[seqfan] Re: detective work related to Creighton Dement's \"Floretions\"
Alonso Del Arte
alonso.delarte at gmail.com
Fri Nov 20 02:48:14 CET 2009
Creighton, maybe you could write the Mathworld entry on floretions.
Something about as long as the Mathworld entry on Julia sets that would
cover most of the basics of floretions and point out some of the most
interesting specimens. Al
On Thu, Nov 19, 2009 at 6:03 PM, Creighton Kenneth Dement <
creighton.k.dement at mail.uni-oldenburg.de> wrote:
> > I read this and got curious:
> >
> > On Mon, 9 Nov 2009, Michael Porter wrote:
> >
> >> I did an OEIS search on "floretion" and the first two pages consisted
> >> of:
> >> 1. sequences related to floretions with few references or links, and 2.
> >> big
> >> sequences like the prime numbers or the squares which have some note
> >> about
> >> floretions buried in them.? It wasn't very useful for someone looking to
> >> learn more about floretions.
> >>
> >
> > I used "Sloandora" (mrob.com/pub/math/sloandora) to find all the
> floretion
> > sequences and related sequences in OEIS (using an August 2007 copy of the
> > eisBTfry00000.txt files).
> >
> > As reported by Michael, there are lots of OEIS entries that have little
> or
> > no explanatory references, and a few (like A000330, A001541, A001792,
> > A001834) that have a reference buried in the middle of their dozens or
> > hundreds of other references.
>
> Unfortunately, I cannot address every statement right now. Let me try to
> go over some of the more important points brought up.
>
> vesseq(X), jes, les, tes, … vespos, vesneg, along with the identities
> ves = jes + les + tes, etc. are all clearly (and easily, in my opinion)
> defined here http://www.scribd.com/doc/14790151/Floretions-2009
> (you can even see these visually by clicking on "ves", "tes", etc. in the
> list selection box on the top left at
> http://www.fumba.eu/sitelayout/floretion.html )
>
> Mr. Munafo and others do have a valid point, which is that parts of the
> documentation are confusing and/or incomplete. Before I continue, I would
> like to thank him and others diving into the subject!
>
> > "1tesforzapseq" is part of an obscure naming system, combining a prefix
> > with
> > one or more three-letter "syllables" representing different types of
> > calculations that can be performed in each step of an iterative process.
>
> Most of the confusing documentation comes from the very beginning
> (2004/2005), when I was mezmerized by "all the different paths a floretion
> could be taken down" and would often wake up each morning with a new idea.
>
> In some cases I was trying to show
> when you do “A”, you get a 4th order sequence
> when you do “B”, you get a 2nd order sequence
> when you do “C” this you get something very strange
>
> and simply choosing one or more representative sequences from each class
> (because I certainly didn't have enough time to submit them all).
>
> My “old, dead website” has been discussed on the seqfan list (a few years
> ago). What happened is that I had a private website with several pages
> dedicated to explaining the terminology and giving examples, but where I
> had to pay what became an overpriced monthly fee to maintain it. After a
> year or two, I ran into financial trouble and could no longer pay the
> monthly rates. Apparently, the site was then shut down very quickly and
> the domain name sold to another "party". Years later, some data is still
> retrievable but it seems several pages are missing and/or are
> non-recoverable.
>
> > Similarly, A105770 makes the bizarre statement "*This sequence is
> > 'tesrokseq' at the link 'Sequences in Context'.*" and also *"Link to
> > Sequences in Context contains futher details on the 'roktype' used"*. The
> > writer clearly did not understand what the "Sequence in context:" links
> > are.
> > A105660 is similar.
>
> When my site was taken offline, a member of the seqfan list wrote to Neil
> and me that my site and been taken down and and that (yikes!) malicious
> code was now being executed from it. Needless to say, I was quite upset.
> At that point, Neil asked me what he should do with all the (more than
> 200, I believe) references to my homepage. I agreed that he should remove
> all links as soon as possible and he did. This explains why some sequences
> say “See sequences in context” or “See FAMP” in the comments but with no
> proper link. Formerly, there would have been a link to my website with a
> batch of related sequences (i.e. all generated by the same floretion under
> the same conditions). There was also a link to download FAMP (the
> Floretion Algebra Multiplication Program). I have since tried to recreate
> an “online version of FAMP” here
> http://www.fumba.eu/sitelayout/floretion.html (however, a lot of the
> functionality is still missing for lack of time and in some cases security
> reasons).
