The forward and backward shift on the Lipschitz space of a tree
Emmanuel Rivera-Guasco, Rub\'en A. Mart\'inez-Avenda\~no

TL;DR
This paper studies the properties of forward and backward shift operators on Lipschitz spaces of trees, establishing boundedness, isometry conditions, spectra, and hypercyclicity criteria for various types of trees.
Contribution
It introduces the analysis of shift operators on Lipschitz spaces of trees, providing boundedness, spectral, and hypercyclicity results, which are new for this setting.
Findings
Forward shift is bounded and an isometry on leafless trees.
Spectrum of the forward shift is computed for leafless trees.
Backward shift's boundedness, norm, spectrum, and hypercyclicity are characterized.
Abstract
We initiate the study of the forward and backward shifts on the Lipschitz space of a tree, , and on the little Lipshitz space of a tree, . We determine that the forward shift is bounded both on and on and, when the tree is leafless, it is an isometry; we also calculate its spectrum. For the backward shift, we determine when it is bounded on and on , we find the norm when the tree is homogeneous, we calculate the spectrum for the case when the tree is homogeneous, and we determine, for a general tree, when it is hypercyclic.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsHolomorphic and Operator Theory · Advanced Banach Space Theory · Mathematical Analysis and Transform Methods
The Forward and Backward Shift on the Lipschitz Space of a Tree
Rubén A. Martínez-Avendaño
Departamento Académico de Matemáticas
Instituto Tecnológico Autónomo de México
Mexico City, Mexico
and
Emmanuel Rivera-Guasco
Centro de Investigación en Matemáticas
Guanajuato, Mexico
Abstract.
We initiate the study of the forward and backward shifts on the Lipschitz space of an undirected tree, , and on the little Lipschitz space of an undirected tree, . We determine that the forward shift is bounded both on and on and, when the tree is leafless, it is an isometry; we also calculate its spectrum. For the backward shift, we determine when it is bounded on and on , we find the norm when the tree is homogeneous, we calculate the spectrum for the case when the tree is homogeneous, and we determine, for a general tree, when it is hypercyclic.
Key words and phrases:
Hypercyclicity, Lipschitz space, trees, shifts
2010 Mathematics Subject Classification:
47A16, 47B37, 05C05, 05C63
We would like to thank the reviewers for their suggestions, which greatly improved this paper. The first author’s research is partially supported by the Asociación Mexicana de Cultura A.C
1. Introduction
In [11], Colonna and Easley introduced the Lipschitz space of a tree, . This is the Banach space of complex-valued functions on a rooted, countably infinite, locally finite and undirected tree (from now on simply referred to as a tree) which are Lipschitz functions, when the tree is endowed with the edge-counting metric. This space may be considered as the discrete analogue of the classical Bloch space: the space of functions which are Lipschitz when the unit disk is given the hyperbolic or Bergman metric (see, e.g., [24]) and the set of complex numbers is given the usual Euclidean metric.
As it turns out, the Lipschitz space of the tree is, roughly speaking, the space of funtions on the tree whose “derivative” remains bounded on the tree. Therefore, there is also the little Lipschitz space, , defined as the space of functions on the tree whose “derivative” tends to zero when far away from the root of the tree (i.e., on the “boundary” of the tree).
The motivation for investigating spaces of functions on trees comes mainly from harmonic analysis. Early studies of harmonic functions on regular trees were done by Cartier in [7, 8]. Also, Cohen and Colonna studied the Bloch space of harmonic functions on a regular tree in [9], characterizing several properties of functions on this space. Later, in [10] Cohen and Colonna showed how to embed certain homogeneous trees in the hyperbolic disk in a “nice way”: for example, in such a way that bounded harmonic functions on the disk correspond to harmonic functions on the tree.
Several operators on the Lipschitz space of a tree have been studied. For instance, in [11], Colonna and Easley characterize boundedness of multiplication operators on and on , as well as establishing other operator-theoretical properties of such operators. In [4], Allen, Colonna and Easley study properties of the composition operators on the Lipschitz space of a tree. There have also been several studies of multiplication and other operators defined on and on other Banach spaces on trees [1, 2, 3, 12, 13, 14].
The shift operators (both the forward and backward shifts) on have been studied for a long time. There are several reasons why researchers have been interested in shift operators: one of them is that they provide a wealth of examples and counterexamples in operator theory (see, e.g., [23]). In [19], Jabłoński, Jung and Stochel initiated the study of shifts on directed trees. In their paper, they investigate several operator theoretic properties of weighted (forward) shifts on the space of an infinite directed tree. Later, in [21], the first author defined (inspired by a result in [19]) the backward shift operator on a weighted space of a directed tree and characterized its hypercyclicity.
