06/14/2021, 12:07 AM
Here I present my latest improved method of the infinite composition method we started talking about in 2021.
Let x be complex with Re(x) > 1.
Consider f(x) = exp(t(x-1) * f(x-1)) = exp(t(x-1) * exp(t(x-2) * f(x-2)) = exp...
One of James last considerations was t(x) = 1/( exp(-x) + 1). (well almost , he considered the isomorphic case f(x) = t(x-1) * exp(f(x-1)) ) )
I came up with t(x) = 1/(gamma(-x,1) + 1).
Now I propose another function t(x).
Alot can be said but I wont go into details yet.
However some pictures might say more than words, so I will add a few.
I will not talk about poles singularities and zero's for now. I will pretent they do not exist at the moment to avoid complications.
This ofcourse follows from the general sigmoid type function ideas ( t(x) = 1/ something ).
I also note that in general t(x) = 1/( too-fast(x) + 1) is generally to chaotic to work for most " too-fast functions " such like triple exp growth rate.
The pictures and proposed function might clarify that.
I worked with sage for the proposed function and plots.
So copied from sage I used :
---
complex_plot(h(x)*h(2*x)*h(3*x)*h(4*x)*h(5*x),(-40,40),(-40,40))
Launched png viewer for Graphics object consisting of 1 graphics primitive
sage: h(x)
1/(e^(2*x^(3/2)*sinh(-sqrt(x))) + 1)
---
where t(x) = h(x)*h(2*x)*h(3*x)*h(4*x)*h(5*x)
This gives faster convergeance in a sufficiently large domain.
The many products were neccessary to avoid infinite switches from near 1 to near 0 , again see pictures.
( black is zero , white is infinity , other colors are arguments )
regards
Tom Marcel Raes
tommy1729
