On the dynamics of weighted composition operators Nilson C. Bernardes Jr, Antonio Bonilla, João V. A. Pinto Abstract We study the properties of power-boundedness, Li-Yorke chaos, distributional chaos, absolutely Cesàro boundedness and mean Li-Yorke chaos for weighted composition operators on L p ​ ( μ ) L^{p}(\mu) spaces and on C 0 ​ ( Ω ) C_{0}(\Omega) spaces. We illustrate the general results by presenting several applications to weighted shifts on the classical sequence spaces c 0 ​ ( ℕ ) c_{0}(\mathbb{N}) , c 0 ​ ( ℤ ) c_{0}(\mathbb{Z}) , ℓ p ​ ( ℕ ) \ell^{p}(\mathbb{N}) and ℓ p ​ ( ℤ ) \ell^{p}(\mathbb{Z}) ( 1 ≤ p < ∞ 1\leq p<\infty ) and to weighted translation operators on the classical function spaces C 0 [ 1 , ∞ ) C_{0}[1,\infty) , C 0 ​ ( ℝ ) C_{0}(\mathbb{R}) , L p [ 1 , ∞ ) L^{p}[1,\infty) and L p ​ ( ℝ ) L^{p}(\mathbb{R}) ( 1 ≤ p < ∞ 1\leq p<\infty ). Keywords: Weighted composition operators, power-boundedness, Li-Yorke chaos, distributional chaos, absolutely Cesàro boundedness, mean Li-Yorke chaos. 2020 Mathematics Subject Classification: Primary 47A16, 47B33; Secondary 46E15, 46E30. 1 Introduction Linear dynamics is a branch of mathematics that lies at the interface between the big areas of dynamical systems and operator theory. Its main focus is to study the dynamics of continuous linear operators on topological vector spaces (often Banach or Fréchet spaces). The area has undergone significant development during the last four decades. We refer the reader to the books [ 4 , 23 ] for an overview of the area up to around 2010. The main objective of these books is the study of chaotic behaviors related to the concept of hypercyclicity (existence of a dense orbit), such as hypercyclicity itself, Devaney chaos, mixing and frequent hypercyclicity. During the 2010s, the series of papers [ 5 , 7 , 8 , 9 , 10 ] laid the foundation for a theory of chaotic behaviors related to the dynamics of pairs of points (including Li-Yorke chaos, distributional chaos and mean Li-Yorke chaos) in the context of linear dynamics. More recently, some fundamental notions in dynamical systems that are not notions of chaos, such as hyperbolicity, expansivity, shadowing and stability, have started to be systematically investigated in the setting of linear dynamics. We refer the reader to the recent articles [ 11 , 13 ] , where many additional references can be found. On the other hand, composition operators constitute a very important class of operators in operator theory and its applications. There is a vast literature on the dynamics of these operators in different contexts (analytic, measure-theoretic, topological). We refer the reader to [ 2 , 15 , 16 , 17 , 21 , 25 , 29 , 32 ] , for instance. In the particular case of composition operators on L^{p}(\mu) spaces, Li-Yorke chaos was studied in [ 12 , 14 ] , topological transitivity and mixing in [ 3 ] , Devaney chaos and frequent hypercyclicity in [ 20 ] , generalized hyperbolicity and shadowing in [ 19 ] , expansivity and strong structural stability in [ 28 ] , distributional chaos in [ 24 ] , and Kitai’s Criterion in [ 22 ] . Our goal in the present work is to study the concepts of power-boundedness, Li-Yorke chaos, distributional chaos, p -absolutely Cesàro boundedness and mean Li-Yorke chaos for weighted composition operators C_{w,f}(\varphi)=(\varphi\circ f)\cdot w on the classical Banach spaces L^{p}(\mu) and C_{0}(\Omega) . Special emphasis will be given to obtaining characterizations of these concepts. In order to illustrate the general results, we will present several applications to two special classes of weighted composition operators, namely, weighted shifts on the classical sequence spaces c_{0}(\mathbb{N}) , c_{0}(\mathbb{Z}) , \ell^{p}(\mathbb{N}) and \ell^{p}(\mathbb{Z}) ( 1\leq p<\infty ) and weighted translation operators on the classical function spaces C_{0}[1,\infty) , C_{0}(\mathbb{R}) , L^{p}[1,\infty) and L^{p}(\mathbb{R}) ( 1\leq p<\infty ). Our results on Li-Yorke chaos will emphasize the close relationship between this concept and the notion of power-boundedness and will complement previous studies developed in [ 12 , 14 ] . In the case of L^{p}(\mu) spaces, our results on distributional chaos for weighted composition operators will extend and complement recent results obtained in the preprint [ 24 ] in the unweighted case, but we will also study here the case of C_{0}(\Omega) spaces. To the best of the authors’ knowledge, the concepts of absolute Cesàro boundedness and mean Li-Yorke chaos have not been studied before in the context of the present article. Throughout \mathbb{K} denotes either the field \mathbb{R} of real numbers or the field \mathbb{C} of complex numbers, \mathbb{Z} denotes the ring of integers, \mathbb{N} denotes the set of all positive integers, and \mathbb{N}_{0}=\mathbb{N}\cup\{0\} . By an operator on a Banach space Y , we mean a bounded linear map T:Y\to Y . Recall that the operator norm of such a map is the number \|T\|=\sup\{\|Ty\|:\|y\|\leq 1\} . 2 Preliminaries 2.1 Weighted composition operators on L^{p}(\mu) Throughout this article we fix a real number p\in[1,\infty) and an arbitrary positive measure space (X,\mathfrak{M},\mu) , unless otherwise specified. L^{p}(\mu) denotes the Banach space over \mathbb{K} of all \mathbb{K} -valued p -integrable functions on (X,\mathfrak{M},\mu) endowed with the p -norm \|\varphi\|_{p}=\Big(\int_{X}|\varphi|^{p}d\mu\Big)^{\frac{1}{p}}. L^{\infty}(\mu) denotes the Banach space over \mathbb{K} of all \mathbb{K} -valued essentially bounded measurable functions on (X,\mathfrak{M},\mu) endowed with the essential supremum norm \|\varphi\|_{\infty}=\operatorname{ess\,sup}|f|. We also consider a measurable map w:X\to\mathbb{K} such that \varphi\cdot w\in L^{p}(\mu)\ \text{ for all }\varphi\in L^{p}(\mu). (1) If the measure \mu is semifinite, that is, every set with infinite measure contains a set with positive finite measure (in particular, if \mu is \sigma -finite), then ( 1 ) is equivalent to w\in L^{\infty}(\mu) [ 14 , Proposition 7] , but this equivalence is not true in general [ 14 , Remark 8] . Given a bimeasurable map f:X\to X (i.e., f(B)\in\mathfrak{M} and f^{-1}(B)\in\mathfrak{M} whenever B\in\mathfrak{M} ), it is not difficult to show that the weighted composition operator C_{w,f}(\varphi)=(\varphi\circ f)\cdot w is a well-defined bounded linear operator on L^{p}(\mu) if and only if there is a constant c\in(0,\infty) such that \int_{B}|w|^{p}d\mu\leq c\,\mu(f(B))\ \text{ for every }B\in\mathfrak{M}. (2) Moreover, in this case, \|C_{w,f}\|^{p}\leq c . Whenever we consider a weighted composition operator C_{w,f} on L^{p}(\mu) , we will implicitly assume that w and f satisfy conditions ( 1 ) and ( 2 ). As in [ 14 , Section 3] , we associate to w and f the following sequence of positive measures on (X,\mathfrak{M}) : \mu_{n}(B)=\int_{B}|w^{(n)}|^{p}d\mu\ \ \ \ (B\in\mathfrak{M},n\in\mathbb{N}), where w^{(1)}=w\ \ \ \text{ and }\ \ \ w^{(n)}=(w\circ f^{n-1})\cdot\ldots\cdot(w\circ f)\cdot w\ \text{ for }n\geq 2. Lemma 1 ( [ 14 ] ) . If ( 2 ) holds, then \mu_{n}(B)\leq c^{n}\mu(f^{n}(B))\ \text{ for all }B\in\mathfrak{M}\text{ and }n\in\mathbb{N}. Example 2 . If X=\mathbb{N} (resp. X=\mathbb{Z} ), \mathfrak{M}=\mathcal{P}(X) (the power set of X ), \mu is the counting measure on \mathfrak{M} and f:n\in X\mapsto n+1\in X , then L^{p}(\mu)=\ell^{p}(\mathbb{N}) (resp. L^{p}(\mu)=\ell^{p}(\mathbb{Z}) ) and C_{w,f} coincides with the unilateral (resp. bilateral ) weighted backward