The study of hypercyclic operators goes as far back as the papers of Birkhoff [6] and MacLane [20], but the first example of a hypercyclic operator on a Banach space was given by Rolewicz [22]: it is a multiple of the backward shift on . For the basic definitions and the history of hypercycicity, we recommend the texts [17] and [5].
The purpose of this paper is to introduce the study of the forward and backward shift operator on the Lipschitz space and on the little Lipschitz space . The paper is organized as follows. After giving the basic definitions and notations we will use throughout this paper in Section 2, we define the forward and backward shifts in Section 3. We observe that the forward shift is always an isometry, when the tree is leafless, and find its spectrum when it acts on and on . Also, we establish that the backward shift is the adjoint operator of the forward shift. In Section 4, we give a sufficient and necessary condition to ensure that the backward shift is bounded: it turns out the backward shift is bounded exactly when the tree is homogeneous by sectors. We summarize this characterization in Theorem 4.10, and we also find some lower estimates for the norm. Later, in Section 5, we find the value of the norm of the backward shift and the value of the norm for powers of the backward shift, in the case of homogeneous trees. In Section 6, we obtain the spectrum for the backward shift in the case where the tree is homogeneous, both for (Theorem 6.2) and for (Theorems 6.3 and 6.4). Lastly, in Section 7, we establish that the forward shift can never be hypercyclic, but the backward shift is hypercyclic exactly when the tree has no free ends (Theorem 7.7): this result is analogous to the one found in [21].
2. Preliminaries
As is customary, , , , and will denote the set of natural numbers, the set of nonnegative integers, the set of real numbers, the set of complex numbers, and the open unit disk in centered at the origin, respectively.
Recall that a graph consists of a nonempty set of vertices and a set of edges . In this paper, the set of vertices will always be countably infinite. If , we say that and are adjacent and we denote this by . For each , the degree of , denoted by , is the number of vertices adjacent to . In this paper, all of our graphs will be locally finite; i.e., for every .
A path of lenght joining two vertices and is a finite sequence of distinct vertices . A graph is a tree if for each pair of vertices there is one and only one path between them. In this paper, for a tree we will denote its set of vertices also by the letter , which should cause no confusion.
Every tree we consider here has a distinguished vertex, which we call the root of and denote by . For a tree , we denote by the length of the unique path between the vertices . For we use the notation . We also denote by the set of all vertices minus the root; i.e., . Ocasionally, we will denote by the set .
For each , we define the parent of , denoted by , as the unique vertex in the path from to with . Observe that every vertex in has a parent, except for the root . Inductively, for , we define the -parent of , denoted by , as follows: if , and for , we set , if has a -parent and . The set of all vertices that have -parents is denoted by .
Note: We should point out that with the above definitions we are giving the tree a directed structure, in the manner of the definitions in [19]: each edge can be thought of as a directed edge going to a vertex from its parent . Also, we have chosen a fixed vertex and we called it a root, which has no parent and thus coincides with the definition of a root in [19] (and hence it is unique). We choose not to follow this point of view in this work, since the spaces we study ahead, were originally defined on undirected trees.
If is the parent of , we say that is a child of and we denote the set of all children of by . If is the -parent of we say that is an -child of and we denote the set of all -children of by . For a vertex , we denote by the number of children it has; i.e., is the cardinality of . Also, is the number of -children of ; i.e., is the cardinality of . We will say a tree is homogeneous of order if for all (this differs a bit from the use of the term in the literature).
If a vertex satisfies that (i.e., has no children) we will say that is a leaf of . A tree with no leaves will be called leafless. Observe that in a leafless tree, every vertex is the parent of some other vertex.
Lastly, for every , we denote by the sector determined by , which consists of and all its -children; i.e., , where we will agree that and . Sometimes we will refer to a sector as a subtree.
Let be a countably infinite, locally finite tree. We denote by the set of all functions . In [11], Colonna and Easley define the Lipschitz space of a tree as follows.
