DYNAMICS OF SOFIC SHIFTS
ALI AKBAR K.
Department of Mathematics, Central University of Kerala, Kasaragod (India).
E-mail: aliakbar.pkd@gmail.com, aliakbar@cukerala.ac.in
ORCID: https://orcid.org/0000-0003-3542-3727
Reception: 13/08/2022 Acceptance: 28/08/2022 Publication: 29/12/2022
Suggested citation:
Ali A. K. (2022). Dynamics of Sofic Shifts. 3C Tecnología. Glosas de innovación aplicada a la pyme,11 (2), 13-23.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
13
ABSTRACT
In this paper, we provide a characterization for the subshifts of finite type (
SFT
) in terms of Cellular
automata (CA). In addition, we prove that
1. The following are equivalent for a non-singleton subshift of finite type XF.
a) XFis transitive and P er(XF), the set of periodic points of XF, is cofinite
b) XFis weak mixing
c) XFis mixing.
2. For non-singleton sofic shifts, only the statements (a)and (b)are equivalent.
KEYWORDS
subshift of finite type, sofic shifts, mixing, strongly connected digraph, labeled digraph, cellular automata
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
1 INTRODUCTION AND PRELIMINARIES
A dynamical system is a pair (
X, f
), where
X
is a metric space and
f
is a continuous self map. For
each dynamical system (
X, f
), the period set
Per
(
f
)=
{nN
:
xX
such that
fn
(
x
)=
x
=
fm
(
x
)
m<n}
consisting of the lengths of the cycles there, is a subset of the set
N
of positive integers.
We say that a point
xX
is periodic if
fnx
=
x
for some
nN
, where
fn
is the composition of
f
with itself
n
times. The smallest such positive integer
n
is called the period of
x
. Two dynamical
systems (
X, f
),(
Y,g
)are said to be topological conjugate if there exists a homeomorphism
h
:
XY
such that
hf
=
gh
. If there is a continuous surjection
h
:
XY
such that
hf
=
gh
then
we say that (
Y,g
)is a factor of (
X, f
). A dynamical system (
X, f
)is said to be transitive if for any
non-empty open sets
U, V
in
X
there exists
nN
such that
fn
(
U
)
V
=
and is said to be mixing if
for any non-empty open sets
U, V
in
X
there exists
nN
such that
fm
(
U
)
V
=
for all
mn
.A
dynamical system (
X, f
)is said to be weak mixing, if given any four nonempty open sets
U1,V
1,U
2,V
2
in Xthere exists mNsuch that fm(U1)V1and fm(U2)V2are non-empty.
The subshifts form an important class of dynamical systems, because almost all dynamical systems
are factor of some subshifts. There are plenty of books that explain how their study would throw light on
still larger classes of dynamical systems (see [7], [10] and [14]). There are some dynamical systems such
that the notions transitivity together with cofiniteness, weak mixing, and mixing which are equivalent
(See [5], [8] for interval maps, See [2] for topological graph maps). It is natural to ask on which spaces a
similar result will be true. We find that the same result is true in the class of non-singleton
SFT
s, and in
the class of continuous 2-dimensional toral automorphisms. Note that
SFT
s, continuous 2-dimensional
toral automorphisms and topological graph maps are different kinds of dynamical systems, and we
cannot hope to have a similarity of proofs. In this paper, we concentrate on two-sided shifts. The case
of one-sided shifts is similar.
Let
A
be a non-empty finite set (called alphabet) with discrete topology and consider the set
AZ
,
which denotes the set of doubly-infinite sequences (
xi
)
iZ
where each
xi∈A
, with product topology. It
is compact and metrizable. The shift is the homeomorphism
σ
:
AZ→A
Z
given by
σ
(
x
)
i
=
xi+1
for all
iZ
. A subset
A⊂A
Z
is called
σ
-invariant if
σ
(
A
)
A
. The pair (
AZ
)forms a dynamical system
called a full shift. A subshift is a
σ
- invariant non-empty closed subset
X
of a full shift, together with
the restriction of
σ
to
X
. We denote the set of periods of all periodic points of
σ
in
X
by
P er
(
X
). We
call
P er
(
X
)the period set of
X
. A word
w
on
A
is a concatenation
w1w2...wk
, where each
wi∈A
and (
w
)
n
=
wr
whenever
nr
(
mod k
). A subshift
X
is said to be a subshift of finite type (SFT) if
X
=
XF
=
{x∈A
Z
:no word in
F
occurs in
x}
for some finite set of words
F
. A subshift
X⊂A
Z
is
called sofic if it is a factor of an SFT.
The notion of strongly connected digraphs is well known. For every SFT, there is an associated
digraph and an associated matrix as described in [7]. This may or may not be strongly connected. Let
G
=(
V,E
)be any directed graph (digraph) with vertex set
V
and edge set
E
. A subgraph
G
=(
V,E
)
of
G
is said to be a full subgraph if
E
=
E∩{
(
v1,v
2
):
v1,v
2V}
. A digraph is said to be simple if
from every vertex
v
to a vertex
w
there is atmost one edge and it is said to be strongly connected if for
every pair of vertices there exists a directed path. It is to be noted that a connected digraph may not
be strongly connected. The SFT associated for a digraph
G
is denoted as
XG
and the SFT associated
for an
m×m
matrix
A
with entries 0or 1is denoted as
XA
. Let
V
=
{vV
:there exists a cycle
through this
v}
. Consider a full subgraph of a simple digraph
G
with vertex set
V
, say
G
=(
V,E
).
Define a relation
R
on
V
in such a way that
xRy
if there is a directed path from
x
to
y
and from
y
to
x
. Then
R
is an equivalence relation on the set of vertices
V
. Let
G
x
be the full subgraph of
G
with [
x
], equivalence class of
x
, as the vertex set. This
G
x
is strongly connected simple digraph and
G
can be written as a finite union of such strongly connected simple digraphs, say
n
i=1 Gi
. The SFT
associated for a simple digraph Gis denoted as XG. Note that P er(XG)=P er(XG).
Let
A
be an alphabet having atleast two elements. Let
rN0
. A function
f
:
A2r+1 →A
is called a
local rule. It induces a function F:AZ→A
Zby the rule
(
F
(
x
))
n
=
f
(
xnr,x
nr1, ..., xn1,x
n,x
n+1, ..., xn+r1,x
n+r
)for all
nZ
. The pair (
AZ,F
)(sim-
ply the map
F
:
AZ→A
Z
) is called a cellular automaton (abbreviated as
CA
). A map
F
:
AZ→A
Z
is
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
14
ABSTRACT
In this paper, we provide a characterization for the subshifts of finite type (
SFT
) in terms of Cellular
automata (CA). In addition, we prove that
1. The following are equivalent for a non-singleton subshift of finite type XF.
a) XFis transitive and P er(XF), the set of periodic points of XF, is cofinite
b) XFis weak mixing
c) XFis mixing.
2. For non-singleton sofic shifts, only the statements (a)and (b)are equivalent.
KEYWORDS
subshift of finite type, sofic shifts, mixing, strongly connected digraph, labeled digraph, cellular automata
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
1 INTRODUCTION AND PRELIMINARIES
A dynamical system is a pair (
X, f
), where
X
is a metric space and
f
is a continuous self map. For
each dynamical system (
X, f
), the period set
Per
(
f
)=
{nN
:
xX
such that
fn
(
x
)=
x
=
fm
(
x
)
m<n}
consisting of the lengths of the cycles there, is a subset of the set
N
of positive integers.
We say that a point
xX
is periodic if
fnx
=
x
for some
nN
, where
fn
is the composition of
f
with itself
n
times. The smallest such positive integer
n
is called the period of
x
. Two dynamical
systems (
X, f
),(
Y,g
)are said to be topological conjugate if there exists a homeomorphism
h
:
XY
such that
hf
=
gh
. If there is a continuous surjection
h
:
XY
such that
hf
=
gh
then
we say that (
Y,g
)is a factor of (
X, f
). A dynamical system (
X, f
)is said to be transitive if for any
non-empty open sets
U, V
in
X
there exists
nN
such that
fn
(
U
)
V
=
and is said to be mixing if
for any non-empty open sets
U, V
in
X
there exists
nN
such that
fm
(
U
)
V
=
for all
mn
.A
dynamical system (
X, f
)is said to be weak mixing, if given any four nonempty open sets
U1,V
1,U
2,V
2
in Xthere exists mNsuch that fm(U1)V1and fm(U2)V2are non-empty.
