Note on Fib(n), L(n)
creigh at o2online.de
creigh at o2online.de
Mon Aug 16 09:59:08 CEST 2004
I find similarities among the following three formula worth noting:
(for an explanation of notation, see below or http://www.crowdog.de ->
Floretions; a Java "FAMP" program given can also be downloaded from
there )
A057681(n) = ves( (E'x)^n )/2
E' = ( - i' - i' + 'ii' + 'jj' + 'kk' + 'jk' + kj' + 1 )/4
x = 'i + 'i + 'ji' + 'ki' + 1 (signs are preserved)
A000032(n) = L(n) = ves( (Ex)^n ) (disregarding signs)
E = ( i' + i' + 'ii' + 'jj' + 'kk' + 'jk' + kj' + 1 )/4
x = 'i + 'i + 'ji' + 'ki' + 1
A000045(n) = Fib(n) = ves( (E''x)^n )
E'' = ( - i' - i' - 'ii' + 'jj' + 'kk' + 'jk' + kj' - 1 )/4
x = 'i + 'i + 'ji' + 'ki' + 1
****************
The following is becoming part of a "standard procedure":
once a formula for any sequence has been found (since forall z
in "Floretion Algebra": jes(z) + les(z) + tes(z) = ves(z) by
definition), "jes", "les", and "tes" can be used to separate (or "disect")
it into the termwise sum of three sequences. This often produces
interesting results (which have only begun to be documented)
Examples:
(1) A001906(n+1) = (A005248(n+1) - A011783(n))/2
(2) A081010 a(n) = 2 + .5[A001906(n+1)]^2 + .5[A011783(n)]^2
To arrive at the above, notice that ves( (Ex)^n ) = A001906(n+1) with
E = .25( - 'i - i' + 'ii' + 'jj' + 'kk' + 'jk' + 'kj' + 1 )
x = 'i + 'k + j' + 'jj' + 1
(this was not a total "shot in the dark"- it was known that "E",
along with a few other elements, have special "symmetric" properties)
It was then verified that
jes( (Ex)^n ) = - A011783(n)/2
les( (Ex)^n ) = tes( (Ex)^n ) = A005248(n+1)/4
ves( (Ex)^n ) = A001906(n+1)
The first formula now follows immediately using
jes(z) + les(z) + tes(z) = ves(z)
The second formula follows from the first using the relation
A005248(n) = sqrt((A005248(n)^2-4)/5) given at
http://www.research.att.com/projects/OEIS?Anum=A001906
**************
Explanation of Formula:
For elements of the factor space Q x Q / {(1,1), (-1,-1)},
Q = Quaternions I write 'i to mean [ (i, 1) ] = [ (-i, -1) ],
i' to mean [ (1, i) ] = [ (-1, -i) ], 'ij' to mean [ (i, j) ], etc.
Let F be the algebra over the reals generated from this group.
A typical element z of F can be written by 16 basis vectors:
x = x(1) 'i + x(2) 'j + x(3) 'k + x(4) i' + x(5) j' + x(6) k' + x(7)
'ii'
+ x(8) 'jj' + x(9) 'kk' + x(10)'ij' + x(11) 'ik' + x(12) 'ji' +
x(13) 'jk'
+ x(14) 'ki' + x(15) 'kj' + x(16) 1
**********
Note: technically, one need only consider the ring F* generated
by Q x Q / {(1,1), (-1,-1)}, however, we often need to consider elements
of the form x = tz, where z in F* and t = 1/n for some n in naturals;
on occasion I've also needed to set n = sqrt(2). Regardless, one could
always deal exclusively with the ring and divide off these numbers
afterwards, ex.:
A057681(n) = ves( (E'x)^n )/2
E' = ( - i' - i' + 'ii' + 'jj' + 'kk' + 'jk' + kj' + 1 )/4
x = 'i + 'i + 'ji' + 'ki' + 1
becomes
A057681(n) = ves( (E'''x)^n )/(2*4^n)
E''' = ( - i' - i' + 'ii' + 'jj' + 'kk' + 'jk' + kj' + 1 )
x = 'i + 'i + 'ji' + 'ki' + 1
***********
Define the (linear) functions:
tes: F -> reals,
ves: F -> reals,
jes: F -> reals,
les: F -> reals
tes( (x_1)'i + (x_2)'j + ... + (x_16)1 ) = x_16
ves( (x_1)'i + (x_2)'j + ... + (x_16)1 ) = x_1 + x_2 + ... + x_16
jes( (x_1)'i + (x_2)'j + ... + (x_16)1 ) = x_1 + x_2 + x_3 + x_4 + x_5
+ x_6
les = ves - tes - jes
Sincerly,
Creighton Dement
More information about the SeqFan
mailing list