Definition 2.1**.**
The Lipschitz space of is the set of all complex-valued Lipschitz functions on ; i.e., is Lipschitz if
[TABLE]
In [11], Colonna and Easley show that
[TABLE]
and hence the Lipschitz space consists of all functions for which
[TABLE]
For , we define as the function
[TABLE]
Thus is in the Lipschitz space if is bounded. We denote by the set of all such functions endowed with the norm
[TABLE]
Colonna and Easley showed in [11] that , with an equivalent norm, is a Banach space. We use the present norm, following [13], to make some calculations cleaner.
The following lemma, which we will use later, can be easily obtained using the results in [11]. For the sake of completeness, we give the proof here (which is a slight modification of the proof in [11]) since we are using a different norm.
Lemma 2.2**.**
Let and . Then
[TABLE]
Proof.
First, we claim that if , and , then for every . We prove this by induction on . It is clear that the claim is true for . Assume that the claim holds for all with , and let with . Then
[TABLE]
which completes the induction step and finishes the proof of the claim.
Now, observe that the lemma is trivial if is the zero function. So assume that is not identically zero. For the moment, assume . Define as . Clearly, . Observe that for all and . Hence
[TABLE]
so aplying the claim to we obtain
[TABLE]
for every . But from this we obtain
[TABLE]
which proves the theorem for functions with . Now let be an arbitrary nonzero function in . Applying the previous argument to we obtain
[TABLE]
which finishes the proof. ∎
Also of interest is the little Lipshitz space of , denoted by , defined as the set of all for which
[TABLE]
Clearly is a subset of and it can be shown (see [11]) that it is a separable closed subspace of .
In Section 7, we will talk about hypercyclicity. Recall that a bounded operator on a Banach space is hypercyclic if there exists a vector (called a hypercyclic vector for ) such that the orbit of under is dense in the Banach space; i.e., the set
[TABLE]
is dense in . Clearly, if is hypercyclic, then must be separable. Also observe that if is a hypercyclic vector, then so is , for any . Thus, if is hypercyclic, then the set of its hypercyclic vectors is dense in .
One way to prove that an operator is hypercyclic is to apply the hypercyclicity criterion. We include here the version we will use in this paper.
Theorem 2.3** (Hypercyclicity Criterion).**
Let be a separable Banach space and a bounded operator on . Assume there exists a set , dense in , and for each there exists a function such that, for every we have
- •
* as ,*
- •
* as , and*
- •
* as .*
Then is hypercyclic.
The proof (of a more general version) of this theorem can be found in [17, p. 74]. A lot more information about the fascinating topic of hypercyclicity can be found in [17] and [5].
3. The forward shift and its adjoint
We now present the two main objects of study in this note. The first operator was originally defined, on , in [19].
Definition 3.1**.**
Let be a a rooted, countably infinite and localy finite tree. The forward shift operator is defined as
[TABLE]
It is clear that is a linear operator on . In [19, Proposition 3.4.1 (iii)], Jabłonski, Jung and Stochel found the form for the (Hilbert space) adjoint of on . Inspired by this result, the first author made the following definition in [21].
Definition 3.2**.**
Let be a rooted, countably infinite and locally finite tree. The backward shift operator is defined as
[TABLE]
where if has no children, the sum is understood to be empty and hence .
Also, it is clear that is a linear operator on . We should point out that the operator is the adjacency operator (or adjacency matrix) of the graph . We will see at the end of this section that is the Banach space adjoint of (on an appropriate space). We start with some results about the forward shift.
Theorem 3.3**.**
Let be a rooted, countable infinite and locally finite tree and let be the forward shift. Then is bounded and . If is leafless, then .
Proof.
Let . It is a straightforward calculation to check that . Hence we get
[TABLE]
and hence . If is leafless, the the inequality in the expression above is an equality. Hence is an isometry, as desired. ∎
The same result holds for as an operator on .
Theorem 3.4**.**
Let be a rooted, countable infinite and locally finite tree and let be the forward shift. Then is bounded and . If is leafless, then .
Proof.
We first observe that if , then . Indeed,
[TABLE]
Since is a closed subspace of , it follows that is a bounded operator on . The calculation in the theorem above then shows that and that is an isometry if is leafless. ∎
If has leaves, we will show later (right after Proposition 3.5) that has nontrivial kernel (both as an operator on and on ) and hence it is not an isometry.
If is leafless, we have shown that is an isometry on . By a theorem of Koehler and Rosenthal (see, for example, [16, p. 11]) there exists a semi-inner product in such that for all . This fact raises a lot of questions about the structure of with this seminorm. We plan to explore this in future research.
