Shift space

Last updated September 09, 2023

In symbolic dynamics and related branches of mathematics, a shift space or subshift is a set of infinite words that represent the evolution of a discrete system. In fact, shift spaces and symbolic dynamical systems are often considered synonyms. The most widely studied shift spaces are the subshifts of finite type and the sofic shifts.

In the classical framework ^[1] a shift space is any subset $\Lambda$ of $A^{\mathbb {Z} }:=\{(x_{i})_{i\in \mathbb {Z} }:\ x_{i}\in A\ \forall i\in \mathbb {Z} \}$ , where $A$ is a finite set, which is closed for the Tychonov topology and invariant by translations. More generally one can define a shift space as the closed and translation-invariant subsets of $A^{\mathbb {G} }$ , where $A$ is any non-empty set and $\mathbb {G}$ is any monoid.^[2]^[3]

Definition

Let $\mathbb {G}$ be a monoid, and given $g,h\in \mathbb {G}$ , denote the operation of $g$ with $h$ by the product $gh$ . Let $\mathbf {1} _{\mathbb {G} }$ denote the identity of $\mathbb {G}$ . Consider a non-empty set $A$ (an alphabet) with the discrete topology, and define $A^{\mathbb {G} }$ as the set of all patterns over $A$ indexed by $\mathbb {G}$ . For $\mathbf {x} =(x_{i})_{i\in \mathbb {G} }\in A^{\mathbb {G} }$ and a subset $N\subset \mathbb {G}$ , we denote the restriction of $\mathbf {x}$ to the indices of $N$ as $\mathbf {x} _{N}:=(x_{i})_{i\in N}$ .

On $A^{\mathbb {G} }$ , we consider the prodiscrete topology, which makes $A^{\mathbb {G} }$ a Hausdorff and totally disconnected topological space. In the case of $A$ being finite, it follows that $A^{\mathbb {G} }$ is compact. However, if $A$ is not finite, then $A^{\mathbb {G} }$ is not even locally compact.

This topology will be metrizable if and only if $\mathbb {G}$ is countable, and, in any case, the base of this topology consists of a collection of open/closed sets (called cylinders), defined as follows: given a finite set of indices $D\subset \mathbb {G}$ , and for each $i\in D$ , let $a_{i}\in A$ . The cylinder given by $D$ and $(a_{i})_{i\in D}\in A^{|D|}$ is the set

 ${\big [}(a_{i})_{i\in D}{\big ]}_{D}:=\{\mathbf {x} \in A^{\mathbb {G} }:\ x_{i}=a_{i},\ \forall i\in D\}.$

When $D=\{g\}$ , we denote the cylinder fixing the symbol $b$ at the entry indexed by $g$ simply as $[b]_{g}$ .

In other words, a cylinder ${\big [}(a_{i})_{i\in D}{\big ]}_{D}$ is the set of all set of all infinite patterns of $A^{\mathbb {G} }$ which contain the finite pattern $(a_{i})_{i\in D}\in A^{|D|}$ .

Given $g\in \mathbb {G}$ , the g-shift map on $A^{\mathbb {G} }$ is denoted by $\sigma ^{g}:A^{\mathbb {G} }\to A^{\mathbb {G} }$ and defined as

 $\sigma ^{g}{\big (}(x_{i})_{i\in \mathbb {G} }{\big )}=(x_{gi})_{i\in \mathbb {G} }$ .

A shift space over the alphabet $A$ is a set $\Lambda \subset A^{\mathbb {G} }$ that is closed under the topology of $A^{\mathbb {G} }$ and invariant under translations, i.e., $\sigma ^{g}(\Lambda )\subset \Lambda$ for all $g\in \mathbb {G}$ .^{[note 1]} We consider in the shift space $\Lambda$ the induced topology from $A^{\mathbb {G} }$ , which has as basic open sets the cylinders ${\big [}(a_{i})_{i\in D}{\big ]}_{\Lambda }:={\big [}(a_{i})_{i\in D}{\big ]}\cap \Lambda$ .

For each $k\in \mathbb {N} ^{*}$ , define ${\mathcal {N}}_{k}:=\bigcup _{N\subset \mathbb {G} \atop \#N=k}A^{N}$ , and ${\mathcal {N}}_{A^{\mathbb {G} }}^{f}:=\bigcup _{k\in \mathbb {N} }{\mathcal {N}}_{k}=\bigcup _{N\subset \mathbb {G} \atop \#N<\infty }A^{N}$ . An equivalent way to define a shift space is to take a set of forbidden patterns $F\subset {\mathcal {N}}_{A^{\mathbb {G} }}^{f}$ and define a shift space as the set

 $X_{F}:=\{\mathbf {x} \in A^{\mathbb {G} }:\ \forall N\subset \mathbb {G} ,\forall g\in \mathbb {G} ,\ \left(\sigma ^{g}(\mathbf {x} )\right)_{N}=\mathbf {x} _{gN}\notin F\}.$

Intuitively, a shift space $X_{F}$ is the set of all infinite patterns that do not contain any forbidden finite pattern of $F$ .

