terms for m=0,

10n=11n3csc2(nπ5)

We are then left with eq(1),

m=110n=11(10m+n)3csc2(nπ5)

Consider, when n=1,

csc2(π5)m=11(10m+1)3

Consider, when n=2,

csc2(2π5)m=11(10m+2)3=csc2(2π5)m=1181(5m+1)3

which is for all even n,

185i=1csc2(2iπ5)m=11(5m+i)3

then (1) is then,

5i=1csc2((2i1)π5)m=11(10m+(2i1))3+185i=1csc2(2iπ5)m=11(5m+i)3

And the origin summation we started with is,

π225[5i=1csc2((2i1)π5)m=11(10m+(2i1))3+185i=1csc2(2iπ5)m=11(5m+i)3+10n=11n3csc2(nπ5)]

Convergent but...NOT Quite!

Addends at π and 2π should cancels leaving 8 points on the circle, but it is not satisfying.

Note:

Flint Miller Series is n=1csc2(n)n3