On the other hand, if has leaves, one may ask how “far” is from an isometry.111We thank a referee for suggesting this question and for providing the reference in the previous paragraph. It is not hard to see (for example, use Theorem 3.6 below and mimic the argument in [18, Problem 150]) that , if is a surjective isometry. Is it possible that there exists a (non surjective) isometry such that ? We leave this question open for future research.
We should point out that if has leaves, we can define
[TABLE]
where ; i.e., is the subspace of all functions that vanish on every leaf and every ancestor of a leaf. Then clearly is an invariant subspace for (of finite codimension if has finitely many leaves) and S\big{|}_{\mathcal{M}} is an isometry. Thus, even if has leaves, there is a subspace in which acts isometrically.
We will now study the spectrum of . We first show that the forward shift has no eigenvalues, not even on . Recall that for an operator , the set of eigenvalues, the approximate point spectrum and the spectrum, are denoted by , , and , respectively. (The relevant definitions can be found in, e.g., [15].)
Proposition 3.5**.**
Let be a rooted, leafless, countably infinite and locally finite tree and let be the forward shift on . Then .
Proof.
Let . Clearly, for all implies that for all and hence for all . Thus is not eigenvalue.
Assume there exists such that , with . Since it follows that . Now, for every we have and hence . Continuing in this manner, we obtain that for all and thus is not an eigenvalue. ∎
Observe that if the tree has a leaf, then [math] is an eigenvalue of as an operator on : indeed, if is a leaf, then , where is the characteristic function of . The proof above shows that, in this case, [math] is the unique eigenvalue of .
As a corollary, it should be noted that in the leafless case, has no eigenvalues as an operator on and as an operator on . If has a leaf, [math] is an eigenvalue for both on and on , with eigenvector . (By the way, this also shows that if has a leaf, then is not an isometry on nor on ; see Theorems 3.3 and 3.4).
In the leafless case, since is an isometry, its approximate point spectrum lies in the unit circle. Indeed, let with ; then
[TABLE]
and hence is bounded below. Therefore, .
The following theorem gives a full description of the spectrum of in the leafless case. It should be noted that it is known that the spectrum of a noninvertible isometry is always (e.g. [15, p. 213]), we prefer to give an independent proof since it gives more information about .
Theorem 3.6**.**
Let be a rooted, countable infinite and locally finite tree and let be the forward shift on or on . Then, . If is leafless, then .
Proof.
Since , it follows that . It is easily verified that the equation has no solution (just evaluate at the root ), hence .
Let , . Then the equation
[TABLE]
has a unique solution given by , as a straightforward calculation shows. But in this case, , for , is given by
[TABLE]
But since , the function is unbounded, and thus (and ). Hence is not surjective and thus . It then follows that and hence .
Lastly, assume is leafless. Recall that for any operator we have (e.g. [15, Prop. 6.7] and hence (since is an isometry, as noted above). Therefore , as desired. ∎
If has leaves, as we showed before, [math] is an eigenvalue of and thus . Also observe that, as a corollary, we obtain that , even in the case where has leaves.
In [13] it is shown that the dual space of is (isometrically isomorphic to) the space and the dual space of is (isometrically isomorphic to) . Using the identification in [13] we can make the following observations.
Proposition 3.7**.**
Let be a rooted, countable infinite and locally finite tree. If is the forward shift, then is given by , where is the backward shift restricted to .
Proof.
It is shown in [13] that, for every , the functional , defined as
[TABLE]
for each , is bounded and the mapping is an isometric isomorphism from onto . Using this identification we obtain
[TABLE]
Observe that for every vertex either is empty and hence
[TABLE]
or there are vertices with . Therefore,
[TABLE]
Hence, Equation (1) implies that
[TABLE]
Hence can be identified with on . ∎
Using the same technique as above, the following result can be shown.
Proposition 3.8**.**
Let be a rooted, countable infinite and locally finite tree. If is the backward shift, then is given by .
The above results show why we choose to call the “backward” shift, as an analogy of what happens with the classical forward and backward shifts on . We study this operator in greater depth in the next section.
4. The backward shift
It will turn out that the backward shift operator is not always bounded on or on , as was the case for the forward shift. We will need the following definition (see Figure 2 for an example).
Definition 4.1**.**
Let be a rooted, countably infinite and locally finite tree. We say that is homogenous by sectors (at the level ) if there exists such that for all with , we have for each .