The subshifts form an important class of dynamical systems, because almost all dynamical systems
are factor of some subshifts. There are plenty of books that explain how their study would throw light on
still larger classes of dynamical systems (see [7], [10] and [14]). There are some dynamical systems such
that the notions transitivity together with cofiniteness, weak mixing, and mixing which are equivalent
(See [5], [8] for interval maps, See [2] for topological graph maps). It is natural to ask on which spaces a
similar result will be true. We find that the same result is true in the class of non-singleton
SFT
s, and in
the class of continuous 2-dimensional toral automorphisms. Note that
SFT
s, continuous 2-dimensional
toral automorphisms and topological graph maps are different kinds of dynamical systems, and we
cannot hope to have a similarity of proofs. In this paper, we concentrate on two-sided shifts. The case
of one-sided shifts is similar.
Let
A
be a non-empty finite set (called alphabet) with discrete topology and consider the set
AZ
,
which denotes the set of doubly-infinite sequences (
xi
)
iZ
where each
xi∈A
, with product topology. It
is compact and metrizable. The shift is the homeomorphism
σ
:
AZ→A
Z
given by
σ
(
x
)
i
=
xi+1
for all
iZ
. A subset
A⊂A
Z
is called
σ
-invariant if
σ
(
A
)
A
. The pair (
AZ
)forms a dynamical system
called a full shift. A subshift is a
σ
- invariant non-empty closed subset
X
of a full shift, together with
the restriction of
σ
to
X
. We denote the set of periods of all periodic points of
σ
in
X
by
P er
(
X
). We
call
P er
(
X
)the period set of
X
. A word
w
on
A
is a concatenation
w1w2...wk
, where each
wi∈A
and (
w
)
n
=
wr
whenever
nr
(
mod k
). A subshift
X
is said to be a subshift of finite type (SFT) if
X
=
XF
=
{x∈A
Z
:no word in
F
occurs in
x}
for some finite set of words
F
. A subshift
X⊂A
Z
is
called sofic if it is a factor of an SFT.
The notion of strongly connected digraphs is well known. For every SFT, there is an associated
digraph and an associated matrix as described in [7]. This may or may not be strongly connected. Let
G
=(
V,E
)be any directed graph (digraph) with vertex set
V
and edge set
E
. A subgraph
G
=(
V,E
)
of
G
is said to be a full subgraph if
E
=
E∩{
(
v1,v
2
):
v1,v
2V}
. A digraph is said to be simple if
from every vertex
v
to a vertex
w
there is atmost one edge and it is said to be strongly connected if for
every pair of vertices there exists a directed path. It is to be noted that a connected digraph may not
be strongly connected. The SFT associated for a digraph
G
is denoted as
XG
and the SFT associated
for an
m×m
matrix
A
with entries 0or 1is denoted as
XA
. Let
V
=
{vV
:there exists a cycle
through this
v}
. Consider a full subgraph of a simple digraph
G
with vertex set
V
, say
G
=(
V,E
).
Define a relation
R
on
V
in such a way that
xRy
if there is a directed path from
x
to
y
and from
y
to
x
. Then
R
is an equivalence relation on the set of vertices
V
. Let
G
x
be the full subgraph of
G
with [
x
], equivalence class of
x
, as the vertex set. This
G
x
is strongly connected simple digraph and
G
can be written as a finite union of such strongly connected simple digraphs, say
n
i=1 Gi
. The SFT
associated for a simple digraph Gis denoted as XG. Note that P er(XG)=P er(XG).
Let
A
be an alphabet having atleast two elements. Let
rN0
. A function
f
:
A2r+1 →A
is called a
local rule. It induces a function F:AZ→A
Zby the rule
(
F
(
x
))
n
=
f
(
xnr,x
nr1, ..., xn1,x
n,x
n+1, ..., xn+r1,x
n+r
)for all
nZ
. The pair (
AZ,F
)(sim-
ply the map
F
:
AZ→A
Z
) is called a cellular automaton (abbreviated as
CA
). A map
F
:
AZ→A
Z
is
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
15
a cellular automaton if and only if it is continuous and commutes with the shift (see [13]).
Our main results prove that:
1.
Let (
AZ,F
)be any
CA
. Then
F ix
(
F
)is an
SFT
or empty set. Conversely given any
SFT X
there exists a CA Fsuch that F ix(F)=X.
2. The following are equivalent for a subset Sof N.
a) S=Per(X)for some mixing subshift of finite type X.
b) Either S={1}or N\Ffor some finite subset Fof N.
3. a) The following are equivalent for a non-singleton SFT X.
i. Xis transitive and P er(X)is cofinite
ii. Xis weak mixing
iii. Xis mixing.
b)
In general, in the case of non-singleton sofic shifts, only the statements 3(a)ii and 3(a)iii are
equivalent.
The results are intuitive in nature and provide a basic understanding of the dynamics of shift spaces.
This review article will be useful to any reader interested in understanding basics of dynamics of shift
spaces.
2
SUBSHIFTS OF FINITE TYPE AND SOFIC SHIFTS: TRANSITIVITY, WEAK
MIXING, MIXING
One of the result in this paper is motivated by the following two known theorems.
Theorem 1. The following are equivalent for a topological graph map
f
:
GG
(See [5], [8] for
interval maps, See [2] for topological graph maps).
1. fis transitive and P er(f)is cofinite (ie., N\P er(f)is finite)
2. fis weak mixing
3. fis mixing.
Theorem 2. The following are equivalent for a continuous toral automorphism T:T2T2.
1. Tis transitive and P er(T)is cofinite (ie., N\P er(T)is finite)
2. Tis weak mixing
3. Tis mixing.
Proof. Proof follows from the main results of [16] and [17].
2.1 SUBSHIFTS OF FINITE TYPE
Let
A
be a
k×k
adjacency matrix (i.e., the matrix with entries 0or 1). We call the matrix primitive if
there exists NNsuch that AN>0. Now we consider the following known proposition.
Proposition 1. [7] An
SFT
induced by a matrix
A
with non-zero rows and columns is mixing if and
only if Ais primitive.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
Next we state the following known theorem. We denote any finite subset
A
of
N
as
A⊂⊂ N
, and
gcd for the greatest common divisor.
Theorem 3. [1] The following are equivalent for a subset Sof N.
1. S
=
Per
(
XG
)for some strongly connected simple digraph
G
containing cycles of lengths
m1, ..., mk
such that gcd(m1,m
2, ..., mk)=1.
2. Either S={1}or S=N\Ffor some F⊂⊂ N.
Next we have:
Theorem 4. A strongly connected simple digraph
G
induced by an adjacency matrix
Ak×k
with non
zero rows and columns contains cycles of lengths
m1,m
2, ..., mk
such that
gcd
(
m1,m
2, ..., mk
)=1if
and only if Ais primitive.
Proof. Suppose that
G
is a strongly connected simple digraph, and contains cycles of lengths
m1,m
2, ..., mk
such that
gcd
(
m1,m
2, ..., mk
)=1. Without loss of generality we can assume that
these are simple cycles of
G
that contain all vertices. By a basic number theory result, there exists
n0N
,
F⊂⊂ N
such that for all
nn0
,
nN\F
there exist
a1,a
2, ..., akN
such that
n
=
a1m1
+
a2m2
+
...