>
> One of the most confusing statements is surely what is meant by a
> "transform of the zero-sequence". First, I will agree that if the
> consensus is this is totally nonsense, then by all means let's call it
> something else!
>
> Here's the idea: start with any floretion X and let (c(n)) be any sequence
> of integers we would like to transform.
>
> Define
> Y = X + (c(0) – ves(X))ee
> and Y(n+1) = Y^n + (c(n) – ves(Y^n))ee
> where “ee” is the unit vector.
>
> Then
> ves(Y(n+1)) = ves(Y^n + (c(n) – ves(Y^n))ee)
> = ves(Y^n) + ves((c(n) – ves(Y^n))ee)
> = ves(Y^n) + ves(c(n)ee) – ves(Y^n)
> = ves(c(n)ee)
> = c(n)
>
> The identity ves = jes + les + tes always holds, thus:
> c(n-1) = ves(Y^n) = jes(Y^n) + les(Y^n) + tes(Y^n)
>
> My terminology (which is certainly open to improvements) would say here
> that the sequence of numbers jesfor(X) = (jes(Y), jes(Y^2), …,) is the
> jesfor-transform of the sequence (c(n)) with respect to the floretion X.
>
> One of the easiest choices for the sequence to be transformed is, of
> course, the zero-sequence itself. This leads to the strange sounding (but
> consistent, imho) formulation "... jesfor-transform of the zero-sequence".
>
> Force transforms are interesting (there are at least two types- I can't
> remember which one is which off hand) and I basically abandoned the entire
> subject for lack of time. Substituting jesfor(X) (or one of the other
> generated sequences, below it is 4dia[I]seq) for the new sequence to be
> transformed led to this past seqfan post:
>
> *************
>
> Generalized Sequence Convergence?
>
> Let J be some subset of the set of all integer sequences, T[ ]: J -> J be
> some mapping "transform", and c an integer sequence.
>
> Define T^n as T applied n times and T[c](m) as the m-th term of the
> sequence T[c]. Assume that for each integer sequence c in J exists integer
> sequence d so that for each m in naturals exists p in naturals so that for
> all n > p : T^n[c](m) = d(m)
>
> In that case, we can define T*[ ]: J -> integer sequences, T*[c] = d for
> some new transform T*. Ex. T[c](m) = floor(c(m)/2), then T*[c] =
> (0,0,0,0,...) for all c. Below, I give a T* such that, apparently (and
> without proof), for all c: T*[c] = A000045 (Fib). However, it is the
> procedure itself that is stressed here and not any particular result.
>
> [snip]
>
> O-th iteration *********
>
> 1vesseq: -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1,
>
> 4tesseq: -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1,
>
> 4lesseq: -5, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3,
>
> 1jesseq: 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
>
> 4dia[I]seq: 1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1,
>
>
> 1st iteration ********* (LoopType: dia[I] )
>
> 1vesforseq: 1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1,
>
> 4tesforseq: 7, 7, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1,
>
> 4lesforseq: -5, -7, -7, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3,
>
> 1jesforseq: 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
>
> 4dia[I]forseq: 1, 3, 3, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1,
>
>
> 2nd iteration ********* (LoopType: dia[I] )
>
> 1vesforseq: 1, 3, 3, 1, -1, 1, -1, 1, -1, 1, -1,
>
> 4tesforseq: 7, 15, 21, 13, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1,
> 1, -1, 1, -1
>
> 4lesforseq: -5, -7, -17, -21, -11, 3, -3, 3, -3, 3, -3, 3, -3, 3, -3, 3,
> -3, 3, -3, 3, -3
>
> 1jesforseq: 0, 1, 2, 3, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
>
> 4dia[I]forseq: 1, 3, 5, 9, 7, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1, -1, 1,
> -1, 1, -1
>
> ... [snip] ...