Intuitively, a tree is homogeneous by sectors if after some level every subtree is a homogeneous tree. For a tree , homogeneous by sectors at the level , we define
[TABLE]
Since the tree is homogeneous by sectors at the level , it turns out the above supremum is finite and, in fact,
[TABLE]
As it turns out, is bounded if the tree is homogeneous by sectors and gives an estimate of the norm of . In the corollary that follows the next proposition, we will find a more manageable estimate.
Proposition 4.2**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . Then is bounded on . Furthermore,
[TABLE]
Proof.
First observe that
[TABLE]
and hence, since for all , we have
[TABLE]
Now, for , we have
[TABLE]
and hence
[TABLE]
Recalling that for all , it follows that
[TABLE]
Since by Lemma 2.2, we obtain
[TABLE]
Therefore, , for all . Since , it follows that
[TABLE]
as desired. ∎
We now provide a more computable estimate of the norm of the backward shift. For a tree , homogeneous by sectors at the level , we define
[TABLE]
and
[TABLE]
Observe that if is homogeneous of order , then and .
The number that we defined above can be estimated using and . Indeed, if and , then
[TABLE]
If and ,
[TABLE]
Hence, and we obtain the following corollary.
Corollary 4.3**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . Then is bounded on . Furthermore,
[TABLE]
Observe that if and only if and this occurs if and only if and . That is, if and only if is a homogeneous tree of order . We obtain the following corollary.
Corollary 4.4**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . If is homogeneous of order then is bounded on and . In any other case, is bounded on and .
The result in the theorem above also holds for the little Lipschitz space.
Theorem 4.5**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . Then is bounded on .
Proof.
Since is a closed subspace of , it is enough to show that, if , then .
Let . Since , there exists such that
[TABLE]
By Equation (2), for we have
[TABLE]
Hence, since for every with , we have that if then
[TABLE]
Hence , as desired. ∎
It is clear that the norm of , as an operator on , satisfies the same estimates as in Proposition 4.2 and Corollary 4.4.
We want to prove the converse of Proposition 4.2. For that, we need the following lemma.
Lemma 4.6**.**
Let be a rooted, countable infinite and locally finite tree. If , then for every there exists such that and .
Proof.
By contradiction, assume that there exists such that for all with we have . It follows that, for all , we have that if , then
[TABLE]
But, since , this implies that
[TABLE]
and hence that , contradicting the hypothesis. ∎
We can now show that homogeneity by sectors is actually a necessary condition for boundedness of .
Theorem 4.7**.**
Let be a rooted, countable infinite and locally finite tree. If is bounded, then is homogeneous by sectors.
Proof.
Consider the function given by . Observe that, for every we have
[TABLE]
First we show that . By contradiction, assume that . By the previous lemma, there exists a sequence in such that and . But then, Equation (3) gives
[TABLE]
so is unbounded and hence contradicting the boundedness of . Hence , as claimed.
Now, let . If is not homogeneous by sectors then, for every , there exists with such that . But then Equation (3) implies
[TABLE]
which implies that is unbounded and hence contradicting the boundedness of . Therefore, must be homogeneous by sectors. ∎
A similar result holds for , with basically the same proof. We include the details for the sake of completeness.
Theorem 4.8**.**
Let be a rooted, countable infinite and locally finite tree. If is bounded, then is homogeneous by sectors.
Proof.
Consider the function given by . Observe that, for every we have
[TABLE]
First we show that . By contradiction, assume that . By Lemma 4.6, there exists a sequence in such that and . But then, Equation (4) gives
[TABLE]
But this expression is unbounded and so is . Hence contradicting the boundedness of .
Now, let . If is not homogeneous by sectors then, for every , there exists with such that . But then Equation (3) implies
[TABLE]
which implies that is unbounded and hence contradicting the boundedness of . Therefore, must be homogeneous by sectors. ∎
The following proposition characterizes trees that are homogeneous by sectors in terms of a combinatorial quantity.
Proposition 4.9**.**
Let be a rooted, countable infinite and locally finite tree. Then if and only if is homogeneous by sectors.
Proof.
First assume that is homogeneous by sectors at the level . Then, for all , we have and hence
[TABLE]
which is a finite set. Hence, .
Now, assume that . If was not homogeneous by sectors, for every , there would exist with and such that . But then
[TABLE]
and therefore , which is a contradition. Hence is homogeneous by sectors. ∎
We can summarize the results of this section in the following theorem.
Theorem 4.10**.**
Let be a rooted, countable infinite and locally finite tree. Let be the backward shift. The following are equivalent.
- •
* is bounded.*
- •
* is bounded.*
- •
* is homogeneous by sectors.*
- •
.