Language of shift space

Given a shift space $\Lambda \subset A^{\mathbb {G} }$ and a finite set of indices $N\subset \mathbb {G}$ , let $W_{\emptyset }(\Lambda ):=\{\epsilon \}$ , where $\epsilon$ stands for the empty word, and for $N\neq \emptyset$ let $W_{N}(\Lambda )\subset A^{N}$ be the set of all finite configurations of $A^{N}$ that appear in some sequence of $\Lambda$ , i.e.,

 $W_{N}(\Lambda ):=\{(w_{i})_{i\in N}\in A^{N}:\ \exists \ \mathbf {x} \in \Lambda {\text{ s.t. }}x_{i}=w_{i}\ \forall i\in N\}.$

Note that, since $\Lambda$ is a shift space, if $M\subset \mathbb {G}$ is a translation of $N\subset \mathbb {G}$ , i.e., $M=gN$ for some $g\in \mathbb {G}$ , then $(w_{j})_{j\in M}\in W_{M}(\Lambda )$ if and only if there exists $(v_{i})_{i\in N}\in W_{N}(\Lambda )$ such that $w_{j}=v_{i}$ if $j=gi$ . In other words, $W_{M}(\Lambda )$ and $W_{N}(\Lambda )$ contain the same configurations modulo translation. We will call the set

 $W(\Lambda ):=\bigcup _{N\subset \mathbb {G}  \atop \#N<\infty }W_{N}(\Lambda )$

the language of $\Lambda$ . In the general context stated here, the language of a shift space has not the same mean of that in Formal Language Theory, but in the classical framework which considers the alphabet $A$ being finite, and $\mathbb {G}$ being $\mathbb {N}$ or $\mathbb {Z}$ with the usual addition, the language of a shift space is a formal language.

Classical framework

The classical framework for shift spaces consists of considering the alphabet $A$ as finite, and $\mathbb {G}$ as the set of non-negative integers ( $\mathbb {N}$ ) with the usual addition, or the set of all integers ( $\mathbb {Z}$ ) with the usual addition. In both cases, the identity element $\mathbf {1} _{\mathbb {G} }$ corresponds to the number 0. Furthermore, when $\mathbb {G} =\mathbb {N}$ , since all $\mathbb {N} \setminus \{0\}$ can be generated from the number 1, it is sufficient to consider a unique shift map given by $\sigma (\mathbf {x} )_{n}=x_{n+1}$ for all $n$ . On the other hand, for the case of $\mathbb {G} =\mathbb {Z}$ , since all $\mathbb {Z}$ can be generated from the numbers {-1, 1}, it is sufficient to consider two shift maps given for all $n$ by $\sigma (\mathbf {x} )_{n}=x_{n+1}$ and by $\sigma ^{-1}(\mathbf {x} )_{n}=x_{n-1}$ .

Furthermore, whenever $\mathbb {G}$ is $\mathbb {N}$ or $\mathbb {Z}$ with the usual addition (independently of the cardinlaity of $A$ ), due to its algebraic structure, it is sufficient consider only cylinders in the form

 $[a_{0}a_{1}...a_{n}]:=\{(x_{i})_{i\in \mathbb {G} }:\ x_{i}=a_{i}\ \forall i=0,..,n\}.$

Moreover, the language of a shift space $\Lambda \subset A^{\mathbb {G} }$ will be given by

 $W(\Lambda ):=\bigcup _{n\geq 0}W_{n}(\Lambda ),$

where $W_{0}:=\{\epsilon \}$ and $\epsilon$ stands for the empty word, and

 $W_{n}(\Lambda ):=\{((a_{i})_{i=0,..n}\in A^{n}:\ \exists \mathbf {x} \in \Lambda \ s.t.\ x_{i}=a_{i}\ \forall i=0,...,n\}.$

In the same way, for the particular case of $\mathbb {G} =\mathbb {Z}$ , it follows that to define a shift space $\Lambda =X_{F}$ we do not need to specify the index of $\mathbb {G}$ on which the forbidden words of $F$ are defined, that is, we can just consider $F\subset \bigcup _{n\geq 1}A^{n}$ and then

 $X_{F}=\{\mathbb {x} \in A^{\mathbb {Z} }:\ \forall i\in \mathbb {Z} ,\ \forall k\geq 0,\ (x_{i}...x_{i+k})\notin F\}.$

However, if $\mathbb {G} =\mathbb {N}$ , if we define a shift space $\Lambda =X_{F}$ as above, without to specify the index of where the words are forbidden, then we will just capture shift spaces which are invariant through the shift map, that is, such that $\sigma (X_{F})=X_{F}$ . In fact, to define a shift space $X_{F}\subset A^{\mathbb {N} }$ such that $\sigma (X_{F})\subsetneq X_{F}$ it will be necessary to specify from which index on the words of $F$ are forbidden.