+
akmk
. Therefore, given any vertex
x
there exists a cycle of length
n
for all
nn0
. Let
m
=
diam
(
G
)=
Max{l
(
x, y
):
x, y V
(
G
)
}
where
l
(
x, y
)denotes the length of a directed
path from
x
to
y
. Let
N
=
n0
+
m
. Hence
AN>
0since for every
x, y V
(
G
)there exists a path of
length
p
for all
pm
and for every
xV
(
G
)there exists a cycle of length
n
for all
nn0
. Write
N=n0+mp+p. Hence Ais primitive.
Conversely, suppose that
A
is primitive and
gcd
(
m1,m
2, ..., ml
)=
p>
1for all cycles of length
mi
,1
ip
,
pN
. Let
k
=
gcd
of lengths of all cycles of
G
. Then there exist cycles of length
m1,m
2, ..., ml
such that
gcd
(
m1,m
2, ..., ml
)=
k
. Then
k
divides the lengths of all cycles. Also, there
exists
sN
such that
As>
0, and for every
x, y V
(
G
)there exists a path of length
s
from
x
to
y
since
A
is primitive and
G
is strongly connected. Therefore
k
divides
s
and
s
+1, which implies
k
=1.
A contradiction. Hence the proof.
Corollary 1. A strongly connected simple digraph
G
contains cycles of length
m1, ..., mk
such that
gcd(m1,m
2, ..., mn)=1if and only if XGis mixing.
Proof. This follows from Proposition 1, Theorems 3 and 4.
Corollary 2. The following are equivalent for a subset Sof N.
1. S=Per(X)for some mixing subshift of finite type X.
2. Either S={1}or N\Ffor some finite subset Fof N.
Proof. Proof follows from Theorems 3, 4, and Corollary 1.
The proof of the following theorem relies mostly on the proof of Proposition 1, as given in [7].
Lemma 1. An
SFT XA
is weak mixing if and only if for every 1
i1,j
1,i
2,j
2k
there exists
nN
such that An(i1,j
1)>0and An(i2,j
2)>0where Adenotes a k×kadjacency matrix.
Proof. Assume that XAis weak mixing. Let U1={(xn)nZXA:x0=i1},V1={(xn)nZXA:
x0
=
j1}
,
U2
=
{
(
yn
)
nZXA
:
y0
=
i2}
and
V2
=
{
(
yn
)
nZXA
:
y0
=
j2}
where 1
i1,j
1,i
2,j
2k
.
These sets are open. Then there exists
nN
such that
σn
(
Ui
)
Vi
=
,
i
=1
,
2. Hence
x0
=
i1,y
0
=
i2,x
n
=
j1
and
yn
=
j2
. Note that
AN
(
i, j
)=
k
r1=1 ... k
rN1=1 A
(
i, r1
)
A
(
r1,r
2
)
...A
(
rN2,r
N1
)
A
(
rN1,j
)
for all
NN
. But
A
(
i1,x
1
)=
A
(
x1,x
2
)=
A
(
x2,x
3
)=
...
=
A
(
xn1,j
1
)=1and
A
(
i2,y
1
)=
A(y1,y
2)=A(y2,y
3)=... =A(yn1,j
2)=1. Therefore An(i1,j
1)>0and An(i2,j
2)>0.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
16
a cellular automaton if and only if it is continuous and commutes with the shift (see [13]).
Our main results prove that:
1.
Let (
AZ,F
)be any
CA
. Then
F ix
(
F
)is an
SFT
or empty set. Conversely given any
SFT X
there exists a CA Fsuch that F ix(F)=X.
2. The following are equivalent for a subset Sof N.
a) S=Per(X)for some mixing subshift of finite type X.
b) Either S={1}or N\Ffor some finite subset Fof N.
3. a) The following are equivalent for a non-singleton SFT X.
i. Xis transitive and P er(X)is cofinite
ii. Xis weak mixing
iii. Xis mixing.
b)
In general, in the case of non-singleton sofic shifts, only the statements 3(a)ii and 3(a)iii are
equivalent.
The results are intuitive in nature and provide a basic understanding of the dynamics of shift spaces.
This review article will be useful to any reader interested in understanding basics of dynamics of shift
spaces.
2
SUBSHIFTS OF FINITE TYPE AND SOFIC SHIFTS: TRANSITIVITY, WEAK
MIXING, MIXING
One of the result in this paper is motivated by the following two known theorems.
Theorem 1. The following are equivalent for a topological graph map
f
:
GG
(See [5], [8] for
interval maps, See [2] for topological graph maps).
1. fis transitive and P er(f)is cofinite (ie., N\P er(f)is finite)
2. fis weak mixing
3. fis mixing.
Theorem 2. The following are equivalent for a continuous toral automorphism T:T2T2.
1. Tis transitive and P er(T)is cofinite (ie., N\P er(T)is finite)
2. Tis weak mixing
3. Tis mixing.
Proof. Proof follows from the main results of [16] and [17].
2.1 SUBSHIFTS OF FINITE TYPE
Let
A
be a
k×k
adjacency matrix (i.e., the matrix with entries 0or 1). We call the matrix primitive if
there exists NNsuch that AN>0. Now we consider the following known proposition.
Proposition 1. [7] An
SFT
induced by a matrix
A
with non-zero rows and columns is mixing if and
only if Ais primitive.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
Next we state the following known theorem. We denote any finite subset
A
of
N
as
A⊂⊂ N
, and
gcd for the greatest common divisor.
Theorem 3. [1] The following are equivalent for a subset Sof N.
1. S
=
Per
(
XG
)for some strongly connected simple digraph
G
containing cycles of lengths
m1, ..., mk
such that gcd(m1,m
2, ..., mk)=1.
2. Either S={1}or S=N\Ffor some F⊂⊂ N.
Next we have:
Theorem 4. A strongly connected simple digraph
G
induced by an adjacency matrix
Ak×k
with non
zero rows and columns contains cycles of lengths
m1,m
2, ..., mk
such that
gcd
(
m1,m
2, ..., mk
)=1if
and only if Ais primitive.
Proof. Suppose that
G
is a strongly connected simple digraph, and contains cycles of lengths
m1,m
2, ..., mk
such that
gcd
(
m1,m
2, ..., mk
)=1. Without loss of generality we can assume that
these are simple cycles of
G
that contain all vertices. By a basic number theory result, there exists
n0N
,
F⊂⊂ N
such that for all
nn0
,
nN\F
there exist
a1,a
2, ..., akN
such that
n
=
a1m1
+
a2m2
+
...
+
akmk
. Therefore, given any vertex
x
there exists a cycle of length
n
for all
nn0
. Let
m
=
diam
(
G
)=
Max{l
(
x, y
):
x, y V
(
G
)
}
where
l
(
x, y
)denotes the length of a directed
path from
x
to
y
. Let
N
=
n0
+
m
. Hence
AN>
0since for every
x, y V
(
G
)there exists a path of
length
p
for all
pm
and for every
xV
(
G
)there exists a cycle of length
n
for all
nn0
. Write
N=n0+mp+p. Hence Ais primitive.
Conversely, suppose that
A
is primitive and
gcd
(
m1,m
2, ..., ml
)=
p>
1for all cycles of length
mi
,1
ip
,
pN
. Let
k
=
gcd
of lengths of all cycles of
G
. Then there exist cycles of length
m1,m
2, ..., ml
such that
gcd
(
m1,m
2, ..., ml
)=
k
. Then
k
divides the lengths of all cycles. Also, there
exists
sN
such that
As>
0, and for every
x, y V
(
G
)there exists a path of length
s
from
x
to
y
since
A
is primitive and
G
is strongly connected. Therefore
k
divides
s
and
s
+1, which implies
k
=1.
A contradiction. Hence the proof.
Corollary 1. A strongly connected simple digraph
G
contains cycles of length
m1, ..., mk
such that
gcd(m1,m
2, ..., mn)=1if and only if XGis mixing.
Proof. This follows from Proposition 1, Theorems 3 and 4.
Corollary 2. The following are equivalent for a subset Sof N.
1. S=Per(X)for some mixing subshift of finite type X.
2. Either S={1}or N\Ffor some finite subset Fof N.
Proof. Proof follows from Theorems 3, 4, and Corollary 1.