Moreover, in this case, .
We can obtain some simple estimates from below for the norm of the backward shift. Indeed, assume is homogeneous by sectors and let be defined as (see Figure 3 for an example)
[TABLE]
Clearly, and . But we also have
[TABLE]
Then and thus . This estimate is part of the next proposition.
Proposition 4.11**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . Then , where .
Proof.
Let such that . Define the function as (see Figure 4 for an example)
[TABLE]
Clearly, and . It is easy to check that and . Hence . Therefore, . By the calculations before the statement of this proposition, it follows that . ∎
Observe that the same estimate holds for as operator on . For a particular type of trees, the above estimate can be improved.
Proposition 4.12**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . Let . Then .
Proof.
Let , with , be such that . Define the function as (see Figure 5 for an example)
[TABLE]
Clearly, and it is straightforward to check that . But also, and . Hence . Therefore, , as desired. ∎
It is easily verified that the estimate in Proposition 4.12 is better than the estimate in Proposition 4.11 if .
How good are the estimates in Propositions 4.2 and 4.11? In the next section, we will show that for a homogeneous tree, we can find the precise value of . For now, we show what happens for the case of a tree that is homogeneous by sectors at the level .
Proposition 4.13**.**
Let be a rooted, countable infinite and locally finite tree. Assume that is homogeneous by sectors at the level . Then
[TABLE]
both as an operator on and on .
Proof.
Let and let be the children of . Furthermore, assume that .
Let us deal with the case first. It is easy to check that if there exists with , and if for all . Hence, by Proposition 4.2, . By proposition 4.11, it follows that and thus , as desired.
So assume for the rest of the proof that . We have two cases:
- •
Assume . We have three cases:
If for some , we have that
[TABLE]
If for some , we have that
[TABLE]
If for some , we have that
[TABLE]
Thus, , and by Proposition 4.2 . But , since . Therefore . But by Proposition 4.11 we also have , thus . Therefore , as desired.
- •
Assume .
If for some , then
[TABLE]
If for some , then
[TABLE]
Hence, , and by Proposition 4.2, we have . But now, by Proposition 4.12, we have that , and therefore . Thus , as desired.∎
Are the estimates in Propositions 4.2, 4.11 and 4.12 sharp for ? We have not been able to obtain an answer and we leave open the question of what the norm of is for general trees.
**Note: **We would like to thank a referee for suggesting we investigate lower estimates for the norm of , which led to improvements of the estimates we had in a previous version of this paper.
5. Norm of and of on Homogeneous Trees
In this section, we find an expression for the norms of and for the case when is a homogeneous tree. Recall that a tree is homogeneous of order if for all .
Theorem 5.1**.**
Let be a rooted homogeneous tree of order and let be the backward shift on . Then
[TABLE]
Proof.
First of all, observe that, by Corollary 4.3, since and we have .
By Proposition 4.11, we have that , and hence the result follows. ∎
We have the same result for the little Lipschitz space.
Theorem 5.2**.**
Let be a rooted homogeneous tree of order and let be the backward shift on . Then
[TABLE]
For the computation of the spectrum of for a rooted homogeneous tree, we will need to find the norm of . Let us do some preliminary computations. First of all, it is clear that, for any we have
[TABLE]
From this, and since each vertex in is the parent of vertices in , it follows that
[TABLE]
In the same manner, we have
[TABLE]
since each vertex in is the parent of vertices in . Proceeding inductively, we get
[TABLE]
In short, we have obtained
[TABLE]
We will use this expression in the proof of the following proposition.
Proposition 5.3**.**
Let be a rooted homogeneous tree of order and let be the backward shift on . Then,
[TABLE]
Proof.
Using Equation (6), we have
[TABLE]
Since for every there are vertices in , and for all , we get
[TABLE]
and therefore, since , we have
[TABLE]
Now, let . Equation (5) is
[TABLE]
Also, we have
[TABLE]
The two equations above give
[TABLE]
Since each vertex in is the parent of vertices in , we have
[TABLE]
Inductively, we obtain
[TABLE]
Since there are vertices in different than , we have
[TABLE]
In short, we have obtained
[TABLE]
Substituting Equation (6) (for ) and (8) into Equation (7) we obtain
[TABLE]
Since, for every , there are vertices in , and for every there are vertices in , we obtain
[TABLE]
Therefore, , as desired. ∎
Using this proposition we can compute the exact value of the norm of .