In particular, in the classical framework of $A$ being finite, and $\mathbb {G}$ being $\mathbb {N}$ ) or $\mathbb {Z}$ with the usual addition, it follows that $M_{F}$ is finite if and only if $F$ is finite, which leds to classical definition of a shift of finite type as those shift spaces $\Lambda \subset A^{\mathbb {G} }$ such that $\Lambda =X_{F}$ for some finite $F$ .

Some types of shift spaces

Among several types of shift spaces, the most widely studied are the shifts of finite type and the sofic shifts.

In the case when the alphabet $A$ is finite, a shift space $\Lambda$ is a shift of finite type if we can take a finite set of forbidden patterns $F$ such that $\Lambda =X_{F}$ , and $\Lambda$ is a sofic shift if it is the image of a shift of finite type under sliding block code ^[1] (that is, a map $\Phi$ that is continuous and invariant for all $g$ -shift maps ). If $A$ is finite and $\mathbb {G}$ is $\mathbb {N}$ or $\mathbb {Z}$ with the usual addition, then the shift $\Lambda$ is a sofic shift if and only if $W(\Lambda )$ is a regular language.

The name "sofic" was coined by Weiss (1973), based on the Hebrew word סופי meaning "finite", to refer to the fact that this is a generalization of a finiteness property.^[4]

When $A$ is infinite, it is possible to define shifts of finite type as shift spaces $\Lambda$ for those one can take a set $F$ of forbidden words such that

 $M_{F}:=\{g\in \mathbb {G} :\ \exists N\subset \mathbb {G} {\text{ s.t. }}g\in N{\text{ and }}(w_{i})_{i\in N}\in F\},$

is finite and $\Lambda =X_{F}$ .^[3] In this context of infinite alphabet, a sofic shift will be defined as the image of a shift of finite type under a particular class of sliding block codes.^[3] Both, the finiteness of $M_{F}$ and the additional conditions the sliding block codes, are trivially satisfied whenever $A$ is finite.

Topological dynamical systems on shift spaces

Shift spaces are the topological spaces on which symbolic dynamical systems are usually defined.

Given a shift space $\Lambda \subset A^{\mathbb {G} }$ and a $g$ -shift map $\sigma ^{g}:\Lambda \to \Lambda$ it follows that the pair $(\Lambda ,\sigma ^{g})$ is a topological dynamical system.

Two shift spaces $\Lambda \subset A^{\mathbb {G} }$ and $\Gamma \subset B^{\mathbb {G} }$ are said to be topologically cojugate (or simply conjugate) if for each $g$ -shift map it follows that the topological dynamical systems $(\Lambda ,\sigma ^{g})$ and $(\Gamma ,\sigma ^{g})$ are topologically cojugate, that is, if there exists a continuous map ${\displaystyle \Phi$ such that $\Phi \circ \sigma ^{g}=\sigma ^{g}\circ \Phi$ . Such maps are known as generalized sliding block codes or just as sliding block codes whenever $\Phi$ is uniformly continuous.^[3]

Although any continuous map $\Phi$ from $\Lambda \subset A^{\mathbb {G} }$ to itself will define a topological dynamical system $(\Lambda ,\Phi )$ , in symbolic dynamics it is usual to consider only continuous maps ${\displaystyle \Phi$ which commute with all $g$ -shift maps, i. e., maps which are generalized sliding block codes. The dynamical system $(\Lambda ,\Phi )$ is known as a ' generalized cellular automaton' (or just as a cellular automaton whenever $\Phi$ is uniformly continuous).

Examples

The first trivial example of shift space (of finite type) is the full shift $A^{\mathbb {N} }$ .

Let $A=\{a,b\}$ . The set of all infinite words over A containing at most one b is a sofic subshift, not of finite type. The set of all infinite words over A whose b form blocks of prime length is not sofic (this can be shown by using the pumping lemma).

The space of infinite strings in two letters, $\{0,1\}^{\mathbb {N} }$ is called the Bernoulli process. It is isomorphic to the Cantor set.