shift B_{w}:(x_{n})_{n\in X}\in\ell^{p}(X)\mapsto(w_{n}x_{n+1})_{n\in X}\in\ell^{p}(X). Example 3 . If X=[1,\infty) (resp. X=\mathbb{R} ), \mathfrak{M} is the \sigma -algebra of all Lebesgue measurable sets in X , \mu is the Lebesgue measure on \mathfrak{M} and f:x\in X\mapsto x+1\in X , then L^{p}(\mu)=L^{p}[1,\infty) (resp. L^{p}(\mu)=L^{p}(\mathbb{R}) ) and C_{w,f} coincides with the unilateral (resp. bilateral ) weighted translation operator T_{w}:\varphi\in L^{p}(X)\mapsto\varphi(\cdot+1)w(\cdot)\in L^{p}(X). Recall that B_{w} (resp. T_{w} ) is a well-defined bounded linear operator exactly when w is a bounded sequence (resp. an essentially bounded measurable function). According to our convention, whenever we consider such an operator B_{w} (resp. T_{w} ), we will implicitly assume that this is the case. 2.2 Weighted composition operators on C_{0}(\Omega) Throughout this article we fix an arbitrary locally compact Hausdorff space \Omega , unless otherwise specified. C_{0}(\Omega) denotes the Banach space over \mathbb{K} of all continuous maps \varphi:\Omega\to\mathbb{K} that vanish at infinity endowed with the supremum norm \|\varphi\|=\sup_{x\in\Omega}|\varphi(x)|. More generally, given any bounded map \varphi:\Omega\to\mathbb{K} and any B\subset\Omega , we define \|\varphi\|_{B}=\sup_{x\in B}|\varphi(x)|, where we consider this supremum to be 0 if B=\varnothing . In the case B=\Omega , we usually write \|\varphi\| instead of \|\varphi\|_{\Omega} even if \varphi\not\in C_{0}(\Omega) . Recall that the support of \varphi:\Omega\to\mathbb{K} is defined by \operatorname{supp}\varphi=\overline{\{x\in\Omega:\varphi(x)\neq 0\}}. C_{c}(\Omega) denotes the subspace of C_{0}(\Omega) consisting of those \varphi\in C_{0}(\Omega) that have compact support. We also consider a continuous map w:\Omega\to\mathbb{K} such that \varphi\cdot w\in C_{0}(\Omega)\ \text{ for all }\varphi\in C_{0}(\Omega). (3) It is easy to show that ( 3 ) holds if and only if w is bounded in \Omega . Given a continuous map f:\Omega\to\Omega , it is not difficult to show that the weighted composition operator C_{w,f}(\varphi)=(\varphi\circ f)\cdot w is a well-defined bounded linear operator on C_{0}(\Omega) if and only if \{x\in\Omega:f(x)\in K\text{ and }|w(x)|\geq\varepsilon\}\text{ is compact, for all }\varepsilon>0\text{ and }K\subset\Omega\text{ compact}. (4) Whenever we consider a weighted composition operator C_{w,f} on C_{0}(\Omega) , we will implicitly assume that w and f satisfy conditions ( 3 ) and ( 4 ). As in [ 14 , Section 2] , we associate to w and f the following sequence of continuous maps from \Omega into \mathbb{K} : w^{(1)}=w\ \ \ \text{ and }\ \ \ w^{(n)}=(w\circ f^{n-1})\cdot\ldots\cdot(w\circ f)\cdot w\ \text{ for }n\geq 2. Example 4 . If \Omega=\mathbb{N} (resp. \Omega=\mathbb{Z} ) endowed with the discrete topology and f:n\in\Omega\mapsto n+1\in\Omega , then C_{0}(\Omega)=c_{0}(\mathbb{N}) (resp. C_{0}(\Omega)=c_{0}(\mathbb{Z}) ) and C_{w,f} coincides with the unilateral (resp. bilateral ) weighted backward shift B_{w}:(x_{n})_{n\in\Omega}\in c_{0}(\Omega)\mapsto(w_{n}x_{n+1})_{n\in\Omega}\in c_{0}(\Omega). Example 5 . If \Omega=[1,\infty) (resp. \Omega=\mathbb{R} ) endowed with its usual topology and f:x\in\Omega\mapsto x+1\in\Omega , then C_{0}(\Omega)=C_{0}[1,\infty) (resp. C_{0}(\Omega)=C_{0}(\mathbb{R}) ) and C_{w,f} coincides with the unilateral (resp. bilateral ) weighted translation operator T_{w}:\varphi\in C_{0}(\Omega)\mapsto\varphi(\cdot+1)w(\cdot)\in C_{0}(\Omega). Recall that B_{w} (resp. T_{w} ) is a well-defined bounded linear operator exactly when w is a bounded sequence (resp. a bounded continuous function). According to our convention, whenever we consider such an operator B_{w} (resp. T_{w} ), we will implicitly assume that this is the case. 3 Power-bounded versus Li-Yorke chaotic weighted composition operators Recall that an operator T on a Banach space Y is said to be power-bounded if there exists C\in(0,\infty) such that \|T^{n}\|\leq C for all n\in\mathbb{N} . More generally, given a subspace Z of Y , we say that T is power-bounded in Z if there exists C\in(0,\infty) such that \|T^{n}|_{Z}\|\leq C\ \text{ for all }n\in\mathbb{N}, where T^{n}|_{Z}:Z\to Y is the bounded linear map obtained by restricting the domain of T^{n} to Z . Given a metric space M , recall that a map f:M\to M is said to be Li-Yorke chaotic if there exists an uncountable set S\subset M such that each pair (x,y) of distinct points in S is a Li-Yorke pair for f , in the sense that \liminf_{n\to\infty}d(f^{n}(x),f^{n}(y))=0\ \ \text{ and }\ \ \limsup_{n\to\infty}d(f^{n}(x),f^{n}(y))>0. If the set S can be chosen to be dense in M , then f is densely Li-Yorke chaotic . An extensive study of the concept of Li-Yorke chaos in the setting of linear dynamics was developed in [ 5 , 8 ] . In particular, the following useful characterizations were obtained: For any operator T on any Banach space Y , the following assertions are equivalent: (i) T is Li-Yorke chaotic; (ii) T admits a semi-irregular vector , that is, a vector y\in Y such that \liminf_{n\to\infty}\|T^{n}y\|=0\ \ \text{ and }\ \ \limsup_{n\to\infty}\|T^{n}y\|>0. (iii) T admits an irregular vector , that is, a vector y\in Y such that \liminf_{n\to\infty}\|T^{n}y\|=0\ \ \text{ and }\ \ \limsup_{n\to\infty}\|T^{n}y\|=\infty. Let us also recall that by an irregular manifold for T we mean a vector subspace of Y consisting, except for the zero vector, of irregular vectors for T . 3.1 The case of the space L^{p}(\mu) Theorem 6 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . For each n\in\mathbb{N} , \|(C_{w,f})^{n}\|=\sup_{0<\mu(B)<\infty}\Big(\frac{\mu_{n}(f^{-n}(B))}{\mu(B)}\Big)^{\frac{1}{p}}, (5) where the supremum is taken over all measurable sets B satisfying 0<\mu(B)<\infty . In particular, C_{w,f} is power-bounded if and only if there exists a constant C\in(0,\infty) such that, for each measurable set B of finite positive \mu -measure, \mu_{n}(f^{-n}(B))\leq C\,\mu(B)\ \text{ for all }n\in\mathbb{N}. (6) Proof. Fix n\in\mathbb{N} and denote the right-hand side of ( 5 ) by r . Given a measurable set B of finite positive \mu -measure, define \phi=\frac{1}{\mu(B)^{1/p}}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B} . Then \|\phi\|_{p}=1 and \displaystyle\|(C_{w,f})^{n}(\phi)\|_{p}^{p} \displaystyle= \displaystyle\frac{1}{\mu(B)}\int_{X}|{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B}\circ f^{n}|^{p}|w\circ f^{n-1}|^{p}\cdots|w\circ f|^{p}|w|^{p}d\mu \displaystyle= \displaystyle\frac{1}{\mu(B)}\int_{f^{-n}(B)}|w\circ f^{n-1}|^{p}\cdots|w\circ f|^{p}|w|^{p}d\mu \displaystyle= \displaystyle\frac{\mu_{n}(f^{-n}(B))}{\mu(B)}\cdot This implies that r\leq\|(C_{w,f})^{n}\| . On the other hand, take any real number t>1 . Given \varphi\in L^{p}(\mu) , \varphi\neq 0 , consider the measurable sets B_{i}=\{x\in X:t^{i-1}\leq|\varphi(x)|1 is arbitrary, we obtain \|(C_{w,f})^{n}\|\leq r , which completes the proof of ( 5 ). The last assertion follows from ( 5 ). ∎ In the particular case of a weighted shift B_{w} on \ell^{p}(\mathbb{N}) , it is well known that \|(B_{w})^{n}\|=\sup_{i\in\mathbb{N}}|w_{i}\cdots w_{i+n-1}|\ \text{ for all }n\in\mathbb{N}, (7) which is a special case of formula ( 5 ). Thus, B_{w} is power-bounded if and only if \sup\{|w_{i}\cdots w_{j}|:i,j\in\mathbb{N},i\leq j\}<\infty. (8) Therefore, [ 14 , Corollary 14] (see also [ 5 , Proposition 27] ) can be rewritten as follows. Corollary 7 . A weighted shift B_{w} on \ell^{p}(\mathbb{N}) is