Possible growth of arithmetical complexity - PowerPoint PPT Presentation

1 / 33
About This Presentation
Title:

Possible growth of arithmetical complexity

Description:

was invented by S. V. Avgustinovich in 1999. but. Theorem ... For the Fibonacci word. A. Frid. Growth of arithmetical complexity. 31. WACaM'04. July 17, 2004 ... – PowerPoint PPT presentation

Number of Views:54
Avg rating:3.0/5.0
Slides: 34
Provided by: afr73
Category:

less

Transcript and Presenter's Notes

Title: Possible growth of arithmetical complexity


1
Possible growth of arithmetical complexity
  • Anna FridSobolev Institute of MathematicsNovosib
    irsk, Russiafrid_at_math.nsc.ruhttp//www.math.nsc
    .ru/LBRT/k4/Frid/fridanna.htm

2
Arithmetical closure
is the arithmetical closure of since
3
Van der Waerden theorem
What else may occur in ?
4
A simple question
Does 010101 always occur in the Thue-Morse word?
0110 1001 1001 0110 1001 0110
5
Subword and arithmetical complexity
  • Subword complexity
  • Arithmetical complexity

number of factors of w of length n
number of words of length n in A(w)
6
Possible growth?
Many examples, no characterization
NO
7
Paperfolding word
wT(P,P,)T(P)
P0?1? a pattern
0
the paperfolding word
aw(n)8n4 for n gt 13
A generalization Toeplitz words
8
First results and classification
Linearar. compl.
Exponentialar. compl.
Fixed pointsof uniform morphisms
Thue-Morse word
Paperfolding word
Avgustinovich, Fon-Der-Flaass, Frid, 00 (03)
9
Arguments for arithmetical complexity
  • Mathematics involved
  • Van der Waerden theorem
  • more number theory Legendre symbol, Dirichlet
    theorem, computations modulo p (for words of
    linear complexity)
  • linear algebra (for the Thue-Morse word etc.)
  • geometry (for Sturmian words)

10
Further results
  • ar. compl. of fixed points of symmetric morphisms
    Frid03
  • characterization of un. rec. words of linear ar.
    compl. Frid03
  • uniformly recurrent words of lowest complexity
    Avgustinovich, Cassaigne, Frid, submitted
  • a family with ar. compl. from a wider class
    (new)
  • on ar. compl. of Sturmian words (Cassaigne,
    Frid, preliminary results published)

11
Symmetric D0L words
Thue-Morse morphism,ar. compl. of the fixed
point is 2n
In general, on the q-letter alphabet
aw(n)q2kn-2, kq.
12
p-adic complexity
is the nimber of words occurring in
subsequences of differences of w
0110 1001 1001 0110 1001 0110
Our technique does not work
13
Regular Toeplitz words
P1ab?cd? a (3-regular) pattern
P2ef? a (3-regular) pattern
14
Linearity
Uniformly recurrent word all factors occur
infinite number of times with bounded gaps
  • Theorem. Let w be a uniformly recurrent infinite
    word. Then aw(n)O(n)
  • iff up to the set of factors
    wT(P1,P2,), where
  • all patterns Pi are p-regular for some fixed
    prime p
  • sequence P1,P2, is ultimately periodic

15
Another example
P0?1? 2-regular paperfolding
pattern
Q23? 3-regular
wQT(P,P,)T(P1,P2, P1,P2,), where
0
3
2
2
3
2
0
1
2
0
2
3
3
0
3
P12?0?3?2?1?3?
P23?0?2?3?1?2?
16
Lowest complexity?
A word w is Sturmian if its subword complexity is
minimal for a non-periodic word
Is arithmetical complexity of a Sturmian word
also minimal?
NO, it is not even linear (Sturmian words are not
Toeplitz words) What words have lowest ar.
complexity?
17
Relatives of period doubling word
Let a be a symbol, p be a prime integer. Define
Rp(a)ap-1?
and
wpT(Rp(0),Rp(1),, Rp(0),Rp(1),)
0100 0101 0100 0100 0100 0101 0100 0100
period doubling word
18
Minimal ar. complexity
and these limits are minimal for uniformly
recurrent words
Avgustinovich, Cassaigne, Frid, submitted
19
Plot for ar. complexity of w
p
ar. compl.
length
20
Not uniformly recurrent?
All results on linearity are valid only for
uniformly recurrent word. Open problem. Are
there (essentially) not uniformly recurrent words
of linear arithmetical complexity?
something un. rec. word
is not considered
21
More classification
Sturmian words,O(n3)
Exponentialar. compl.
Symm. D0L words
22
Words with aw(n)O(nfu(logp(n)))
Recall that for a symbol a and a prime p
Rp(a)ap-1?.
u0010
000000001 000000001 000000000
23
A theorem
Theorem. For all u (on a finite alphabet) and
each prime pgt2, aw(u)(n)O(nfu(logp(n
))).
for p2, the situation is more complicated
since 01010101... may occur both in
and
24
Particular cases
  • If u is periodic, then aw(n)O(n) which
    agrees with the characterization above
  • If fu(n)O(n), then aw(n)O(n log n) for
    example, when u is a Sturmian word, or the
    Thue-Morse word, or 0 1 00 11 0000 1111
  • If fu(n)O(n log n), then aw(n)O(n log n
    log log n)
  • If fu(n)O(na), then aw(n)O(n (log n)a)

25
Particular cases - 2
  • If fu(n)O(an), then aw(n)O(n1log a)
    so, on the binary alphabet we can reach
    aw(n)O(n1log 2) for larger alphabets, the
    degree grows.
  • If fu(n) grows intermediately between
    polynomials and exponentials, then aw(n) grows
    intermediately between n log n and polynomials.

p
3
26
Geometric definition of Sturmian words
27
Subsequence of difference 2
Factors of an arithmetical subsequence also can
be represented as intersections of a line with
the grid
28
Dual picture gates and faces
Berstel, Pocchiola, 93
29
Counting faces
By Euler formula, fe-v1

where is the Euler function
We have
30
Computational results
It seems that for 1/3lt lt2/3,
where is a simple function, ultimately
periodic on .
31
The current state
linear subword complexity
Sturmian words,O(n3)
Exponentialar. compl.
Symm. D0L words
32
Other complexities
Only complexities which are not less than the
subword one
  • d-complexity, Ivanyi, 1987
  • pattern complexity, Restivo and Salemi, 2002
  • maximal pattern complexity, Kamae and
    Zamboni, 2002
  • modified complexity, Nakashima, Tamura,
    Yasutomi, 1999

33
Thank you
Write a Comment
User Comments (0)
About PowerShow.com