Theorem 5.4**.**
Let be a rooted homogeneous tree of order and let be the backward shift on . Then, for every ,
[TABLE]
Proof.
First of all, observe that, by Proposition 5.3 we have , which equals if , and if .
Choose a fixed . Define the function (see Figure 6 for an example) as
[TABLE]
It is clear that
[TABLE]
and therefore and . Also, a straightforward computation shows that
[TABLE]
Hence,
[TABLE]
If , Equation (9) simplifies to
[TABLE]
Hence, if , we have then that , which together with Proposition 5.3 gives that , as desired.
If , Equation (9) simplifies to
[TABLE]
It can be checked that
[TABLE]
and hence , which together with Proposition 5.3 gives
[TABLE]
as desired. ∎
Observe that, in the previous proof, the function is also in . Hence we also obtain
Theorem 5.5**.**
Let be a rooted homogeneous tree of order and let be the backward shift on . Then
[TABLE]
6. Spectrum of on Homogeneous Trees
In this section, we compute the spectrum of for both the Lipschitz and the little Lipschitz space in the case where is a rooted homogeneous tree. First, we obtain part of the set of eigenvalues. We will show later that we actually have an equality.
Theorem 6.1**.**
Let be a rooted homogeneous tree of order . If is the backward shift on , then
[TABLE]
If is the backward shift on , then
[TABLE]
Proof.
Define as . Then,
[TABLE]
For we have
[TABLE]
Hence, if and only if and if and only if or . The result now follows immediately. ∎
With this, we are ready to prove the following theorem.
Theorem 6.2**.**
Let be a rooted homogeneous tree of order . If is the backward shift on , then
[TABLE]
Proof.
First, we will compute the spectral radius of . By Theorem 5.4 if then
[TABLE]
If , Theorem 5.4 gives
[TABLE]
Therefore, . This and the previous theorem imply that
[TABLE]
and hence the result follows. ∎
We obtain a similar result for the backward shift on .
Theorem 6.3**.**
Let be a rooted homogeneous tree of order . If is the backward shift on , then
[TABLE]
Proof.
As was the case in the theorem above, by Theorem 5.5 we have and hence
[TABLE]
This, and Theorem 6.1 give that
[TABLE]
Since, for any operator we have (see, e.g. [15, p. 210]), we have that
[TABLE]
and, again, by Theorem 6.1, we obtain
[TABLE]
which completes the proof. ∎
With the previous result showing what the spectrum of the backward shift is, we can determine the point spectrum.
Theorem 6.4**.**
Let be a rooted homogeneous tree of order . If is the backward shift on , then
[TABLE]
Proof.
By Theorems 6.1 and 6.3, it suffices to show that if and , then . So let with and assume then that for a nonzero .
First, since is not zero, there exists a vertex such that . Dividing by a constant, if necessary, we may assume that
Now, we claim that for all there exists with . Indeed, suppose this was not the case. Then, for some we would have for all . But since , we have
[TABLE]
and hence we obtain
[TABLE]
which is a contradiction, so the claim is true.
Now, since , there exists such that, for all we have
[TABLE]
By the claim, there exists with . Hence
[TABLE]
But then,
[TABLE]
since every satisfies . But the last display implies that , which is a contradiction. Hence there cannot be with and for a nonzero , which completes the proof of the theorem. ∎
7. Hypercyclicity
In [13], it is shown that (with an equivalent norm) is not separable, while is separable (this was originally shown in [11]). So, in order to study hypercyclicity of operators, we need to restrict ourselves to , which we do from now on.
First, we get rid of the question of whether is hypercyclic. It is not since the norm of is one and therefore can never be hypercyclic. We offer an alternative proof.
Theorem 7.1**.**
Let be a rooted, countably infinite and locally finite tree and let be the forward shift on . Then is not hypercyclic.
Proof.
If were hypercyclic, then there would exist and a natural number such that
[TABLE]
where is the characteristic function of the root . The definition of the norm in then would imply that
[TABLE]
But, since for every , this is a contradiction. Therefore is not hypercyclic. ∎
We will use the following lemma, which is proved in [14].
Lemma 7.2**.**
Let be the set of all functions in with finite support. Then is dense in .
The following definition will be useful to characterize hypercyclicity.
Definition 7.3**.**
Let a rooted tree and . We say that is a free end (at ) if for all we have .
Recall that denotes the set of vertices that have -parents; i.e., if there exists with . Also, recall that denotes the number of vertices in the set .
We define the function as
[TABLE]
The following lemma will be used later.