The proof of the following theorem relies mostly on the proof of Proposition 1, as given in [7].
Lemma 1. An
SFT XA
is weak mixing if and only if for every 1
i1,j
1,i
2,j
2k
there exists
nN
such that An(i1,j
1)>0and An(i2,j
2)>0where Adenotes a k×kadjacency matrix.
Proof. Assume that XAis weak mixing. Let U1={(xn)nZXA:x0=i1},V1={(xn)nZXA:
x0
=
j1}
,
U2
=
{
(
yn
)
nZXA
:
y0
=
i2}
and
V2
=
{
(
yn
)
nZXA
:
y0
=
j2}
where 1
i1,j
1,i
2,j
2k
.
These sets are open. Then there exists
nN
such that
σn
(
Ui
)
Vi
=
,
i
=1
,
2. Hence
x0
=
i1,y
0
=
i2,x
n
=
j1
and
yn
=
j2
. Note that
AN
(
i, j
)=
k
r1=1 ... k
rN1=1 A
(
i, r1
)
A
(
r1,r
2
)
...A
(
rN2,r
N1
)
A
(
rN1,j
)
for all
NN
. But
A
(
i1,x
1
)=
A
(
x1,x
2
)=
A
(
x2,x
3
)=
...
=
A
(
xn1,j
1
)=1and
A
(
i2,y
1
)=
A(y1,y
2)=A(y2,y
3)=... =A(yn1,j
2)=1. Therefore An(i1,j
1)>0and An(i2,j
2)>0.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
17
Conversely, assume that for every 1
i1,j
1,i
2,j
2k
there exists
NN
such that
AN
(
i1,j
1
)
>
0and
AN
(
i2,j
2
)
>
0. Given non-empty open sets
U1,V
1,U
2,V
2
we can choose (
i(l)
n
)
nZUl
and (
j(l)
n
)
nZVl
such that for
M>
0sufficiently large; and
Ul⊃{
(
xn
)
nZXA
:
xk
=
i(l)
k,MkM}
,
Vl⊃{
(
xn
)
nZXA
:
xk
=
j(l)
k,MkM}
for
l
=1
,
2(It is possible since the set of symmetric
cylinders form a base for the topology on AZ).
By hypothesis, there exists
N>
0such that
AN
(
i(l)
M,j(l)
M
)
>
0for
l
=1
,
2. This means we can find
a word xl
1...xl
N1such that A(i(l)
M,x
l
1)=A(xl
1,x
l
2)=... =A(xl
N1,j(l)
M)=1.
Define x(l)
n=
i(l)
nif nM
xl
nMif M+1nM+N1
j(l)
n(2M+N)if M+Nn
Then σ2M+N(Ul)Vl=for l=1,2. Hence XAis weak mixing.
Next we have:
Theorem 5. An SFT is weak mixing if and only if it is mixing.
Proof. Let
A
be an adjacency matrix of order
k
. Assume that
XA
is weak mixing. We have to prove
that there exist cycles of lengths
m1,m
2, ..., mp
such that
gcd
(
m1,m
2, ..., mp
)=1. Suppose not. Then
there exists
sN\{
1
}
such that
s
divides the lengths of all cycles (let
s
=
gcd
of lengths all cycles).
Let 1
v1,w
1,v
2k
be such that
v1w1
is a block in
x
for some
xXA
. Then there exist a cycle of
length
n
through
v2
and a path of length
n
from
w1
to
v1
, which implies
s
divides
n
and
n
+1. Hence
s=1. A contradiction. Hence XAis mixing. Converse part is easy.
Hence we have:
Theorem 6. The following are equivalent for a non-singleton SFT XF.
(i)XFis transitive and Per(XF)is cofinite.
1. XFis weak mixing.
2. XFis mixing.
Proof. We first observe that the period set
Per
(
XF
)of a finite
SFT XF
is finite. So
Per
(
XF
)is not
cofinite. Except singleton
SFT
s all other finite
SFT
s are not weak mixing and hence not mixing. Hence
the theorem follows from Theorems 3, 4 and 5, and Proposition 1.
Remark 1. Suppose some rows of
A
or columns of
A
is of full of zeros, say
i
th row and
j
th column.
Then remove
i
th row and
j
th column. Doing this for all such
i
and
j
, we obtain another matrix
˜
A
of
smaller size. Then
XA
and
X˜
A
are in a sense one and the same. Therefore the equivalence of (2) and
(3) is true for all subshifts induced by adjacency matrices.
Next we consider:
Theorem 7. (see [15]) (Blokh, Barge-Martin) Let
f
:
II
be an interval map such that the periodic
points are dense in
I
. Then the interval
I
decomposes into transitive components
Cn
in the following
way.
1. Cn
is a closed non-degenerate interval or
Cn
is the union of two disjoint closed non degenerate
intervals,
2. f|Cnis transitive,
3. the complement set of Cnis included in {xX:f2(x)=x}.
In addition, the number of transitive components
Cn
is finite or countable and their interiors are
pairwise disjoint.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
As similar to Theorem 7, now we have:
Theorem 8. Let
XF
be an
SFT
for some finite set of words
F
over an alphabet
A
with dense set of
periodic points. Then there exists some finite set of words
G⊃F
, and an
SFT XG
with dense set of
periodic points and it is a finite union of transitive SFTs, and Per(XF)=Per(XG).
Proof. Let
XG
denotes the subshift of finite type associated for a simple digraph
G
such that
XG
=
XF
. Let
V
be the set of all vertices of
G
. For
v1,v
2V
, we say that
v1v2
whenever there is a
path from
v1
to
v2
and vice-versa. Then
forms an equivalence relation on
V
. Each equivalence class
corresponds to a strongly connected simple digraphs. Let
G
be the union of all such strongly connected
simple digraphs. Observe that
P er
(
XG
)=
P er
(
XG
). Now consider
G⊃F
such that
XG
=
XG
. Hence
the proof.
Consider a countable set
{
1
,
2
, ...}
. With the discrete topology it is a non-compact metrizable space.
Let
=
{
1
,
2
, ...}Z
. With product topology
is a totally disconnected, perfect and non-compact
metric space. As in the finite case, the cylinder sets form a countable basis of clopen sets. The shift,
σ
, is a homeomorphism of the space to itself. The dynamical system (
)is the full shift on the
symbols. If
A
is a countable, zero-one matrix, then as in the finite case, we use transition rules to
define a shift-invariant subset of the full shift on countably many symbols, denoted by
A
. Then the
subspace
A
of
is non-compact, metrizable and
σ
:
AA
is the countable state Markov shift
defined by A.
Next consider the following two known propositions.
Proposition 2. [12] A countable state Markov shift
A
is topologically transitive if and only if
A
is
irreducible.
Proposition 3. [12] A countable state Markov shift
A
is topologically mixing if and only if
A
is
primitive.
Now we have:
Theorem 9. [12] The following are equivalent for countable Markov shift σ:AA.
(i)σ:AAis transitive and Per(σ)is cofinite.
(ii)σ:AAis weak mixing.
(iii)σ:AAis mixing.
Proof. The proof follows from Propositions 2 and 3.
2.2 SOFIC SHIFTS
The notion of labeled digraph is well known (see [14], [7]). For every sofic shift there is a labeled digraph
and vice versa (see [7]). As similar to digraphs, we can define simple labeled digraph and strongly
connected labeled digraph. First we have to define it for corresponding digraphs. Then consider the
corresponding labeled digraphs. As similar to digraphs, for every labeled digraph Γthere exists another
labeled digraph Γ
such that
Per
(
XΓ
)=
Per
(
XΓ
)and Γ
is a finite union of strongly connected simple
labeled digraphs where
XΓ
denotes the subshift induced by Γ. Let Γbe a finite labeled digraph, the
edges of Γare labeled by an alphabet
Am
=
{
1
,
2
, ..., m}
. Note that we do not assume that the different
edges of Γare labeled differently. Let
E
(Γ) denotes the set of all edges of Γ. The subset
XΓ⊂A
Z
m
consisting of all infinite directed paths in Γis closed and shift invariant. If a subshift
X
is topologically
conjugate to XΓfor some labeled digraph Γ, then we say that Γis a presentation of X.