The bi-infinite space of strings in two letters, $\{0,1\}^{\mathbb {Z} }$ is commonly known as the Baker's map, or rather is homomorphic to the Baker's map.

Footnotes

↑ It is common to reffer to a shift space using just the expression shift or subshift. However, some authors use the terms shift and subshift for sets of infinite parterns that are just invariant under the $g$ -shift maps, and reserve the term shift space for those that are also closed for the prodiscrete topology.

Related Research Articles

In mathematics, any vector space has a corresponding dual vector space consisting of all linear forms on together with the vector space structure of pointwise addition and scalar multiplication by constants.

In mathematical logic, model theory is the study of the relationship between formal theories, and their models. The aspects investigated include the number and size of models of a theory, the relationship of different models to each other, and their interaction with the formal language itself. In particular, model theorists also investigate the sets that can be defined in a model of a theory, and the relationship of such definable sets to each other. As a separate discipline, model theory goes back to Alfred Tarski, who first used the term "Theory of Models" in publication in 1954. Since the 1970s, the subject has been shaped decisively by Saharon Shelah's stability theory.

In mathematical analysis and in probability theory, a σ-algebra on a set X is a nonempty collection Σ of subsets of X closed under complement, countable unions, and countable intersections. The ordered pair $is called a measurable space.$

In mathematics, the $L p$ spaces are function spaces defined using a natural generalization of the $p$ -norm for finite-dimensional vector spaces. They are sometimes called Lebesgue spaces, named after Henri Lebesgue, although according to the Bourbaki group they were first introduced by Frigyes Riesz.

In mathematics, the unitary group of degree n, denoted U(n), is the group of n × n unitary matrices, with the group operation of matrix multiplication. The unitary group is a subgroup of the general linear group GL(n, C). Hyperorthogonal group is an archaic name for the unitary group, especially over finite fields. For the group of unitary matrices with determinant 1, see Special unitary group.

In mathematics, a self-adjoint operator on an infinite-dimensional complex vector space V with inner product $is a linear map A that is its own adjoint. If V is finite-dimensional with a given orthonormal basis, this is equivalent to the condition that the matrix of A is a Hermitian matrix, i.e., equal to its conjugate transpose A * . By the finite-dimensional spectral theorem, V has an orthonormal basis such that the matrix of A relative to this basis is a diagonal matrix with entries in the real numbers. This article deals with applying generalizations of this concept to operators on Hilbert spaces of arbitrary dimension.$

In mathematics, particularly in functional analysis, the spectrum of a bounded linear operator is a generalisation of the set of eigenvalues of a matrix. Specifically, a complex number $is said to be in the spectrum of a bounded linear operator if$

In mathematics, the adele ring of a global field is a central object of class field theory, a branch of algebraic number theory. It is the restricted product of all the completions of the global field and is an example of a self-dual topological ring.

<span class="mw-page-title-main">Lattice model (physics)</span>

In mathematical physics, a lattice model is a mathematical model of a physical system that is defined on a lattice, as opposed to a continuum, such as the continuum of space or spacetime. Lattice models originally occurred in the context of condensed matter physics, where the atoms of a crystal automatically form a lattice. Currently, lattice models are quite popular in theoretical physics, for many reasons. Some models are exactly solvable, and thus offer insight into physics beyond what can be learned from perturbation theory. Lattice models are also ideal for study by the methods of computational physics, as the discretization of any continuum model automatically turns it into a lattice model. The exact solution to many of these models includes the presence of solitons. Techniques for solving these include the inverse scattering transform and the method of Lax pairs, the Yang–Baxter equation and quantum groups. The solution of these models has given insights into the nature of phase transitions, magnetization and scaling behaviour, as well as insights into the nature of quantum field theory. Physical lattice models frequently occur as an approximation to a continuum theory, either to give an ultraviolet cutoff to the theory to prevent divergences or to perform numerical computations. An example of a continuum theory that is widely studied by lattice models is the QCD lattice model, a discretization of quantum chromodynamics. However, digital physics considers nature fundamentally discrete at the Planck scale, which imposes upper limit to the density of information, aka Holographic principle. More generally, lattice gauge theory and lattice field theory are areas of study. Lattice models are also used to simulate the structure and dynamics of polymers.

In mathematics, an invariant subspace of a linear mapping T : V → V i.e. from some vector space V to itself, is a subspace W of V that is preserved by T; that is, T(W) ⊆ W.