Li-Yorke chaotic if and only if it is not power-bounded. Similarly, in the particular case of a weighted shift B_{w} on \ell^{p}(\mathbb{Z}) , it is well known that \|(B_{w})^{n}\|=\sup_{i\in\mathbb{Z}}|w_{i}\cdots w_{i+n-1}|\ \text{ for all }n\in\mathbb{N}, (9) which is also a special case of formula ( 5 ). Thus, B_{w} is power-bounded if and only if \sup\{|w_{i}\cdots w_{j}|:i,j\in\mathbb{Z},i\leq j\}<\infty. (10) Hence, [ 14 , Corollary 15] (see also [ 12 , Corollary 1.6] ) can be rewritten as follows. Corollary 8 . A weighted shift B_{w} on \ell^{p}(\mathbb{Z}) with nonzero weights is Li-Yorke chaotic if and only if it is not power-bounded and \displaystyle\liminf_{n\to\infty}|w_{-n}\cdots w_{-1}|=0 . We now return to general weighted composition operators on L^{p}(\mu) . Theorem 9 . Given a weighted composition operator C_{w,f} on L^{p}(\mu) , let \mathfrak{X} be the closed subspace of L^{p}(\mu) generated by the characteristic functions {\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B} such that B is a measurable set of finite \mu -measure satisfying \liminf_{n\to\infty}\mu_{n}(f^{-n}(B))=0. (11) Then, C_{w,f} is Li-Yorke chaotic if and only if it is not power-bounded in \mathfrak{X} . Proof. Suppose that C_{w,f} is Li-Yorke chaotic. By [ 14 , Theorem 10] , there exist a nonempty countable family (B_{i})_{i\in I} of measurable sets of finite positive \mu -measure and an increasing sequence (n_{j})_{j\in\mathbb{N}} of positive integers such that \sup\Big\{\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}:i\in I,n\in\mathbb{N}\Big\}=\infty (12) and \lim_{j\to\infty}\mu_{n_{j}}(f^{-n_{j}}(B_{i}))=0\ \text{ for all }i\in I. (13) For each i\in I , let \phi_{i}=\frac{1}{\mu(B_{i})^{1/p}}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{i}} . By ( 13 ), \phi_{i}\in\mathfrak{X} for all i\in I . Since \|\phi_{i}\|_{p}=1 and \|(C_{w,f})^{n}(\phi_{i})\|_{p}^{p}=\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}\ \ \ (i\in I,n\in\mathbb{N}), it follows from ( 12 ) that C_{w,f} is not power-bounded in \mathfrak{X} . Conversely, suppose that C_{w,f} is not Li-Yorke chaotic. Then, C_{w,f} does not admit a semi-irregular vector. Thus, ( 11 ) is equivalent to \lim_{n\to\infty}\mu_{n}(f^{-n}(B))=0. This implies that the set \mathcal{R}_{1} of all \varphi\in\mathfrak{X} whose orbit under C_{w,f} has a subsequence converging to zero is dense in \mathfrak{X} . Hence, by [ 8 , Corollary 4] , \mathcal{R}_{1} is residual in \mathfrak{X} . If C_{w,f} was not power-bounded in \mathfrak{X} , the Banach-Steinhaus theorem [ 31 , Theorem 2.5] would imply that the set \mathcal{R}_{2} of all \varphi\in\mathfrak{X} whose orbit under C_{w,f} is unbounded is also residual in \mathfrak{X} . This would imply the existence of an irregular vector for C_{w,f} (take any vector in \mathcal{R}_{1}\cap\mathcal{R}_{2} ). Since we are assuming that C_{w,f} is not Li-Yorke chaotic, we conclude that C_{w,f} must be power-bounded in \mathfrak{X} . ∎ Theorem 10 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . If for every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \ \liminf_{n\to\infty}\mu_{n}(f^{-n}(B))=0, then the following assertions are equivalent: (i) C_{w,f} is Li-Yorke chaotic; (ii) C_{w,f} has a residual set of irregular vectors; (iii) C_{w,f} is not power-bounded. Proof. Let \mathfrak{X}_{0} be the set of all simple functions of the form \sum_{k=1}^{m}b_{k}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{k}} , where b_{1},\ldots,b_{m} are scalars and B_{1},\ldots,B_{m} are measurable sets of finite \mu -measure satisfying \liminf_{n\to\infty}\mu_{n}(f^{-n}(B_{1}\cup\ldots\cup B_{m}))=0. (14) The assumption implies that \mathfrak{X}_{0} is dense in L^{p}(\mu) . Moreover, by ( 14 ), \liminf_{n\to\infty}(C_{w,f})^{n}(\varphi)=0\ \text{ for all }\varphi\in\mathfrak{X}_{0}. Hence, the set \mathcal{R}_{1} of all \varphi\in L^{p}(\mu) whose orbit under C_{w,f} has a subsequence converging to zero is residual in L^{p}(\mu) . If (iii) holds, then the set \mathcal{R}_{2} of all \varphi\in L^{p}(\mu) whose orbit under C_{w,f} is unbounded is also residual in L^{p}(\mu) , which implies (ii). The other implications are clear. ∎ Remark 11 . (a) In the case L^{p}(\mu) is separable, [ 8 , Theorem 10] says that (ii) is equivalent to (ii’) C_{w,f} is densely Li-Yorke chaotic. (b) If the space L^{p}(\mu) is separable and there is an increasing sequence (n_{j})_{j\in\mathbb{N}} of positive integers such that, for every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \ \lim_{j\to\infty}\mu_{n_{j}}(f^{-n_{j}}(B))=0, then [ 8 , Theorem 31] implies that (i)–(iii) are also equivalent to (iv) C_{w,f} admits a dense irregular manifold. Let us now present some applications to weighted translation operators. Corollary 12 . Let X=[1,\infty) or \mathbb{R} . For any weighted translation operator T_{w} on L^{p}(X) , \|(T_{w})^{n}\|=\|w^{(n)}\|_{\infty}\ \text{ for all }n\in\mathbb{N}. In particular, T_{w} is power-bounded if and only if \sup_{n\in\mathbb{N}}\|w^{(n)}\|_{\infty}<\infty . Proof. Fix n\in\mathbb{N} . In the present case, \mu is the Lebesgue measure (which is translation invariant) and f:x\in X\mapsto x+1\in X . By Theorem 6 , \|(T_{w})^{n}\|=\sup_{0<\mu(B)<\infty}\Big(\frac{\mu_{n}(f^{-n}(B))}{\mu(B)}\Big)^{\frac{1}{p}}\leq\sup_{0<\mu(B)<\infty}\Big(\frac{\|w^{(n)}\|_{\infty}^{p}\,\mu(f^{-n}(B))}{\mu(B)}\Big)^{\frac{1}{p}}=\|w^{(n)}\|_{\infty}. On the other hand, take 0<\delta<\|w^{(n)}\|_{\infty} (the case \|w^{(n)}\|_{\infty}=0 is trivial) and define B=\{x\in X:|w^{(n)}(x)|\geq\|w^{(n)}\|_{\infty}-\delta\}. Since \mu(B)>0 , we can choose k\in\mathbb{N} such that the set B^{\prime}=B\cap[-k,k] has positive \mu -measure. Hence, by Theorem 6 , \|(T_{w})^{n}\|\geq\Big(\frac{\mu_{n}(f^{-n}(B^{\prime}+n))}{\mu(B^{\prime}+n)}\Big)^{\frac{1}{p}}=\Big(\frac{\mu_{n}(B^{\prime})}{\mu(B^{\prime})}\Big)^{\frac{1}{p}}\geq\|w^{(n)}\|_{\infty}-\delta. By letting \delta\to 0^{+} , we obtain \|(T_{w})^{n}\|\geq\|w^{(n)}\|_{\infty} . ∎ Corollary 13 . For weighted shifts B_{w} on \ell^{p}(\mathbb{N}) and weighted translation operators T_{w} on L^{p}[1,\infty) , properties (i)–(iv) above are always equivalent to each other. Proof. It is enough to observe that the conditions in Remark 11 (b) are satisfied with (n_{j})_{j\in\mathbb{N}} being the full sequence (n)_{n\in\mathbb{N}} . ∎ Corollary 14 . Given a weighted translation operator T_{w} on L^{p}(\mathbb{R}) , let \mathfrak{X} be the closed subspace of L^{p}(\mathbb{R}) generated by the characteristic functions {\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B} such that B is a Lebesgue measurable set in \mathbb{R} of finite Lebesgue measure satisfying \liminf_{n\to\infty}\int_{B}|w(x-n)\cdots w(x-1)|^{p}dx=0. Then, the following assertions are equivalent: (i) T_{w} is Li-Yorke chaotic; (ii) There exists a nonempty countable family (B_{i})_{i\in I} of Lebesgue measurable sets in \mathbb{R} of finite positive Lebesgue measure such that \liminf_{n\to\infty}\int_{B_{i}}|w(x-n)\cdots w(x-1)|^{p}dx=0\ \text{ for all }i\in I and \sup\Big\{\frac{1}{\mu(B_{i})}\int_{B_{i}}|w(x-n)\cdots w(x-1)|^{p}dx:i\in I,n\in\mathbb{N}\Big\}=\infty. (iii) T_{w} is not power-bounded in \mathfrak{X} . Proof. The equivalence (i) \Leftrightarrow (ii) follows from [ 14 , Theorem 10] and [ 14 , Remark 11] . The equivalence (i) \Leftrightarrow (iii) is a restatement of Theorem 9 in the present particular situation. ∎ 3.2 The case of the space C_{0}(\Omega) Theorem 15 . For any weighted