Proposition 4. [7] A subshift
X⊂A
Z
is sofic if and only if it admits a presentation by a finite
labeled digraph.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
18
Conversely, assume that for every 1
i1,j
1,i
2,j
2k
there exists
NN
such that
AN
(
i1,j
1
)
>
0and
AN
(
i2,j
2
)
>
0. Given non-empty open sets
U1,V
1,U
2,V
2
we can choose (
i(l)
n
)
nZUl
and (
j(l)
n
)
nZVl
such that for
M>
0sufficiently large; and
Ul⊃{
(
xn
)
nZXA
:
xk
=
i(l)
k,MkM}
,
Vl⊃{
(
xn
)
nZXA
:
xk
=
j(l)
k,MkM}
for
l
=1
,
2(It is possible since the set of symmetric
cylinders form a base for the topology on AZ).
By hypothesis, there exists
N>
0such that
AN
(
i(l)
M,j(l)
M
)
>
0for
l
=1
,
2. This means we can find
a word xl
1...xl
N1such that A(i(l)
M,x
l
1)=A(xl
1,x
l
2)=... =A(xl
N1,j(l)
M)=1.
Define x(l)
n=
i(l)
nif nM
xl
nMif M+1nM+N1
j(l)
n(2M+N)if M+Nn
Then σ2M+N(Ul)Vl=for l=1,2. Hence XAis weak mixing.
Next we have:
Theorem 5. An SFT is weak mixing if and only if it is mixing.
Proof. Let
A
be an adjacency matrix of order
k
. Assume that
XA
is weak mixing. We have to prove
that there exist cycles of lengths
m1,m
2, ..., mp
such that
gcd
(
m1,m
2, ..., mp
)=1. Suppose not. Then
there exists
sN\{
1
}
such that
s
divides the lengths of all cycles (let
s
=
gcd
of lengths all cycles).
Let 1
v1,w
1,v
2k
be such that
v1w1
is a block in
x
for some
xXA
. Then there exist a cycle of
length
n
through
v2
and a path of length
n
from
w1
to
v1
, which implies
s
divides
n
and
n
+1. Hence
s=1. A contradiction. Hence XAis mixing. Converse part is easy.
Hence we have:
Theorem 6. The following are equivalent for a non-singleton SFT XF.
(i)XFis transitive and Per(XF)is cofinite.
1. XFis weak mixing.
2. XFis mixing.
Proof. We first observe that the period set
Per
(
XF
)of a finite
SFT XF
is finite. So
Per
(
XF
)is not
cofinite. Except singleton
SFT
s all other finite
SFT
s are not weak mixing and hence not mixing. Hence
the theorem follows from Theorems 3, 4 and 5, and Proposition 1.
Remark 1. Suppose some rows of
A
or columns of
A
is of full of zeros, say
i
th row and
j
th column.
Then remove
i
th row and
j
th column. Doing this for all such
i
and
j
, we obtain another matrix
˜
A
of
smaller size. Then
XA
and
X˜
A
are in a sense one and the same. Therefore the equivalence of (2) and
(3) is true for all subshifts induced by adjacency matrices.
Next we consider:
Theorem 7. (see [15]) (Blokh, Barge-Martin) Let
f
:
II
be an interval map such that the periodic
points are dense in
I
. Then the interval
I
decomposes into transitive components
Cn
in the following
way.
1. Cn
is a closed non-degenerate interval or
Cn
is the union of two disjoint closed non degenerate
intervals,
2. f|Cnis transitive,
3. the complement set of Cnis included in {xX:f2(x)=x}.
In addition, the number of transitive components
Cn
is finite or countable and their interiors are
pairwise disjoint.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
As similar to Theorem 7, now we have:
Theorem 8. Let
XF
be an
SFT
for some finite set of words
F
over an alphabet
A
with dense set of
periodic points. Then there exists some finite set of words
G⊃F
, and an
SFT XG
with dense set of
periodic points and it is a finite union of transitive SFTs, and Per(XF)=Per(XG).
Proof. Let
XG
denotes the subshift of finite type associated for a simple digraph
G
such that
XG
=
XF
. Let
V
be the set of all vertices of
G
. For
v1,v
2V
, we say that
v1v2
whenever there is a
path from
v1
to
v2
and vice-versa. Then
forms an equivalence relation on
V
. Each equivalence class
corresponds to a strongly connected simple digraphs. Let
G
be the union of all such strongly connected
simple digraphs. Observe that
P er
(
XG
)=
P er
(
XG
). Now consider
G⊃F
such that
XG
=
XG
. Hence
the proof.
Consider a countable set
{
1
,
2
, ...}
. With the discrete topology it is a non-compact metrizable space.
Let
=
{
1
,
2
, ...}Z
. With product topology
is a totally disconnected, perfect and non-compact
metric space. As in the finite case, the cylinder sets form a countable basis of clopen sets. The shift,
σ
, is a homeomorphism of the space to itself. The dynamical system (
)is the full shift on the
symbols. If
A
is a countable, zero-one matrix, then as in the finite case, we use transition rules to
define a shift-invariant subset of the full shift on countably many symbols, denoted by
A
. Then the
subspace
A
of
is non-compact, metrizable and
σ
:
AA
is the countable state Markov shift
defined by A.
Next consider the following two known propositions.
Proposition 2. [12] A countable state Markov shift
A
is topologically transitive if and only if
A
is
irreducible.
Proposition 3. [12] A countable state Markov shift
A
is topologically mixing if and only if
A
is
primitive.
Now we have:
Theorem 9. [12] The following are equivalent for countable Markov shift σ:AA.
(i)σ:AAis transitive and Per(σ)is cofinite.
(ii)σ:AAis weak mixing.
(iii)σ:AAis mixing.
Proof. The proof follows from Propositions 2 and 3.
2.2 SOFIC SHIFTS
The notion of labeled digraph is well known (see [14], [7]). For every sofic shift there is a labeled digraph
and vice versa (see [7]). As similar to digraphs, we can define simple labeled digraph and strongly
connected labeled digraph. First we have to define it for corresponding digraphs. Then consider the
corresponding labeled digraphs. As similar to digraphs, for every labeled digraph Γthere exists another
labeled digraph Γ
such that
Per
(
XΓ
)=
Per
(
XΓ
)and Γ
is a finite union of strongly connected simple
labeled digraphs where
XΓ
denotes the subshift induced by Γ. Let Γbe a finite labeled digraph, the
edges of Γare labeled by an alphabet
Am
=
{
1
,
2
, ..., m}
. Note that we do not assume that the different
edges of Γare labeled differently. Let
E
(Γ) denotes the set of all edges of Γ. The subset
XΓ⊂A
Z
m
consisting of all infinite directed paths in Γis closed and shift invariant. If a subshift
X
is topologically
conjugate to XΓfor some labeled digraph Γ, then we say that Γis a presentation of X.
Proposition 4. [7] A subshift
X⊂A
Z
is sofic if and only if it admits a presentation by a finite
labeled digraph.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
19
Theorem 10. [1] The following are equivalent for a subset Sof N.
(1)
S
=
Per
(
XΓ
)for some strongly connected labeled digraph Γcontaining cycles of length
m1,m
2, ..., mk
such that gcd(m1,m
2, ..., mk)=1.
(2) Either S={1}or S=N\Ffor some F⊂⊂ N.
From the definition of transitivity, mixing, weak mixing and by using some ideas from Lemma 1, we
can prove the following lemma for any directed labeled graph Γ. Recall that for every non-empty open
set
U
in
XΓ
we can choose (
in
)
nZU
such that for
M>
0sufficiently large;
U⊃{
(
xn
)
nZ
:
xk
=
ik,MkM}.
Lemma 2. Let Γbe a labeled digraph. Then the following are true.
1. XΓ
is transitive if and only if for every
i, j E
(Γ) there exists a directed path of length
n
from
i
to jfor some nN.