>
> 8th iteration ********* (LoopType: dia[I] )
>
> 1vesforseq: 1, 3, 5, 11, 21, 43, 85, 171, 339, 649, 1103, 1505, 1471, 897,
> 255, 1, -1, 1, -1, 1, -1
>
> 4tesforseq: 7, 15, 29, 59, 117, 235, 469, 939, 1869, 3613, 6359, 9329,
> 10399, 8001, 3711, 769, -1, 1, -1, 1, -1
>
> 4lesforseq: -5, -7, -17, -31, -65, -127, -257, -511, -1025, -2037, -3923,
> -6813, -9731,
>
> 1jesforseq: 0, 1, 2, 4, 8, 16, 32, 64, 128, 255, 494, 876, 1304, 1488,
> 1184, 576, 128, 0, 0, 0, 0
>
> 4dia[I]forseq: 1, 3, 5, 11, 21, 43, 85, 171, 341, 681, 1327, 2401, 3711,
> 4481, 3839, 2049, 511, 1, -1, 1, -1
>
> Jacobsthal Sequence!
>
> ***************
>
>
> On to the next question:
>
> > I spent a few hours in Sloandora collecting these results. It found a lot
> > of
> > sequences that were authored by Creighton Dement but have nothing to do
> > with
> > his floretion work. After a while I started down-rating sequences like
> > A100545 and A113166 that say they were floretion-generated but provide no
> > further clues about the floretion definition or algorithm.
>
> Let's see how to generate A100545:
> Proposition 3.3 (Pure Quaternions and Second Order Recurrence Relations)
> states:
>
> Let Y = A'i + B'j + C'k and E = .25('i + i' + ii + jj + kk + jk + kj + ee)
>
> Then all sequences generated by the floretion E*Y satisfy the linear
> recurrence relation:
>
> a(n) = -A*a(n-1) – B*C*a(n-2)
>
> and, in general, Proposition 3.4 states
>
> Let Y = A'i + B'j + C'k + Di' + Ej' + Fk' + G'ii' + H'jj' + I'kk' +
> J'ij' + K'ik' + L'ji' + M'jk' + N'ki' + O'kj'
>
> then all sequences satisfy
>
> a(n) = (-A-D+G+H+I+M+O)*a(n-1) + ((A+D-M-O)(G+H+I)
> +(N+E-B-K)(J+C-F-L))*a(n-2)
>
> Now let's really try and find the sequence. Go to
> http://www.fumba.eu/sitelayout/floretion.html and, on the list at the
> right, choose "Basic Fibonacci". You should see that the floretion "E"
> defined in the above in Prop. 3.3 is automatically set in the top row. The
> second row contains the numbers (-1,-1,1) which corresponds to the
> quaternion Y = -'i -'j + 'k
>
> Will prop 3.3 suffice to find the sequence A100545? Well, we see it
> satisfies the recurrence a(n)=3*a(n-1)-a(n-2) so let's try the quaternion
> Y = -3'i + 'j + 'k by changing the numbers in the second row to (-3,1,1)
> and hitting the button "Go Python!" then scrolling down the page to see a
> list of sequences generated by E*Y.
>
> In this case, this returns a batch of sequences which all satisfy the
> recurrence relation a(n) = 3*a(n-1) – a(n-2), but not the sequence A100545
> example we are looking for. That said, we also chose the simplest possible
> example. The next simplest case (using Prop. 3.4) would be to choose Y =
> -2'i + 'j + 'k - i'
>
> Scan down the batch of sequences and you should see these lines:
> 2mixseq: [1, 3, 8, 21, 55, 144, 377, 987, 2584, 6765]
> 4mixposseq: [3, 8, 23, 61, 160, 419, 1097, 2872, 7519, 19685]
> 4mixnegseq: [-1, -2, -7, -19, -50, -131, -343, -898, -2351, -6155]
>
> And there's our sequence... not only that, the (dynamic) identity
> mix = mixpos + mixneg returns A055273 + A100545 = 2*A001906 (bisection of
> the Fibonacci sequence)
>
> With all that said, there may be a few sequences which have to be deleted
> because the documentation on them has been completely lost. As stated
> above, often I was choosing a representative sequence from a sea of
> possibilities. A good example of this is the sequence Neil and I
> considered for a themesong: A124856. I will have to check with past posts
> to see where the documentation is, and if it can't be found, it should be
> deleted, imo.
>
> Sincerely,
> Creighton
>
>
>
>
>
>
>
> _______________________________________________
>
> Seqfan Mailing list - http://list.seqfan.eu/
>
More information about the SeqFan
mailing list