composition operator C_{w,f} on C_{0}(\Omega) , \|(C_{w,f})^{n}\|=\|w^{(n)}\|\ \text{ for all }n\in\mathbb{N}. (15) In particular, C_{w,f} is power-bounded if and only if there is a constant C\in(0,\infty) such that \|w^{(n)}\|\leq C\ \text{ for all }n\in\mathbb{N}. (16) Proof. Since \|(C_{w,f})^{n}(\varphi)\|=\|(\varphi\circ f^{n})\cdot w^{(n)}\|\leq\|w^{(n)}\|\|\varphi\|\ \text{ for all }\varphi\in C_{0}(\Omega), we obtain \|(C_{w,f})^{n}\|\leq\|w^{(n)}\| . On the other hand, given \varepsilon>0 , there exists x_{n}\in\Omega such that |w^{(n)}(x_{n})|\geq\|w^{(n)}\|-\varepsilon . Choose an open neighborhood V_{n} of f^{n}(x_{n}) in \Omega such that \overline{V_{n}} is compact. By Uryshon’s lemma, there exists a continuous map \phi_{n}:\Omega\to[0,1] such that \operatorname{supp}\phi_{n}\subset V_{n} and \phi_{n}(f^{n}(x_{n}))=1 . Hence, \phi_{n}\in C_{0}(\Omega) and \|\phi_{n}\|=1 . Since \|(C_{w,f})^{n}(\phi_{n})\|\geq|\phi_{n}(f^{n}(x_{n}))\,w^{(n)}(x_{n})|=|w^{(n)}(x_{n})|\geq\|w^{(n)}\|-\varepsilon, we obtain \|(C_{w,f})^{n}\|\geq\|w^{(n)}\|-\varepsilon . Since \varepsilon>0 is arbitrary, the proof of ( 15 ) is complete. The last assertion follows from ( 15 ). ∎ For weighted shifts B_{w} on c_{0}(\mathbb{N}) and c_{0}(\mathbb{Z}) , the norms of the iterates of B_{w} are also given by formulas ( 7 ) and ( 9 ), respectively. Hence, the conditions for B_{w} to be power-bounded are also given by ( 8 ) and ( 10 ), respectively. Thus, [ 14 , Corollary 4] (see also [ 5 , Proposition 27] ) and [ 14 , Corollary 5] can be rewritten as follows. Corollary 16 . A weighted shift B_{w} on c_{0}(\mathbb{N}) is Li-Yorke chaotic if and only if it is not power-bounded. Corollary 17 . A weighted shift B_{w} on c_{0}(\mathbb{Z}) with nonzero weights is Li-Yorke chaotic if and only if it is not power-bounded and \displaystyle\liminf_{n\to\infty}|w_{-n}\cdots w_{-1}|=0 . Theorem 18 . Given a weighted composition operator C_{w,f} on C_{0}(\Omega) , let \mathfrak{X} be the closed subspace of C_{0}(\Omega) generated by the functions \varphi\in C_{0}(\Omega) whose support is contained in a relatively compact open set B in \Omega satisfying \liminf_{n\to\infty}\|w^{(n)}\|_{f^{-n}(B)}=0. (17) Then, C_{w,f} is Li-Yorke chaotic if and only if it is not power-bounded in \mathfrak{X} . Proof. Suppose that C_{w,f} is Li-Yorke chaotic. By [ 14 , Theorem 1] and [ 14 , Remark 3] , there exist a sequence (B_{i})_{i\in\mathbb{N}} of nonempty relatively compact open sets in \Omega and an increasing sequence (n_{j})_{j\in\mathbb{N}} of positive integers such that \overline{B_{i}}\subset B_{i+1} for all i\in\mathbb{N} , \sup\big\{\|w^{(n)}\|_{f^{-n}(B_{i})}:i,n\in\mathbb{N}\big\}=\infty (18) and \lim_{j\to\infty}\|w^{(n_{j})}\|_{f^{-n_{j}}(B_{i})}=0\ \text{ for all }i\in\mathbb{N}. (19) Since \overline{B_{i}}\subset B_{i+1} , there is a continuous map \phi_{i}:\Omega\to[0,1] such that \operatorname{supp}\phi_{i}\subset B_{i+1} and \phi_{i}=1 on \overline{B_{i}} . By ( 19 ), \phi_{i}\in\mathfrak{X} for all i\in\mathbb{N} . Since \|\phi_{i}\|=1 and \|(C_{w,f})^{n}(\phi_{i})\|\geq\|(\phi_{i}\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B_{i})}=\|w^{(n)}\|_{f^{-n}(B_{i})}\ \ \ (i,n\in\mathbb{N}), it follows from ( 18 ) that C_{w,f} is not power-bounded in \mathfrak{X} . Conversely, suppose that C_{w,f} is not Li-Yorke chaotic. If \varphi\in C_{0}(\Omega) has support contained in a relatively compact open set B in \Omega satisfying ( 17 ), then \liminf_{n\to\infty}\|(C_{w,f})^{n}(\varphi)\|=\liminf_{n\to\infty}\|(\varphi\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B)}\leq\|\varphi\|\liminf_{n\to\infty}\|w^{(n)}\|_{f^{-n}(B)}=0. Since C_{w,f} does not admit a semi-irregular vector, we conclude that \lim_{n\to\infty}(C_{w,f})^{n}(\varphi)=0. This implies that the set \mathcal{R}_{1} of all \varphi\in\mathfrak{X} whose orbit under C_{w,f} has a subsequence converging to zero is dense, hence residual, in \mathfrak{X} . Now, simply continue arguing as at the end of the proof of Theorem 9 to conclude that C_{w,f} is power-bounded in \mathfrak{X} . ∎ Theorem 19 . Consider a weighted composition operator C_{w,f} on C_{0}(\Omega) . If \liminf_{n\to\infty}\|w^{(n)}\|_{f^{-n}(B)}=0 for every relatively compact open set B in \Omega , then the following assertions are equivalent: (i) C_{w,f} is Li-Yorke chaotic; (ii) C_{w,f} has a residual set of irregular vectors; (iii) C_{w,f} is not power-bounded. Proof. Given \varphi\in C_{c}(\Omega) , take a relatively compact open set B in \Omega with \operatorname{supp}\varphi\subset B . By the hypothesis, \liminf_{n\to\infty}\|(C_{w,f})^{n}(\varphi)\|\leq\|\varphi\|\liminf_{n\to\infty}\|w^{(n)}\|_{f^{-n}(B)}=0. Since C_{c}(\Omega) is dense in C_{0}(\Omega) , it follows that the set \mathcal{R}_{1} of all \varphi\in C_{0}(\Omega) whose orbit under C_{w,f} has a subsequence converging to zero is residual in C_{0}(\Omega) . If (iii) holds, then the set \mathcal{R}_{2} of all \varphi\in C_{0}(\Omega) whose orbit under C_{w,f} is unbounded is also residual in C_{0}(\Omega) , which implies (ii). The other implications are clear. ∎ Remark 20 . (a) In the case C_{0}(\Omega) is separable, [ 8 , Theorem 10] says that (ii) is equivalent to (ii’) C_{w,f} is densely Li-Yorke chaotic. (b) If the space C_{0}(\Omega) is separable and there is an increasing sequence (n_{j})_{j\in\mathbb{N}} of positive integers such that \lim_{j\to\infty}\|w^{(n_{j})}\|_{f^{-n_{j}}(B)}=0 for every relatively compact open set B in \Omega , then [ 8 , Theorem 31] implies that (i)–(iii) are also equivalent to (iv) C_{w,f} admits a dense irregular manifold. Let us now present some applications to weighted translation operators. Corollary 21 . Let \Omega=[1,\infty) or \mathbb{R} . For any weighted translation operator T_{w} on C_{0}(\Omega) , \|(T_{w})^{n}\|=\sup_{x\in\Omega}|w(x)\cdots w(x+n-1)|\ \text{ for all }n\in\mathbb{N}. In particular, T_{w} is power-bounded if and only if \sup\{|w(x)\cdots w(x+n-1)|:x\in\Omega,n\in\mathbb{N}\}<\infty. Proof. This is a particular case of Theorem 15 . ∎ Corollary 22 . For weighted shifts B_{w} on c_{0}(\mathbb{N}) and weighted translation operators T_{w} on C_{0}[1,\infty) , properties (i)–(iv) above are always equivalent to each other. Proof. The conditions in Remark 20 (b) are satisfied with (n_{j})_{j\in\mathbb{N}} being the full sequence (n)_{n\in\mathbb{N}} . ∎ Corollary 23 . Given a weighted translation operator T_{w} on C_{0}(\mathbb{R}) , let \mathfrak{X} be the closed subspace of C_{0}(\mathbb{R}) generated by the functions \varphi\in C_{0}(\mathbb{R}) whose support is contained in a bounded open set B in \mathbb{R} satisfying \liminf_{n\to\infty}\Big(\sup_{x\in B}|w(x-n)\cdots w(x-1)|\Big)=0. Then, the following assertions are equivalent: (i) T_{w} is Li-Yorke chaotic; (ii) There exists a sequence (B_{i})_{i\in\mathbb{N}} of bounded open sets in \mathbb{R} such that \overline{B_{i}}\subset B_{i+1}\ \text{ for all }i\in\mathbb{N}, \liminf_{n\to\infty}\Big(\sup_{x\in B_{i}}|w(x-n)\cdots w(x-1)|\Big)=0\ \text{ for all }i\in\mathbb{N} and \sup\Big\{\sup_{x\in B_{i}}|w(x-n)\cdots w(x-1)|:i,n\in\mathbb{N}\Big\}=\infty. (iii) T_{w} is not power-bounded in \mathfrak{X} . Proof. The equivalence (i) \Leftrightarrow (ii) follows from [ 14 , Theorem 1] and [ 14 , Remark 3] . The equivalence (i) \Leftrightarrow (iii) is a restatement of Theorem 18 in the present particular situation. ∎ 4 Distributionally chaotic weighted composition operators Recall that the lower density and the upper