2. XΓ
is weak mixing if and only if for every
i1,j
1,i
2,j
2E
(Γ) there exist directed paths of length
nfrom i1to j1and from i2to j2for some nN.
3. XΓ
is mixing if and only if for every
i, j E
(Γ) there exists
NN
such that for all
nN
there
is a directed path of length nfrom ito j.
Theorem 11. Let Γbe a strongly connected labeled digraph. Then Γcontains cycles of lengths
m1,m
2, ..., mksuch that gcd(m1,m
2, ..., mk)=1if and only if XΓis mixing.
Proof. We can provide a proof similar to that of Theorem 4. Here while proceeding the proof without
loss of generality it is not possible to assume the cycles are simple. Still the conclusion is true.
Corollary 3. The period set of a mixing SFT is
either {1}or N\Ffor some F⊂⊂ N.
Proof. Proof follows from Theorems 10 and 11.
Theorem 12. A sofic shift is weak mixing if and only if it is mixing.
Proof. Because of Lemma 2 and Theorem 11, we can provide a proof similar to that of Theorem 5.
Remark 2. There exists a sofic shift
XΓ
which is transitive and its period set is cofinite, but it is not
Mixing.
Proof. Let
XΓ
be the sofic shift based on the directed graph Γwith vertices 0and 1, arcs labeled
a, b, c
from 0to 1, and arcs labeled
a, b, d
from 1to 0. Then
XΓ
is the image of the topologically
transitive subshift of finite type, based on Γbut with distinctly labeled edges. The period set of
XΓ
is
N. But XΓis not topologically mixing by Theorems 11. Hence the remark follows.
Remark 3. If
XΓ
is a transitive non-singleton sofic shift, then the set of periodic points of
σ
in
XΓ
is
dense in
XΓ
. But a compact dynamical system which is totally transitive and has a dense set of periodic
points is weak mixing (See [4]). Therefore
XΓ
is totally transitive if and only if
XΓ
is weak mixing. In
general, for a subshift, the conclusion of Theorem 12 need not be true. There is a subshift which is weak
mixing but not mixing (Chacon shift, See [11]).
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3 A CHARACTERIZATION OF AN SFT
The cellular automata play an important role in various contexts such as computer graphics, parallel
computing and cell biology. It is natural to ask for a neat description of the sets of periodic points of
cellular automata, unfortunately we do not have a complete answer. There have been some papers that
discussed about the sets of periodic points for continuous self maps (See [3], [7], [9]). It is natural to
ask: Which sets will arise as the set of all periodic points of continuous self maps? This question is too
abstract. If we ask the same question in the class of some nice class of maps then we can expect a nice
answer. In this section, we consider in the case of
CA
. Characterization of the sets of periodic points
for a continuous self map of an interval is incomplete. J.-P. Delahaye gave partial results in this context
(see Propositions 5, 6). This is our first motivation for considering
CA
. We completely solved in the
case of a continuous 2-dimensional toral automorphism in [16] (see Theorem 13). This is our second
motivation for considering
CA
. In this section we give a partial answer in the case
CA
. Our result is
similar to the following propositions 5 and 6, and Theorem 13. It characterizes an
SFT
in terms of a
CA.
Proposition 5. [9] (i) The set of fixed points of a continuous function from [0
,
1] to [0
,
1] is a closed
subset of [0,1].
(ii) For every closed subset
F
of [0
,
1] there exists a continuous function
f
whose fixed point set is
F
.
Definition 1. A subset Fof [0,1] is symmetric if for x[0,1],1
2+xF1
2xF.
Proposition 6. [9] (
i
)The set of periodic points of period 1or 2of a continuous function from [0
,
1]
to [0,1] is a closed subset of [0,1].
(
ii
)For every symmetric closed subset of [0
,
1] there exists a continuous function from [0
,
1]
[0
,
1]
whose set of periodic points period 1or 2is F∪{1
2}.
Theorem 13. [16]
For any continuous toral automorphism
T
, the set
P
(
T
)of periodic points of
T
is one of the following:
1. Q1×Q1, where Q1denotes the set of all rational points in [0,1).
2. Srfor some rQ {∞}; where Sr={(x, y)T2:rx +yis rational }.
3. T2.
Definition 2. A dynamical system (
X, f
)has the shadowing property, if for any
ϵ>
0there exists
δ>
0such that any finite
δ
-chain is
ϵ
-shadowed by some point. A point
xXϵ
-shadows a finite
sequence
x0,x
1, ..., xn
, if for all
in
,
d
(
Fi
(
x
)
,x
i
)
. A (finite or infinite) sequence (
xn
)
n0
is a
δ-chain, if d(F(xn),x
n+1)for all n.
ie., ϵ>0,δ>0,x0, ..., xn,(i, d(F(xi),x
i+1) =⇒∃x, i, d(Fi(x),x
i)).
Definition 3. A dynamical system (X, f)is open, if f(U)is open for any open UX.
There are two distinct topological characterization of SFT known in literature as follows.
Theorem 14. [13] A subset X⊂A
Nis an SFT if and only if Xhas the shadowing property.
Theorem 15. [13] A subset X⊂A
Nis an SFT if and only if Xis an open subset of AN.
Next we have:
Lemma 3. For every
SFT X
, there exists a finite set of words
G
having odd length such that
X
=
XG
.
Proof. Let
X
be a
k
-step
SFT
. Then there exists a finite set of words
F
having length atmost
k
such that
X
=
XF
. If
k
is odd then consider
G
=
{xWk
(
AZ
):
y
is a subword of
x
for some
yF}
.
If kis even then consider G={xWk+1(AZ):yis a subword of xfor some yF}.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
20
Theorem 10. [1] The following are equivalent for a subset Sof N.
(1)
S
=
Per
(
XΓ
)for some strongly connected labeled digraph Γcontaining cycles of length
m1,m
2, ..., mk
such that gcd(m1,m
2, ..., mk)=1.
(2) Either S={1}or S=N\Ffor some F⊂⊂ N.
From the definition of transitivity, mixing, weak mixing and by using some ideas from Lemma 1, we
can prove the following lemma for any directed labeled graph Γ. Recall that for every non-empty open
set
U
in
XΓ
we can choose (
in
)
nZU
such that for
M>
0sufficiently large;
U⊃{
(
xn
)
nZ
:
xk
=
ik,MkM}.
Lemma 2. Let Γbe a labeled digraph. Then the following are true.
1. XΓ
is transitive if and only if for every
i, j E
(Γ) there exists a directed path of length
n
from
i
to jfor some nN.
2. XΓ
is weak mixing if and only if for every
i1,j
1,i
2,j
2E
(Γ) there exist directed paths of length
nfrom i1to j1and from i2to j2for some nN.
3. XΓ
is mixing if and only if for every
i, j E
(Γ) there exists
NN
such that for all
nN
there
is a directed path of length nfrom ito j.
Theorem 11. Let Γbe a strongly connected labeled digraph. Then Γcontains cycles of lengths
m1,m
2, ..., mksuch that gcd(m1,m
2, ..., mk)=1if and only if XΓis mixing.
Proof. We can provide a proof similar to that of Theorem 4. Here while proceeding the proof without
loss of generality it is not possible to assume the cycles are simple. Still the conclusion is true.
Corollary 3. The period set of a mixing SFT is
either {1}or N\Ffor some F⊂⊂ N.
Proof. Proof follows from Theorems 10 and 11.
Theorem 12. A sofic shift is weak mixing if and only if it is mixing.
Proof. Because of Lemma 2 and Theorem 11, we can provide a proof similar to that of Theorem 5.
Remark 2. There exists a sofic shift
XΓ
which is transitive and its period set is cofinite, but it is not
Mixing.