Lemma 7.4**.**
Let be a rooted, countably infinite and locally finite tree. If is homogeneous by sectors and has no free ends then
[TABLE]
Proof.
Since is homogeneous by sectors, there exists such that for every with , we have for every . Since is locally finite, there exist finitely many such , say . For each , define . Since has no free ends, .
Let and let .
- •
If , then and hence
[TABLE]
- •
If , then and hence . Therefore,
[TABLE]
- •
If , let . Then, since , we have and hence
[TABLE]
But clearly which implies that and hence . Since this gives
[TABLE]
Hence, for every , we have
[TABLE]
if . Therefore, if , then
[TABLE]
and thus
[TABLE]
We can now give a sufficient condition for hypercyclicity of .
Theorem 7.5**.**
Let be an countably infinite and locally finite tree and assume that is bounded on . If has no free ends, then is hypercyclic.
Proof.
To show hypercyclicity of , we will use the Hypercyclicity Criterion (Theorem 2.3). Let be the set of all functions with finite support and for each , define the function as
[TABLE]
(Observe that also has finite support.)
- (1)
First, let . Choose such that for all with . Then, for all ,
[TABLE]
if and, similarly, for all we have if . Hence as long as . Therefore , as , as desired. 2. (2)
Let and . Since is of finite support, there exists such that for all .
- •
If , then ; while if , then . Hence if .
- •
If , then, since , then
[TABLE]
- •
If . Then,
[TABLE]
Therefore, for all , we have
[TABLE]
and hence
[TABLE]
Therefore, since has no free ends, by Lemma 7.4 (since is bounded and hence is homogeneous by sectors) we have that , as desired. 3. (3)
Now, let . We then have
[TABLE]
Therefore, as , as desired.
Since all conditions in the hyperciclicity criterion hold, it follows that is hypercyclic. ∎
The following theorem shows that the condition on the previous theorem actually characterizes hypercyclicity.
Theorem 7.6**.**
Let be an countably infinite and locally finite tree and assume that is bounded on . If is hypercyclic, then has no free ends.
Proof.
Asumme that has a free end. Let be a vertex on the free end such that and . Then, for each each of the sets and has a unique element.
Since is hypercyclic there exists a hypercyclic vector . In fact, by the density of the hypercyclic vectors, we may assume that . Let be the characteristic funtion of . By hypercyclicity of , there exists , such that
[TABLE]
But then
[TABLE]
where is the unique element in the set . Hence,
[TABLE]
and therefore
[TABLE]
But, since , we have
[TABLE]
which is a contradiction. Therefore, cannot have free ends. ∎
We summarize the previous two theorems to obtain a full characterization of the hypercyclicity of .
Theorem 7.7**.**
Let be a rooted, countably infinite and locally finite tree and assume that is bounded on . Then is hypercyclic if and only if has no free ends.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] R. F. Allen, F. Colonna and G. R. Easley, Multiplication Operators between Lipschitz-type Spaces on a Tree Int. J. Math. Math. Sci. (2011) Art. ID 472495, 36 pp.
- 2[2] R. F. Allen, F. Colonna and G. R. Easley, Multiplication Operators on the Iterated Logarithmic Lipschitz Spaces of a Tree , Mediterr. J. Math. 9 (2012) 575–600.
- 3[3] R. F. Allen, F. Colonna and G. R. Easley, Multiplication Operators on the Weighted Lipschitz Space of a Tree , J. Operator Theory 69 (2013) 209–231.
- 4[4] R. F. Allen, F. Colonna and G. R. Easley, Composition Operators on the Lipschitz Space of a Tree , Mediterr. J. Math. 11 (2014) 97–108.
- 5[5] F. Bayart and É. Matheron, Dynamics of Linear Operators , Cambridge University Press, Cambridge, 2009.
- 6[6] G. D. Birkhoff, Démonstration d’un théorème élémentaire sur les fonctions entières , C. R. Acad. Sci. Paris 189 (1929) 473–475
- 7[7] P. Cartier, Fonctions harmoniques sur un arbre , in Symposia Mathematica, Vol. IX (Convegno di Calcolo delle Probabilità, INDAM, Rome, 1971) Academic Press, London, 1972.
- 8[8] P. Cartier, Géométrie et analyse sur les arbres , in Séminaire Bourbaki, 24ème année (1971/1972), Exp. No. 407, pp. 123–40. Lecture Notes in Math., Vol. 317, Springer, Berlin, 1973.