density of a set D\subset\mathbb{N} are defined by \operatorname{\underline{dens}}(D)=\liminf_{n\to\infty}\frac{\operatorname{card}(D\cap[1,n])}{n}\ \ \text{ and }\ \ \operatorname{\overline{dens}}(D)=\limsup_{n\to\infty}\frac{\operatorname{card}(D\cap[1,n])}{n}, respectively. Given a metric space M , recall that a map f:M\to M is said to be distributionally chaotic if there exist an uncountable set \Gamma\subset M and \varepsilon>0 such that each pair (x,y) of distinct points in \Gamma is a distributionally chaotic pair for f , in the sense that \operatorname{\underline{dens}}\{n\in\mathbb{N}:d(f^{n}(x),f^{n}(y))<\varepsilon\}=0 and \operatorname{\overline{dens}}\{n\in\mathbb{N}:d(f^{n}(x),f^{n}(y))<\tau\}=1\ \text{ for all }\tau>0. If the set \Gamma can be chosen to be dense in M , then f is densely distributionally chaotic . An extensive study of the concept of distributional chaos in the setting of linear dynamics was developed in [ 5 , 7 , 9 ] . In particular, the following useful characterizations were obtained: For any operator T on any Banach space Y , the following assertions are equivalent: (i) T is distributionally chaotic; (ii) T admits a distributionally irregular vector , that is, a vector y\in Y for which there exist D,E\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=\operatorname{\overline{dens}}(E)=1 such that \lim_{n\in D}T^{n}y=0\ \ \text{ and }\ \ \lim_{n\in E}\|T^{n}y\|=\infty; (iii) T satisfies the Distributional Chaos Criterion , that is, there exist sequences (x_{k})_{k\in\mathbb{N}} and (y_{k})_{k\in\mathbb{N}} in Y such that: (a) There exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that \lim_{n\in D}T^{n}x_{k}=0 for all k . (b) y_{k}\in\overline{\operatorname{span}\{x_{n}:n\in\mathbb{N}\}} for all k\in\mathbb{N} , \|y_{k}\|\to 0 as k\to\infty , and there exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that \operatorname{card}\{1\leq n\leq N_{k}:\|T^{n}y_{k}\|>\varepsilon\}\geq\varepsilon N_{k}\ \text{ for all }k\in\mathbb{N}. Let us also recall that by a distributionally irregular manifold for T we mean a vector subspace of Y consisting, except for the zero vector, of distributionally irregular vectors for T . 4.1 The case of the space L^{p}(\mu) Theorem 24 . A weighted composition operator C_{w,f} on L^{p}(\mu) is distributionally chaotic if and only if there exists a nonempty countable family (B_{i})_{i\in I} of measurable sets of finite positive \mu -measure such that the following properties hold: (a) There exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that \lim_{n\in D}\mu_{n}(f^{-n}(B_{i}))=0\ \text{ for all }i\in I. (b) There exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that, for each k\in\mathbb{N} , there are r\in\mathbb{N} , i_{1},\ldots,i_{r}\in I and b_{1},\ldots,b_{r}\in(0,\infty) with \operatorname{card}\Big\{1\leq n\leq N_{k}:\frac{b_{1}\mu_{n}(f^{-n}(B_{i_{1}}))+\cdots+b_{r}\mu_{n}(f^{-n}(B_{i_{r}}))}{b_{1}\mu(B_{i_{1}})+\cdots+b_{r}\mu(B_{i_{r}})}>k\Big\}\geq\varepsilon N_{k}. Proof. ( \Rightarrow ): Let \varphi\in L^{p}(\mu) be a distributionally irregular vector for C_{w,f} . There exist D,E\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=\operatorname{\overline{dens}}(E)=1 such that \lim_{n\in D}\|(C_{w,f})^{n}(\varphi)\|_{p}=0\ \ \text{ and }\ \ \lim_{n\in E}\|(C_{w,f})^{n}(\varphi)\|_{p}=\infty. Consider the measurable sets B_{i}=\{x\in X:2^{i-1}\leq|\varphi(x)|<2^{i}\}\ \ \ \ \ (i\in\mathbb{Z}) and let I=\{i\in\mathbb{Z}:\mu(B_{i})>0\} . Since 2^{(i-1)p}\mu_{n}(f^{-n}(B_{i}))\leq\|(C_{w,f})^{n}(\varphi)\|_{p}^{p}\,, we have that property (a) holds. On the other hand, for each k\in\mathbb{N} , since \operatorname{\overline{dens}}\{n\in\mathbb{N}:\|(C_{w,f})^{n}(\varphi)\|_{p}^{p}>2^{p}k\|\varphi\|_{p}^{p}\}\geq\operatorname{\overline{dens}}(E)=1, there exists N_{k}\in\mathbb{N} such that \operatorname{card}\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}(\varphi)\|_{p}^{p}>2^{p}k\|\varphi\|_{p}^{p}\}\geq N_{k}\Big(1-\frac{1}{2k}\Big). Moreover, it is clear that the N_{k} ’s can be chosen so that the sequence (N_{k})_{k\in\mathbb{N}} is increasing. Fix k\in\mathbb{N} and let J=\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}(\varphi)\|_{p}^{p}>2^{p}k\|\varphi\|_{p}^{p}\}. Since 2^{p}k\|\varphi\|_{p}^{p}<\|(C_{w,f})^{n}(\varphi)\|_{p}^{p}\leq\sum_{i\in\mathbb{Z}}2^{ip}\mu_{n}(f^{-n}(B_{i}))\ \text{ for all }n\in J, we can take j\in\mathbb{N} large enough so that \sum_{i=-j}^{j}2^{ip}\mu_{n}(f^{-n}(B_{i}))>2^{p}k\|\varphi\|_{p}^{p}\geq k\sum_{i=-j}^{j}2^{ip}\mu(B_{i})\ \text{ for all }n\in J. This shows that property (b) holds with \varepsilon=1/2 . ( \Leftarrow ): By property (b), for each k\in\mathbb{N} , there are r_{k}\in\mathbb{N} , i_{k,1},\ldots,i_{k,r_{k}}\in I and b_{k,1},\ldots,b_{k,r_{k}}\in(0,\infty) such that \operatorname{card}\Big\{1\leq n\leq N_{k}:\frac{b_{k,1}\mu_{n}(f^{-n}(B_{i_{k,1}}))+\cdots+b_{k,r_{k}}\mu_{n}(f^{-n}(B_{i_{k,r_{k}}}))}{b_{k,1}\mu(B_{i_{k,1}})+\cdots+b_{k,r_{k}}\mu(B_{i_{k,r_{k}}})}>k\Big\}\geq\varepsilon N_{k}. (20) For each k\in\mathbb{N} , take pairwise disjoint measurable sets A_{k,1},\ldots,A_{k,s_{k}} such that A_{k,1}\cup\ldots\cup A_{k,s_{k}}=B_{i_{k,1}}\cup\ldots\cup B_{i_{k,r_{k}}} and each B_{i_{k,j}} is a union of some of the A_{k,\ell} ’s. It follows from property (a) that \lim_{n\in D}\,(C_{w,f})^{n}({\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k,\ell}})=0\ \text{ for all }k\in\mathbb{N}\text{ and }\ell\in\{1,\ldots,s_{k}\}. For each k\in\mathbb{N} and each \ell\in\{1,\ldots,s_{k}\} , let a_{k,\ell}=\sum_{j}b_{k,j} , where the sum is taken over all j\in\{1,\ldots,r_{k}\} satisfying A_{k,\ell}\subset B_{i_{k,j}} . Consider the measurable simple functions s_{k}=\frac{(a_{k,1})^{\frac{1}{p}}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k,1}}+\cdots+(a_{k,s_{k}})^{\frac{1}{p}}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k,s_{k}}}}{k^{\frac{1}{p}}\big(a_{k,1}\mu(A_{k,1})+\cdots+a_{k,s_{k}}\mu(A_{k,s_{k}})\big)^{\frac{1}{p}}}\ \ \ \ \ (k\in\mathbb{N}). Clearly, \|s_{k}\|_{p}\to 0 as k\to\infty . Moreover, since \displaystyle\|(C_{w,f})^{n}(s_{k})\|_{p}^{p} \displaystyle=\frac{a_{k,1}\mu_{n}(f^{-n}(A_{k,1}))+\cdots+a_{k,s_{k}}\mu_{n}(f^{-n}(A_{k,s_{k}}))}{k\,\big(a_{k,1}\mu(A_{k,1})+\cdots+a_{k,s_{k}}\mu(A_{k,s_{k}})\big)} \displaystyle=\frac{b_{k,1}\mu_{n}(f^{-n}(B_{i_{k,1}}))+\cdots+b_{k,r_{k}}\mu_{n}(f^{-n}(B_{i_{k,r_{k}}}))}{k\,\big(b_{k,1}\mu(B_{i_{k,1}})+\cdots+b_{k,r_{k}}\mu(B_{i_{k,r_{k}}})\big)}\,, it follows from ( 20 ) that \operatorname{card}\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}(s_{k})\|_{p}>1\}\geq\varepsilon N_{k}. This shows that the operator C_{w,f} satisfies the Distributional Chaos Criterion, and so it is distributionally chaotic. ∎ Remark 25 . Note that the countable family (B_{i})_{i\in I} constructed in the proof of Theorem 24 has the additional property that its terms are pairwise disjoint. The next corollaries will give a simple necessary condition and a simple sufficient condition for C_{w,f} to be distributionally chaotic. Corollary 26 . If a weighted composition operator C_{w,f} on L^{p}(\mu) is distributionally chaotic, then there exists a nonempty countable family (B_{i})_{i\in I} of pairwise disjoint measurables sets of finite positive \mu -measure such that • there exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 and \displaystyle\lim_{n\in D}\mu_{n}(f^{-n}(B_{i}))=0 for all i\in I , • for every constant C>0 , \displaystyle\operatorname{\overline{dens}}\Big\{n\in\mathbb{N}:\sup_{i\in