Proof. Let
XΓ
be the sofic shift based on the directed graph Γwith vertices 0and 1, arcs labeled
a, b, c
from 0to 1, and arcs labeled
a, b, d
from 1to 0. Then
XΓ
is the image of the topologically
transitive subshift of finite type, based on Γbut with distinctly labeled edges. The period set of
XΓ
is
N. But XΓis not topologically mixing by Theorems 11. Hence the remark follows.
Remark 3. If
XΓ
is a transitive non-singleton sofic shift, then the set of periodic points of
σ
in
XΓ
is
dense in
XΓ
. But a compact dynamical system which is totally transitive and has a dense set of periodic
points is weak mixing (See [4]). Therefore
XΓ
is totally transitive if and only if
XΓ
is weak mixing. In
general, for a subshift, the conclusion of Theorem 12 need not be true. There is a subshift which is weak
mixing but not mixing (Chacon shift, See [11]).
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3 A CHARACTERIZATION OF AN SFT
The cellular automata play an important role in various contexts such as computer graphics, parallel
computing and cell biology. It is natural to ask for a neat description of the sets of periodic points of
cellular automata, unfortunately we do not have a complete answer. There have been some papers that
discussed about the sets of periodic points for continuous self maps (See [3], [7], [9]). It is natural to
ask: Which sets will arise as the set of all periodic points of continuous self maps? This question is too
abstract. If we ask the same question in the class of some nice class of maps then we can expect a nice
answer. In this section, we consider in the case of
CA
. Characterization of the sets of periodic points
for a continuous self map of an interval is incomplete. J.-P. Delahaye gave partial results in this context
(see Propositions 5, 6). This is our first motivation for considering
CA
. We completely solved in the
case of a continuous 2-dimensional toral automorphism in [16] (see Theorem 13). This is our second
motivation for considering
CA
. In this section we give a partial answer in the case
CA
. Our result is
similar to the following propositions 5 and 6, and Theorem 13. It characterizes an
SFT
in terms of a
CA.
Proposition 5. [9] (i) The set of fixed points of a continuous function from [0
,
1] to [0
,
1] is a closed
subset of [0,1].
(ii) For every closed subset
F
of [0
,
1] there exists a continuous function
f
whose fixed point set is
F
.
Definition 1. A subset Fof [0,1] is symmetric if for x[0,1],1
2+xF1
2xF.
Proposition 6. [9] (
i
)The set of periodic points of period 1or 2of a continuous function from [0
,
1]
to [0,1] is a closed subset of [0,1].
(
ii
)For every symmetric closed subset of [0
,
1] there exists a continuous function from [0
,
1]
[0
,
1]
whose set of periodic points period 1or 2is F∪{1
2}.
Theorem 13. [16]
For any continuous toral automorphism
T
, the set
P
(
T
)of periodic points of
T
is one of the following:
1. Q1×Q1, where Q1denotes the set of all rational points in [0,1).
2. Srfor some rQ {∞}; where Sr={(x, y)T2:rx +yis rational }.
3. T2.
Definition 2. A dynamical system (
X, f
)has the shadowing property, if for any
ϵ>
0there exists
δ>
0such that any finite
δ
-chain is
ϵ
-shadowed by some point. A point
x
-shadows a finite
sequence
x0,x
1, ..., xn
, if for all
in
,
d
(
Fi
(
x
)
,x
i
)
. A (finite or infinite) sequence (
xn
)
n0
is a
δ-chain, if d(F(xn),x
n+1)for all n.
ie., ϵ>0,δ>0,x0, ..., xn,(i, d(F(xi),x
i+1) =⇒∃x, i, d(Fi(x),x
i)).
Definition 3. A dynamical system (X, f)is open, if f(U)is open for any open UX.
There are two distinct topological characterization of SFT known in literature as follows.
Theorem 14. [13] A subset X⊂A
Nis an SFT if and only if Xhas the shadowing property.
Theorem 15. [13] A subset X⊂A
Nis an SFT if and only if Xis an open subset of AN.
Next we have:
Lemma 3. For every
SFT X
, there exists a finite set of words
G
having odd length such that
X
=
XG
.
Proof. Let
X
be a
k
-step
SFT
. Then there exists a finite set of words
F
having length atmost
k
such that
X
=
XF
. If
k
is odd then consider
G
=
{xWk
(
AZ
):
y
is a subword of
x
for some
yF}
.
If kis even then consider G={xWk+1(AZ):yis a subword of xfor some yF}.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
21
Claim: XF=XG.
Let
xXF
. Suppose
x/XG
. Then for some
y∈F
,
y
is a subword of
x
. A contradiction. Therefore
xXG
. Next, let
xXG
. Which implies
y
is not a subword of
x
for all
y∈F
. Then
xXF
. Hence
the claim.
Next we have:
Theorem 16. Let (
AZ,F
)be any
CA
. Then
F ix
(
F
)is an
SFT
or an empty set. Conversely given any
SFT Xthere exists a CA Fsuch that F ix(F)=X.
Proof. Let (
AZ,F
)be a
CA
defined by the local rule
f
:
A2r+1 →A
. Let
F
=
{w∈A
2r+1
:
f
(
w
)
=the middle term of
w}
. Note that
F
is finite (it may be empty or
A2r+1
). First, let
xXF
.
Which implies (
F
(
x
))
i
=
xi
for all
i
. ie.,
F
(
x
)=
x
. Next, let
x∈A
Z
such that
F
(
x
)=
x
. Then
f(xirxir+1...x0...xi+r1xi+r)=xifor all i. ie., xXF. Hence F ix(F)=XF.
Conversely, given any
SFT XF
without loss of generality assume that
F
contains words of same
length (odd length) because of Lemma 3.
Define f:A2r+1 Asuch that
f(w)=the middle term of w if w is forbidden
some other alphabet otherwise
Then F ix(F)=XF.
Remark 4. In the statement of Theorem 16, we can replace F ix(F)by F ix(Fn).
Proof. Let
f
:
A2r+1 A
be the local rule of a
CA F
:
AZ→A
Z
. The local rule
f
:
A2r+1 →A
induces a function
˜
f
:
A2s+2r+1 →A
2s+1
for all
s
. Then by inductively, define
fn
:
A2nr+1 →A
such
that
fn
(
w
)=
f
(
˜
fn1
(
w
)) where
˜
fm
:
A2r+2(m1)r+1 →A
2r+1
denotes the induced function of
fm
for
s
=
r
. Note that the length of
˜
f
(
w
)is equal to the difference between the length of
w
and 2
r
. Let
Fn={w:fn(w)=the middle term of w}. Then F ix(Fn)=XFn.
Converse part follows easily.
Remark 5. Let (
AZ,F
)be any
CA
. Then
F ix
(
Fn
)is an
SFT
for each
nN
. Conversely, given any
SFT X there exists CA Fn:AZ→A
Zsuch that F ix(Fn
n)=X.
4 CONCLUSIONS
For each self-map
f
on a set
X
, we associate a subset of
N
namely,
P er
(
f
). If
f
belongs to a certain
nice class of functions, then not all subsets of
N
may arise as the set of periods. Which subsets of N
will come in this class? We answer this question for mixing SFT?s, and for mixing sofic shifts.
It is natural to ask: On which class of dynamical systems the following statements are equivalent?
1. fis transitive and P er(f)is cofinite.
2. fis weak mixing
3. fis mixing.
We have proved that the above statements are equivalent in the case of non-singleton SFT’s but not
true in the case of sofic shifts. Also we have obtained a characterization for
SFT
’s in terms of Cellular
automata.
ACKNOWLEDGMENT
The author is very thankful to the referee for giving valuable suggestions.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
REFERENCES
[1]
K. Ali Akbar and V. Kannan (2018), Set of periods of subshifts, Pro. Indian Acad. Sci. (Math.
Sci.), 128:63.
[2]
LL. Alseda,M.A. Del Rio and J. A. Rodrigues (2003), A survey on the Relation Between
Transitivity and Dense periodicity for graph maps, Journal of Difference equations and Applications,
Vol.9(3/4), 281-288.
[3]
I.N. Baker (1964), Fixpoints of polynomials and rational functions, J. London Math. Soc., 39,
615-622.
[4]
J. Banks (1997), Regular Periodic decomposition for topologically transitive maps, Ergodic Theory
and Dynamical Systems, 17, 505-529.