In mathematics, the total variation identifies several slightly different concepts, related to the (local or global) structure of the codomain of a function or a measure. For a real-valued continuous function f, defined on an interval [a, b] ⊂ R, its total variation on the interval of definition is a measure of the one-dimensional arclength of the curve with parametric equation x ↦ f(x), for x ∈ [a, b]. Functions whose total variation is finite are called functions of bounded variation.

The spectrum of a linear operator $that operates on a Banach space is a fundamental concept of functional analysis. The spectrum consists of all scalars such that the operator does not have a bounded inverse on . The spectrum has a standard decomposition into three parts:$

In mathematics, the Gibbs measure, named after Josiah Willard Gibbs, is a probability measure frequently seen in many problems of probability theory and statistical mechanics. It is a generalization of the canonical ensemble to infinite systems. The canonical ensemble gives the probability of the system X being in state x as

In the mathematical discipline of functional analysis, the concept of a compact operator on Hilbert space is an extension of the concept of a matrix acting on a finite-dimensional vector space; in Hilbert space, compact operators are precisely the closure of finite-rank operators in the topology induced by the operator norm. As such, results from matrix theory can sometimes be extended to compact operators using similar arguments. By contrast, the study of general operators on infinite-dimensional spaces often requires a genuinely different approach.

In probability theory, a Markov kernel is a map that in the general theory of Markov processes plays the role that the transition matrix does in the theory of Markov processes with a finite state space.

In mathematics, lifting theory was first introduced by John von Neumann in a pioneering paper from 1931, in which he answered a question raised by Alfréd Haar. The theory was further developed by Dorothy Maharam (1958) and by Alexandra Ionescu Tulcea and Cassius Ionescu Tulcea (1961). Lifting theory was motivated to a large extent by its striking applications. Its development up to 1969 was described in a monograph of the Ionescu Tulceas. Lifting theory continued to develop since then, yielding new results and applications.

Within bayesian statistics for machine learning, kernel methods arise from the assumption of an inner product space or similarity structure on inputs. For some such methods, such as support vector machines (SVMs), the original formulation and its regularization were not Bayesian in nature. It is helpful to understand them from a Bayesian perspective. Because the kernels are not necessarily positive semidefinite, the underlying structure may not be inner product spaces, but instead more general reproducing kernel Hilbert spaces. In Bayesian probability kernel methods are a key component of Gaussian processes, where the kernel function is known as the covariance function. Kernel methods have traditionally been used in supervised learning problems where the input space is usually a space of vectors while the output space is a space of scalars. More recently these methods have been extended to problems that deal with multiple outputs such as in multi-task learning.

<span class="mw-page-title-main">Lie algebra extension</span> Creating a "larger" Lie algebra from a smaller one, in one of several ways

In the theory of Lie groups, Lie algebras and their representation theory, a Lie algebra extension $e$ is an enlargement of a given Lie algebra $g$ by another Lie algebra $h$ . Extensions arise in several ways. There is the trivial extension obtained by taking a direct sum of two Lie algebras. Other types are the split extension and the central extension. Extensions may arise naturally, for instance, when forming a Lie algebra from projective group representations. Such a Lie algebra will contain central charges.

Low-rank matrix approximations are essential tools in the application of kernel methods to large-scale learning problems.

In mathematics, specifically in spectral theory, an eigenvalue of a closed linear operator is called normal if the space admits a decomposition into a direct sum of a finite-dimensional generalized eigenspace and an invariant subspace where $has a bounded inverse. The set of normal eigenvalues coincides with the discrete spectrum.$

References

1 2 Lind, Douglas A.; Marcus, Brian (1995). An introduction to symbolic dynamics and coding. Cambridge: Cambridge University press. ISBN 978-0-521-55900-3.
↑ Ceccherini-Silberstein, T.; Coornaert, M. (2010). Cellular automata and groups Springer Monographs in Mathematics. Springer Monographs in Mathematics. Springer Verlag. doi:10.1007/978-3-642-14034-1. ISBN 978-3-642-14033-4.
1 2 3 4 Sobottka, Marcelo (September 2022). "Some Notes on the Classification of Shift Spaces: Shifts of Finite Type; Sofic Shifts; and Finitely Defined Shifts". Bulletin of the Brazilian Mathematical Society. New Series. 53 (3): 981–1031. arXiv: 2010.10595 . doi:10.1007/s00574-022-00292-x. ISSN 1678-7544. S2CID 254048586.
↑ Weiss, Benjamin (1973), "Subshifts of finite type and sofic systems", Monatsh. Math., 77 (5): 462–474, doi:10.1007/bf01295322, MR 0340556, S2CID 123440583 . Weiss does not describe the origin of the word other than calling it a neologism; however, its Hebrew origin is stated by MathSciNet reviewer R. L. Adler.