I}\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}\geq C\Big\}>0 . Proof. Let (B_{i})_{i\in I} be the family given by Theorem 24 . We may assume that the B_{i} ’s are pairwise disjoint (Remark 25 ). For each k\in\mathbb{N} , let J_{k}\subset\{1,\ldots,N_{k}\} be the set that appears in (b). Let J=\bigcup_{k=1}^{\infty}J_{k} . Since \operatorname{card}J_{k}\geq\varepsilon N_{k} for all k\in\mathbb{N} , we have that \operatorname{\overline{dens}}(J)\geq\varepsilon. Fix C>0 and define H=\Big\{n\in\mathbb{N}:\sup_{i\in I}\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}\geq C\Big\}. If n\in J_{k}\backslash H and we use the notation in (b), then k<\frac{b_{1}\mu_{n}(f^{-n}(B_{i_{1}}))+\cdots+b_{r}\mu_{n}(f^{-n}(B_{i_{r}}))}{b_{1}\mu(B_{i_{1}})+\cdots+b_{r}\mu(B_{i_{r}})}0 . ∎ Corollary 27 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . If there is a nonempty countable family (B_{i})_{i\in I} of measurable sets of finite positive \mu -measure such that • there exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 and \displaystyle\lim_{n\in D}\mu_{n}(f^{-n}(B_{i}))=0 for all i\in I , • there exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that, for each k\in\mathbb{N} , there exists i\in I with \operatorname{card}\Big\{1\leq n\leq N_{k}:\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}>k\Big\}\geq\varepsilon N_{k}, then C_{w,f} is distributionally chaotic. Proof. It follows immediately from Theorem 24 . ∎ The fact that the weighted shift on \ell^{p}(\mathbb{N}) considered in the next example is distributionally chaotic is due to Frédéric Bayart and appeared in [ 10 , Theorem 27] . In order to illustrate Theorem 24 in a concrete situation, we will give below a different proof of this fact. Example 28 . The weighted shift B_{w} on \ell^{p}(\mathbb{N}) with weights w_{n}=\big(\frac{n+1}{n}\big)^{\frac{1}{p}} is distributionally chaotic. Proof. Regard B_{w} as C_{w,f} by considering X=\mathbb{N} , \mathfrak{M}=\mathcal{P}(X) , \mu the counting measure on \mathfrak{M} and f:n\in\mathbb{N}\to n+1\in\mathbb{N} . Let (n_{k})_{k\in\mathbb{N}} be an increasing sequence of positive integers such that \ln n_{k}>k\ln k for every k\in\mathbb{N} . We define I as the set of all integers n with n_{k}\leq n\leq k\,n_{k} for some k\in\mathbb{N} and put B_{i}=\{i\} for each i\in I . Since condition (a) in Theorem 24 holds trivially, let us establish condition (b). Let N_{k}=k\,n_{k} for each k\in\mathbb{N} . Given k\geq 2 , let r=(k-1)n_{k} , i_{j}=n_{k}+j and b_{j}=\frac{1}{i_{j}} ( j\in\{1,\ldots,r\} ). Then b_{1}\mu(B_{i_{1}})+\cdots+b_{r}\mu(B_{i_{r}})=b_{1}+\cdots+b_{r}=\sum_{j=n_{k}+1}^{k\,n_{k}}\frac{1}{j}\leq\ln(k\,n_{k})-\ln n_{k}=\ln k. Moreover, for every n\in\{n_{k},\ldots,(k-1)n_{k}\} , b_{1}\mu_{n}(f^{-n}(B_{i_{1}}))+\cdots+b_{r}\mu_{n}(f^{-n}(B_{i_{r}}))\geq\sum_{j=1}^{n_{k}}\frac{1}{j}\geq\ln n_{k}. Thus, \operatorname{card}\Big\{1\leq n\leq N_{k}:\frac{b_{1}\mu_{n}(f^{-n}(B_{i_{1}}))+\cdots+b_{r}\mu_{n}(f^{-n}(B_{i_{r}}))}{b_{1}\mu(B_{i_{1}})+\cdots+b_{r}\mu(B_{i_{r}})}>k\Big\}\geq\Big(\frac{k-2}{k}\Big)N_{k}. Hence, by Theorem 24 , B_{w} is distributinally chaotic. ∎ For general weighted shifts on \ell^{p}(\mathbb{N}) , Theorem 24 gives us the following characterization. Corollary 29 . A weighted shift B_{w} on \ell^{p}(\mathbb{N}) is distributionally chaotic if and only if there exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that, for each k\in\mathbb{N} , there are r\in\mathbb{N} , i_{1},\ldots,i_{r}\in\mathbb{N} and b_{1},\ldots,b_{r}\in(0,\infty) with \operatorname{card}\Big\{1\leq n\leq N_{k}:\frac{1}{b_{1}+\cdots+b_{r}}\sum_{1\leq j\leq r,i_{j}>n}b_{j}|w_{i_{j}-n}\cdots w_{i_{j}-1}|^{p}>k\Big\}\geq\varepsilon N_{k}. Proof. This corollary is essentially a restatement of Theorem 24 in the present particular situation. For the sufficience of the condition, it is enough to consider B_{i}=\{i\} for each i\in\mathbb{N} . Since \frac{b_{1}\mu_{n}(f^{-n}(B_{i_{1}}))+\cdots+b_{r}\mu_{n}(f^{-n}(B_{i_{r}}))}{b_{1}\mu(B_{i_{1}})+\cdots+b_{r}\mu(B_{i_{r}})}=\frac{1}{b_{1}+\cdots+b_{r}}\sum_{1\leq j\leq r,i_{j}>n}b_{j}|w_{i_{j}-n}\cdots w_{i_{j}-1}|^{p}, the hypothesis implies that condition (b) in Theorem 24 holds; therefore, B_{w} is distributionally chaotic. For the converse, it is enough to decompose each set B_{i} given by Theorem 24 as a union of singletons. ∎ In a similar way, Theorem 24 implies the following result for weighted shifts on \ell^{p}(\mathbb{Z}) . Corollary 30 . A weighted shift B_{w} on \ell^{p}(\mathbb{Z}) is distributionally chaotic if and only if there exists a set S\subset\mathbb{Z} such that the following properties hold: (a) There exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 and \displaystyle\lim_{n\in D}|w_{i-n}\cdots w_{i-1}|=0 for all i\in S . (b) There exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that, for each k\in\mathbb{N} , there are r\in\mathbb{N} , i_{1},\ldots,i_{r}\in S and b_{1},\ldots,b_{r}\in(0,\infty) with \operatorname{card}\Big\{1\leq n\leq N_{k}:\frac{b_{1}|w_{i_{1}-n}\cdots w_{i_{1}-1}|^{p}+\cdots+b_{r}|w_{i_{r}-n}\cdots w_{i_{r}-1}|^{p}}{b_{1}+\cdots+b_{r}}>k\Big\}\geq\varepsilon N_{k}. Remark 31 . If the weights w_{n} are nonzero and \lim_{n\to\infty}\prod_{j=-n+1}^{0}w_{j}=0, then condition (a) above holds for S=\mathbb{Z} and D=\mathbb{N} . Hence, in this case, the weighted shift B_{w} on \ell^{p}(\mathbb{Z}) is distributionally chaotic if and only if condition (b) above holds for S=\mathbb{Z} . Below we will obtain some additional sufficient conditions for C_{w,f} to be distributionally chaotic. Theorem 32 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . Suppose that there is a sequence (A_{k})_{k\in\mathbb{N}} of measurable sets of finite \mu -measure such that the following properties hold: (a) There exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that, for each k\in\mathbb{N} and each \delta>0 , there is a measurable set B\subset A_{k} with \mu(A_{k}\backslash B)<\delta\ \ \text{ and }\ \ \lim_{n\in D}\mu_{n}(f^{-n}(B))=0. (b) \displaystyle\lim_{k\to\infty}\mu(A_{k})=0 and there exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that \operatorname{card}\{1\leq n\leq N_{k}:\mu_{n}(f^{-n}(A_{k}))>\varepsilon\}\geq\varepsilon N_{k}\ \text{ for all }k\in\mathbb{N}. Then, C_{w,f} is distributionally chaotic. Proof. By property (a), for every k,r\in\mathbb{N} , there is a measurable set B_{k,r}\subset A_{k} such that \mu(A_{k}\backslash B_{k,r})<\frac{1}{r}\ \ \text{ and }\ \ \lim_{n\in D}\mu_{n}(f^{-n}(B_{k,r}))=0. Hence, {\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k}}\in\overline{\operatorname{span}\{{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{j,r}}:j,r\in\mathbb{N}\}} for all k\in\mathbb{N} , and \lim_{n\in D}(C_{w,f})^{n}({\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{k,r}})=0\ \text{ for all }k,r\in\mathbb{N}. Moreover, by property (b), \|{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k}}\|_{p}\to 0 as k\to\infty , and \operatorname{card}\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}({\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k}})\|_{p}>\varepsilon^{\frac{1}{p}}\}\geq\varepsilon N_{k}\ \text{ for all }k\in\mathbb{N}. This shows that C_{w,f} satisfies the Distributional Chaos Criterion. ∎ Theorem 33 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) with positive weight function w:X\to(0,\infty) . Suppose that there is a sequence (A_{k})_{k\in\mathbb{N}} of measurable sets of finite positive \mu -measure