[5]
L.S. Block and W.A. Coppel (1992), Dynamics in One Dimension, Springer-Verlag, Berline,
Volume 1513 of Lecture Notes in Mathematics.
[6]
J.A. Bondy and U.S.R. Murty (1982), Graph theory with application, Elsevier science publishing
Co., Inc, New York.
[7]
M. Brin and G. Stuck (2002), Introduction to Dynamical Systems, Cambridge University Press.
[8]
E.M. Coven and M.C. Hidalgo (1991), On the topological entropy of transitive maps of the
interval, Bull. Aust. Math. Soc., 44, 207-213.
[9] J.-P. Delahaye (1981), The set of periodic points, Amer. Math. Monthly., 88, 646-651.
[10]
R.L. Devaney(1989), An Introduction to Chaotic Dynamical Systems, Addison-wesley Publishing
Company Advanced Book Program., Redwood City, CA, second edition.
[11]
Eli Glasner (2003), Ergodic Theory Via Joinigs, American Mathematical Society, Mathematical
surveys and monographs, Vol 101.
[12] B.P. Kitchens (1997), Symbolic Dynamics, Springer.
[13] P. Kurka (2003), Topological and symbolic dynamics, Societe Mathematique de France, Paris.
[14]
D.A. Lind and B. Marcus (1995), An introduction to symbolic dynamics and coding, Cambridge
University Press., New York, NY, USA.
[15]
Sylvie Ruette (2003), Chaos for continuous interval maps, A survey of relationship between the
various forms of chaos.
[16]
I. Subramania Pillai,K. Ali Akbar,V. Kannan and B. Sankararao (2010), The set of
periodic points of a toral automorphism, Journal of Mathematical Analysis and Applications., 366,
367-371.
[17]
I. Subramania Pillai,K. Ali Akbar,V. Kannan and B. Sankararao (2011), The set of
periods of periodic points of a toral automorphism, Topology Procedings, Volume 37, 1-14.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
22
Claim: XF=XG.
Let
xXF
. Suppose
x/XG
. Then for some
y∈F
,
y
is a subword of
x
. A contradiction. Therefore
xXG
. Next, let
xXG
. Which implies
y
is not a subword of
x
for all
y∈F
. Then
xXF
. Hence
the claim.
Next we have:
Theorem 16. Let (
AZ,F
)be any
CA
. Then
F ix
(
F
)is an
SFT
or an empty set. Conversely given any
SFT Xthere exists a CA Fsuch that F ix(F)=X.
Proof. Let (
AZ,F
)be a
CA
defined by the local rule
f
:
A2r+1 →A
. Let
F
=
{w∈A
2r+1
:
f
(
w
)
=the middle term of
w}
. Note that
F
is finite (it may be empty or
A2r+1
). First, let
xXF
.
Which implies (
F
(
x
))
i
=
xi
for all
i
. ie.,
F
(
x
)=
x
. Next, let
x∈A
Z
such that
F
(
x
)=
x
. Then
f(xirxir+1...x0...xi+r1xi+r)=xifor all i. ie., xXF. Hence F ix(F)=XF.
Conversely, given any
SFT XF
without loss of generality assume that
F
contains words of same
length (odd length) because of Lemma 3.
Define f:A2r+1 Asuch that
f(w)=the middle term of w if w is forbidden
some other alphabet otherwise
Then F ix(F)=XF.
Remark 4. In the statement of Theorem 16, we can replace F ix(F)by F ix(Fn).
Proof. Let
f
:
A2r+1 A
be the local rule of a
CA F
:
AZ→A
Z
. The local rule
f
:
A2r+1 →A
induces a function
˜
f
:
A2s+2r+1 →A
2s+1
for all
s
. Then by inductively, dene
fn
:
A2nr+1 →A
such
that
fn
(
w
)=
f
(
˜
fn1
(
w
)) where
˜
fm
:
A2r+2(m1)r+1 →A
2r+1
denotes the induced function of
fm
for
s
=
r
. Note that the length of
˜
f
(
w
)is equal to the difference between the length of
w
and 2
r
. Let
Fn={w:fn(w)=the middle term of w}. Then F ix(Fn)=XFn.
Converse part follows easily.
Remark 5. Let (
AZ,F
)be any
CA
. Then
F ix
(
Fn
)is an
SFT
for each
nN
. Conversely, given any
SFT X there exists CA Fn:AZ→A
Zsuch that F ix(Fn
n)=X.
4 CONCLUSIONS
For each self-map
f
on a set
X
, we associate a subset of
N
namely,
P er
(
f
). If
f
belongs to a certain
nice class of functions, then not all subsets of
N
may arise as the set of periods. Which subsets of N
will come in this class? We answer this question for mixing SFT?s, and for mixing sofic shifts.
It is natural to ask: On which class of dynamical systems the following statements are equivalent?
1. fis transitive and P er(f)is cofinite.
2. fis weak mixing
3. fis mixing.
We have proved that the above statements are equivalent in the case of non-singleton SFT’s but not
true in the case of sofic shifts. Also we have obtained a characterization for
SFT
’s in terms of Cellular
automata.
ACKNOWLEDGMENT
The author is very thankful to the referee for giving valuable suggestions.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
REFERENCES
[1]
K. Ali Akbar and V. Kannan (2018), Set of periods of subshifts, Pro. Indian Acad. Sci. (Math.
Sci.), 128:63.
[2]
LL. Alseda,M.A. Del Rio and J. A. Rodrigues (2003), A survey on the Relation Between
Transitivity and Dense periodicity for graph maps, Journal of Difference equations and Applications,
Vol.9(3/4), 281-288.
[3]
I.N. Baker (1964), Fixpoints of polynomials and rational functions, J. London Math. Soc., 39,
615-622.
[4]
J. Banks (1997), Regular Periodic decomposition for topologically transitive maps, Ergodic Theory
and Dynamical Systems, 17, 505-529.
[5]
L.S. Block and W.A. Coppel (1992), Dynamics in One Dimension, Springer-Verlag, Berline,
Volume 1513 of Lecture Notes in Mathematics.
[6]
J.A. Bondy and U.S.R. Murty (1982), Graph theory with application, Elsevier science publishing
Co., Inc, New York.
[7]
M. Brin and G. Stuck (2002), Introduction to Dynamical Systems, Cambridge University Press.
[8]
E.M. Coven and M.C. Hidalgo (1991), On the topological entropy of transitive maps of the
interval, Bull. Aust. Math. Soc., 44, 207-213.
[9] J.-P. Delahaye (1981), The set of periodic points, Amer. Math. Monthly., 88, 646-651.
[10]
R.L. Devaney(1989), An Introduction to Chaotic Dynamical Systems, Addison-wesley Publishing
Company Advanced Book Program., Redwood City, CA, second edition.
[11]
Eli Glasner (2003), Ergodic Theory Via Joinigs, American Mathematical Society, Mathematical
surveys and monographs, Vol 101.
[12] B.P. Kitchens (1997), Symbolic Dynamics, Springer.
[13] P. Kurka (2003), Topological and symbolic dynamics, Societe Mathematique de France, Paris.
[14]
D.A. Lind and B. Marcus (1995), An introduction to symbolic dynamics and coding, Cambridge
University Press., New York, NY, USA.
[15]
Sylvie Ruette (2003), Chaos for continuous interval maps, A survey of relationship between the
various forms of chaos.
[16]
I. Subramania Pillai,K. Ali Akbar,V. Kannan and B. Sankararao (2010), The set of
periodic points of a toral automorphism, Journal of Mathematical Analysis and Applications., 366,
367-371.
[17]
I. Subramania Pillai,K. Ali Akbar,V. Kannan and B. Sankararao (2011), The set of
periods of periodic points of a toral automorphism, Topology Procedings, Volume 37, 1-14.
https://doi.org/10.17993/3ctecno.2022.v11n2e42.13-23
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254-4143
Ed. 42 Vol. 11 N.º 2 August - December 2022
23