Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
some questions about sexp
#13
(07/05/2010, 09:59 PM)tommy1729 Wrote: you can find g that can't be written as f°f°...°f for any n > 1.
Namely, suppose g has exactly one fixed point p_0, and exactly one p_1 <> p_0 such that
g(p_1) = p_0,....and g = f°f°...°f,
it's easy to see that .... f(p_1) = p_0

> f(p_1) = p_0
I don't understand your equation, but it isn't true if g=e^x. However, that could be because e^x has an infinite number of p_1 alternatives such that e^p_1=p_0, the fixed point of base "e". For exp(x), p_1=p_0 + i2pi*n
(07/05/2010, 09:59 PM)tommy1729 Wrote: A suitable g is g(x) = x + 2 - 2 exp(x).
g(x) has a fixed point at zero, (the other fixed points are 0+i2pi*n). The slope of g(x) at x=0 is -1. This is a strange oscillatory fixed point; actually it appears to be an attracting fixed point for real values, and a repelling fixed point for complex values. I wouldn't know how to expand g(x) into a superfunction .... unless there is another more well behaved fixed point. But if there was another fixed point, that would contradict Tommy's assumption, "suppose g has exactly one fixed point p_0". Because the slope of the fixed point of zero is -1, there isn't a well defined superfunction, and because there isn't a well defined superfunction, then there isn't a well defined half iterate (or n-th iterate).

The "p_1" value for g(x) is approximately -1.5936. g(-1.5936)~=0. But, again, we can't calculate the half iterate of p_1. Nor can we calculate the half-iterate of p_0.

added stuff I discovered upon looking a little closer; 2nd update for clarity
Now for the surprise; the function g(x) does half another fixed point -- actually a fixed slope -- of sorts, and that can be used to define an unexpected alternative definition of the half iterate of p_1=-1.5936. Consider the sequence p(0), p(-1), p(-2), p(-3), p(-4), p(-5), p(-6) .... p(-n), where p(0)=0, and g(p(-1))=p(0)=0, and p(n+1)=g(p(n)). Here's how the sequence looks, accurate to four decimal places. Notice the pattern? For large enough negative values of n, p(n+1)=p(n)+2, because the exponential term becomes insignificant.

p(-9 )=-17.5261
p(-8 )=-15.5261
p(-7 )=-13.5261
p(-6 )=-11.5261
p(-5 )=-9.5261
p(-4 )=-7.5263
p(-3 )=-5.5274
p(-2 )=-3.5353
p(-1 )=-1.5936
p( 0 )=0.0000
p( 1 )=0.0000
p( n )=0.0000

This sequence can be extended to half iterates, with results as follows, and can also be used to generate a superfunction (the limit definition is obvious, but I can post it if others are interested). With this definition of the superfunction, the half iterate of p(-1)=p(-0.5)=0.7071. Then there are an infinite number of half iterates of zero depending on the path, and beginning with p(0.5)=0.3068. p(1.5), p(2.5), p(3.5) ... are all alternative half iterates of zero.

p(-9.5)=-18.5261
p(-8.5)=-16.5261
p(-7.5)=-14.5261
p(-6.5)=-12.5261
p(-5.5)=-10.5261
p(-4.5)=-8.5262
p(-3.5)=-6.5266
p(-2.5)=-4.5295
p(-1.5)=-2.5511
p(-0.5)=-0.7071
p(0.5)=0.3068
p(1.5)=-0.4113
p(2.5)=0.2631
p(3.5)=-0.3388
p(4.5)=0.2360
p(5.5)=-0.2963
p(6.5)=0.2166
p(7.5)=-0.2670
p(8.5)=0.2017
p(9.5)=-0.2452
- Sheldon
Reply


Messages In This Thread
some questions about sexp - by tommy1729 - 06/28/2010, 11:18 PM
RE: some questions about sexp - by tommy1729 - 06/29/2010, 10:55 PM
RE: some questions about sexp - by bo198214 - 07/01/2010, 09:39 AM
RE: some questions about sexp - by mike3 - 07/01/2010, 11:05 AM
RE: some questions about sexp - by bo198214 - 07/02/2010, 07:14 AM
RE: some questions about sexp - by tommy1729 - 07/01/2010, 10:06 PM
RE: some questions about sexp - by mike3 - 07/01/2010, 10:28 PM
RE: some questions about sexp - by bo198214 - 07/02/2010, 07:17 AM
RE: some questions about sexp - by mike3 - 07/02/2010, 07:55 AM
RE: some questions about sexp - by tommy1729 - 07/02/2010, 12:27 PM
RE: some questions about sexp - by sheldonison - 07/02/2010, 03:52 PM
RE: some questions about sexp - by tommy1729 - 07/05/2010, 09:59 PM
RE: some questions about sexp - by sheldonison - 07/06/2010, 01:45 PM

Possibly Related Threads...
Thread Author Replies Views Last Post
  Sexp redefined ? Exp^[a]( - 00 ). + question ( TPID 19 ??) tommy1729 0 1,513 09/06/2016, 04:23 PM
Last Post: tommy1729
  Can sexp(z) be periodic ?? tommy1729 2 3,674 01/14/2015, 01:19 PM
Last Post: tommy1729
  pseudo2periodic sexp. tommy1729 0 1,750 06/27/2014, 10:45 PM
Last Post: tommy1729
  [2014] tommy's theorem sexp ' (z) =/= 0 ? tommy1729 1 2,721 06/17/2014, 01:25 PM
Last Post: sheldonison
  Multiple exp^[1/2](z) by same sexp ? tommy1729 12 13,235 05/06/2014, 10:55 PM
Last Post: tommy1729
  entire function close to sexp ?? tommy1729 8 9,117 04/30/2014, 03:49 PM
Last Post: JmsNxn
  Is sexp(z) pseudounivalent for Re(z) > 0 ? tommy1729 3 3,304 03/26/2014, 01:24 PM
Last Post: tommy1729
  Vincent's theorem and sin(sexp) ? tommy1729 0 1,679 03/22/2014, 11:46 PM
Last Post: tommy1729
  sexp for base (1/e)^e ~= 0.0660? sheldonison 10 10,464 11/22/2013, 11:20 PM
Last Post: mike3
  tetration base conversion, and sexp/slog limit equations sheldonison 44 52,921 02/27/2013, 07:05 PM
Last Post: sheldonison



Users browsing this thread: 1 Guest(s)