such that the following properties hold: (a) There exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that \lim_{n\in D}\mu_{n}(f^{-n}(A_{k}))=0\ \text{ for all }k\in\mathbb{N}. (b) There exists a set E\subset\mathbb{N} with \operatorname{\overline{dens}}(E)>0 such that \sum_{n\in E}\left(\frac{\mu(A_{n})}{\mu_{n}(f^{-n}(A_{n}))}\right)^{\frac{1}{p}}<\infty. Then, C_{w,f} is distributionally chaotic. Proof. We will apply the Distributional Chaos Criterion. For condition (a) in the criterion, we define x_{k}={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k}} and observe that \lim_{n\in D}\|(C_{w,f})^{n}(x_{k})\|_{p}=\lim_{n\in D}\mu_{n}(f^{-n}(A_{k}))^{\frac{1}{p}}=0\ \text{ for all }k\in\mathbb{N}, because of hypothesis (a). By using hypothesis (b), we will prove the existence of a vector y\in\overline{\operatorname{span}\{x_{k}:k\in\mathbb{N}\}} such that \lim_{n\in E}\|(C_{w,f})^{n}(y)\|_{p}=\infty. In view of [ 7 , Proposition 8] , this will imply that condition (b) in the criterion also holds, which will complete the proof. First, let us prove that hypothesis (b) implies the existence of a sequence (c_{n})_{n\in\mathbb{N}} of non-negative scalars such that \sum_{n\in E}c_{n}\,\mu(A_{n})^{\frac{1}{p}}<\infty\ \ \text{ and }\ \ \lim_{n\in E}c_{n}\,\mu_{n}(f^{-n}(A_{n}))^{\frac{1}{p}}=\infty. For this purpose, we will use an argument similar to one used in [ 18 ] . Let \displaystyle a_{n} \displaystyle=\left\{\begin{array}[]{cl}\left(\frac{\mu(A_{n})}{\mu_{n}(f^{-n}(A_{n}))}\right)^{\frac{1}{p}}&\text{ if }n\in E\\ 0&\text{ if }n\notin E,\end{array}\right. \displaystyle r_{n} \displaystyle=\displaystyle\sum_{i\geq n}a_{i}\,, \displaystyle c_{n} \displaystyle=\left\{\begin{array}[]{cl}\frac{1}{\sqrt{r_{n}}\,\mu_{n}(f^{-n}(A_{n}))^{\frac{1}{p}}}&\text{ if }n\in E\\ 0&\text{ if }n\notin E.\end{array}\right. Then, \lim_{n\in E}c_{n}\,\mu_{n}(f^{-n}(A_{n}))^{\frac{1}{p}}=\lim_{n\in E}\frac{1}{\sqrt{r_{n}}}=\infty, since \sum_{n\in E}a_{n} is convergent. Moreover, we have that \sum_{n\in\mathbb{N}}\frac{a_{n}}{\sqrt{r_{n}}}\leq 2\sqrt{r_{1}}\,, since \frac{a_{n}}{\sqrt{r_{n}}}=\frac{(\sqrt{r_{n}}+\sqrt{r_{n+1}})(\sqrt{r_{n}}-\sqrt{r_{n+1}})}{\sqrt{r_{n}}}\leq 2(\sqrt{r_{n}}-\sqrt{r_{n+1}}). Hence, \sum_{n\in E}c_{n}\,\mu(A_{n})^{\frac{1}{p}}=\sum_{n\in E}\frac{a_{n}}{\sqrt{r_{n}}}=\sum_{n\in\mathbb{N}}\frac{a_{n}}{\sqrt{r_{n}}}\leq 2\sqrt{r_{1}}\leq 2\sqrt{\sum_{n\in E}\left(\frac{\mu(A_{n})}{\mu_{n}(f^{-n}(A_{n}))}\right)^{\frac{1}{p}}}<\infty. Now, by defining y=\displaystyle\sum_{n\in E}c_{n}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{n}} , we have that \|y\|_{p}=\Big\|\sum_{n\in E}c_{n}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{n}}\Big\|_{p}\leq\sum_{n\in E}c_{n}\,\mu(A_{n})^{\frac{1}{p}}<\infty. Thus, y\in\overline{\operatorname{span}\{x_{k}:k\in\mathbb{N}\}} . Moreover, \displaystyle\lim_{n\in E}\|(C_{w,f})^{n}(y)\|_{p} \displaystyle=\lim_{n\in E}\Big\|\sum_{k\in E}c_{k}\,(C_{w,f})^{n}({\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{k}})\Big\|_{p} \displaystyle\geq\lim_{n\in E}\|c_{n}\,(C_{w,f})^{n}({\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A_{n}})\|_{p} \displaystyle=\lim_{n\in E}c_{n}\,\mu_{n}(f^{-n}(A_{n}))^{\frac{1}{p}}=\infty. This completes the proof. ∎ Corollary 34 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) with positive weight function w:X\to(0,\infty) . If there is a measurable set A of finite positive \mu -measure such that • there exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 and \displaystyle\lim_{n\in D}\mu_{n}(f^{-n}(A))=0 , • there exists E\subset\mathbb{N} with \operatorname{\overline{dens}}(E)>0 and \displaystyle\sum_{n\in E}\frac{1}{\mu_{n}(f^{-n}(A))^{\frac{1}{p}}}<\infty , then C_{w,f} is distributionally chaotic. In the sequel we will present several sufficient conditions for C_{w,f} to be densely distributionally chaotic. Theorem 35 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . If the space L^{p}(\mu) is separable and for every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \lim_{n\to\infty}\mu_{n}(f^{-n}(B))=0, then the following assertions are equivalent: (i) C_{w,f} is distributionally chaotic; (ii) C_{w,f} is densely distributionally chaotic; (iii) C_{w,f} admits a dense distributionally irregular manifold; (iv) There exist \phi\in L^{p}(\mu) and \delta>0 such that \operatorname{\overline{dens}}\{n\in\mathbb{N}:\|(C_{w,f})^{n}(\phi)\|_{p}\geq\delta\}>0. Proof. (i) \Rightarrow (iv): Take a distributionally irregular vector \phi\in L^{p}(\mu) for C_{w,f} . (iv) \Rightarrow (iii): Let \mathfrak{X}_{0} be the set of all simple functions of the form \sum_{j=1}^{m}b_{j}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{j}} , where b_{1},\ldots,b_{m} are scalars and each B_{j} is a measurable set of finite \mu -measure satisfying \lim_{n\to\infty}\mu_{n}(f^{-n}(B_{j}))=0. By the assumption, \mathfrak{X}_{0} is dense in L^{p}(\mu) . Moreover, \lim_{n\to\infty}(C_{w,f})^{n}(\varphi)=0\ \text{ for all }\varphi\in\mathfrak{X}_{0}. Hence, by (iv) and [ 9 , Theorem 33] , (iii) holds. (iii) \Rightarrow (ii) \Rightarrow (i): Obvious. ∎ Corollary 36 . Assume the hypotheses of the previous theorem. If there exist a constant C>0 and a measurable set A of finite positive \mu -measure such that \operatorname{\overline{dens}}\{n\in\mathbb{N}:\mu_{n}(f^{-n}(A))\geq C\}>0, then C_{w,f} is densely distributionally chaotic. Proof. Property (iv) in the previous theorem holds with \phi={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{A} and \delta=C^{\frac{1}{p}} . ∎ Recall that a measurable system is a 4 -tuple (X,\mathfrak{M},\mu,f) such that: ( \alpha ) (X,\mathfrak{M},\mu) is a \sigma -finite measure space with \mu(X)>0 ; ( \beta ) f:X\to X is a non-singular bimeasurable bijective map (where f non-singular means that \mu(f^{-1}(B))=0 if and only if \mu(B)=0 ); ( \gamma ) there exists c>0 such that \mu(f^{-1}(B))\leq c\,\mu(B) for every B\in\mathfrak{M} . Note that condition ( \gamma ) implies that the composition operator C_{f}(\varphi)=\varphi\circ f is a well-defined bounded linear operator on L^{p}(\mu) . Recall also that a measurable system (X,\mathfrak{M},\mu,f) (equivalently, the map f ) is said to be: • conservative if for each measurable set B of positive \mu -measure, there exists n\in\mathbb{N} such that \mu(B\cap f^{-n}(B))>0 ; • dissipative if there is a measurable set W (called a wandering set ) such that the sets f^{n}(W) , n\in\mathbb{Z} , are pairwise disjoint and X=\bigcup_{n\in\mathbb{Z}}f^{n}(W) . Finally, recall that a set A\subset X is said to be f -invariant if f^{-1}(A)=A . The following classic result can be found in [ 26 , Theorem 3.2] : Hopf Decomposition Theorem. If (X,\mathfrak{M},\mu) is a \sigma -finite measure space and f:X\to X is a non-singular measurable map, then X can be written as the union of two disjoint f -invariant sets \mathcal{C}(f) and \mathcal{D}(f) , called the conservative and the dissipative parts of f , respectively, such that f|_{\mathcal{C}(f)} is conservative and f|_{\mathcal{D}(f)} is dissipative. Theorem 37 . Consider a measurable system (X,\mathfrak{M},\mu,f) and assume the following conditions: (a) For every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \lim_{n\to\infty}\mu(f^{-n}(B))=0. (b) There is a measurable set B^{\prime}\subset W\subset\mathcal{D}(f) of finite positive \mu -measure, W being a wandering set of \mathcal{D}(f) , such that \sum_{n\in\mathbb{Z}}\mu(f^{n}(B^{\prime}))\text{ converges}. Then, the composition operator C_{f} is densely distributionally chaotic. Proof. Let \mathfrak{X}_{0} be the set of all simple functions of the form \sum_{j=1}^{m}b_{j}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{j}} , where b_{1},\ldots,b_{m} are scalars and each B_{j} is a measurable set of finite \mu -measure satisfying \lim_{n\to\infty}\mu(f^{-n}(B_{j}))=0. By property (a), \mathfrak{X}_{0} is dense in L^{p}(\mu) . Moreover, \lim_{n\to\infty}(C_{f})^{n}(\varphi)=0 for all \varphi\in\mathfrak{X}_{0} . Therefore, by [ 7 , Theorem 19] , it remains to show that there exist a subset Y of L^{p}(\mu) , a map S:Y\to Y with C_{f}(S(\varphi))=\varphi on Y , and a vector \phi\in Y\backslash\{0\} such that the series \sum_{n=1}^{\infty}(C_{f})^{n}(\phi)\ \text{ and }\sum_{n=1}^{\infty}S^{n}(\phi)\ \text{ converge unconditionally}. For this purpose, we put Y=\Big\{{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B}:B\in\mathfrak{M}\text{ and }\sum_{n\in\mathbb{Z}}\mu(f^{n}(B))\text{ converges}\Big\} and \phi={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B^{\prime}}\in Y\backslash\{0\} . Since B^{\prime}\subset W\subset\mathcal{D}(f) and W is a wandering set, we have that the sets f^{n}(B^{\prime}) , n\in\mathbb{Z} , are pairwise disjoint. Hence, the convergence of the series in (b) implies that, for any increasing sequence (n_{j})_{j\in\mathbb{N}} of positive integers, {\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{\bigcup_{j=1}^{m}f^{-n_{j}}(B^{\prime})}\in L^{p}(\mu)\ \text{ for all }m\in\mathbb{N} and \sum_{j=1}^{m}(C_{f})^{n_{j}}(\phi)=\sum_{j=1}^{m}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{f^{-n_{j}}(B^{\prime})}={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{\bigcup_{j=1}^{m}f^{-n_{j}}(B^{\prime})}\to{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{\bigcup_{j=1}^{\infty}f^{-n_{j}}(B^{\prime})}\text{ in }L^{p}(\mu)\text{ as }m\to\infty. Hence, the series \sum_{j=1}^{\infty}(C_{f})^{n_{j}}(\phi) converges in L^{p}(\mu) . This implies that the series \sum_{n=1}^{\infty}(C_{f})^{n}(\phi) is unconditionally convergent. Now, we define S:Y\to Y by S(\varphi)={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{f(B)},\text{ for each }\varphi={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B}\in Y. Clearly, C_{f}(S(\varphi))=\varphi . By arguing as above, it follows that the series \sum_{n=1}^{\infty}S^{n}(\phi)=\sum_{n=1}^{\infty}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{f^{n}(B^{\prime})}\ \text{ converges unconditionally}. This completes the proof. ∎ Corollary 38 . Let (X,\mathfrak{M},\mu,f) be a non-conservative measurable system with \mu(X)<\infty . If for every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon and \lim_{n\to\infty}\mu(f^{-n}(B))=0 , the C_{f} is densely distributionally chaotic. Proof. Since the measurable system is not conservative, there is a measurable set B^{\prime}\subset W\subset\mathcal{D}(f) with \mu(B^{\prime})>0 . Since \sum_{n\in\mathbb{Z}}\mu(f^{n}(B^{\prime}))\leq\mu(X)<\infty, the result follows from the previous theorem. ∎ Theorem 39 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) with positive weight function w:X\to(0,\infty) . Suppose that the measure space (X,\mathfrak{M},\mu) is \sigma -finite, the bimeasurable map f is injective and non-singular, and the following properties hold: (a) For every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \lim_{n\to\infty}\mu_{n}(f^{-n}(B))=0. (b) There is a measurable set B^{\prime}\subset W\subset\mathcal{D}(f) of finite positive \mu -measure, W being a wandering set of \mathcal{D}(f) , and a subset D of \mathbb{N} with \operatorname{\overline{dens}}(D)>0 such that \sum_{n\in D}\int_{f^{n}(B^{\prime})}\left(\prod_{j=0}^{n-1}(w\circ f^{j-n})\right)^{-p}d\mu<\infty. Then, C_{w,f} is densely distributionally chaotic. Proof. For each n\in\mathbb{N} , we define v_{n}:X\to(0,\infty) by v_{n}=\prod_{j=0}^{n-1}(w\circ f^{j-n})\text{ on }f^{n}(X)\ \ \ \text{ and }\ \ \ v_{n}=1\text{ on }X\backslash f^{n}(X). Note that v_{n} is well-defined because f is injective. For each k\in\mathbb{N} , let \phi_{k}=\sum_{n\in D,n\geq k}\frac{1}{v_{n}}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{f^{n}(B^{\prime})}. The convergence of the series in (b) implies that each \phi_{k} belongs to L^{p}(\mu) and \phi_{k}\to 0 as k\to\infty . Moreover, (C_{w,f})^{n}(\phi_{k})=(C_{w,f})^{n}\Big(\frac{1}{v_{n}}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{f^{n}(B^{\prime})}\Big)+\cdots={\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B^{\prime}}+\cdots, whenever n\in D and n\geq k . Hence, by defining \varepsilon=\mu(B^{\prime})^{\frac{1}{p}}>0 , we have that \|(C_{w,f})^{n}(\phi_{k})\|_{p}>\varepsilon\ \text{ for all }n\in D\text{ with }n\geq k. Since \delta=\operatorname{\overline{dens}}(D)>0 , there is an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that \operatorname{card}\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}(\phi_{k})\|_{p}>\varepsilon\}\geq\frac{\delta}{2}\,N_{k}\ \text{ for all }k\in\mathbb{N}. By [ 7 , Proposition 8] , C_{w,f} admits a distributionally unbounded orbit. Thus, by [ 7 , Theorem 15] , C_{w,f} is densely distributionally chaotic. ∎ Corollary 40 . Consider a weighted shift B_{w} on \ell^{p}(\mathbb{N}) with positive weights. If there exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)>0 such that \sum_{n\in D}\left(\prod_{j=1}^{n}w_{j}\right)^{-p}<\infty, then B_{w} is densely distributionally chaotic. Proof. In the present case, \mu is the counting measure on \mathbb{N} and f:n\in\mathbb{N}\to n+1\in\mathbb{N} . Condition (a) in Theorem 39 is trivially true. By defining B^{\prime}=\{1\} , we have that B^{\prime} is a wandering set with finite positive \mu -measure such that \sum_{n\in D}\int_{f^{n}(B^{\prime})}\left(\prod_{j=0}^{n-1}(w\circ f^{j-n})\right)^{-p}d\mu=\sum_{n\in D}\left(\prod_{j=1}^{n}w_{j}\right)^{-p}<\infty. Hence, condition (b) in Theorem 39 is also true. Thus, B_{w} is densely distributionally chaotic. ∎ Corollary 41 . Consider a weighted shift B_{w} on \ell^{p}(\mathbb{Z}) with positive weights. If \lim_{n\to\infty}\prod_{j=-n+1}^{0}w_{j}=0 and there exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)>0 such that \sum_{n\in D}\left(\prod_{j=1}^{n}w_{j}\right)^{-p}<\infty, then B_{w} is densely distributionally chaotic. Proof. In the present case, \mu is the counting measure on \mathbb{Z} and f:n\in\mathbb{Z}\to n+1\in\mathbb{Z} . Since \lim_{n\to\infty}(w_{-n+1}\cdots w_{0})=0 , for every finite set A\subset\mathbb{Z} , \lim_{n\to\infty}\mu_{n}(f^{-n}(A))=\lim_{n\to\infty}\sum_{i\in A}\mu_{n}(f^{-n}(\{i\}))=\lim_{n\to\infty}\sum_{i\in A}(w_{-n+i}\cdots w_{-1+i})=0. Now, B^{\prime}=\{1\} is a wandering set with finite positive \mu -measure such that \sum_{n\in D}\int_{f^{n}(B^{\prime})}\left(\prod_{j=0}^{n-1}(w\circ f^{j-n})\right)^{-p}d\mu=\sum_{n\in D}\left(\prod_{j=1}^{n}w_{j}\right)^{-p}<\infty. Thus, by Theoren 39 , B_{w} is densely distributionally chaotic. ∎ Let us mention that reference [ 27 ] contains some sufficient conditions for weighted shifts on Köthe sequence spaces to be distributionally chaotic. 4.2 The case of the space C_{0}(\Omega) Theorem 42 . A weighted composition operator C_{w,f} on C_{0}(\Omega) is distributionally chaotic if and only if there exists a sequence (B_{i})_{i\in\mathbb{N}} of relatively compact open sets in \Omega such that the following properties hold: (a) There exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that \lim_{n\in D}\|w^{(n)}\|_{f^{-n}(B_{i})}=0\ \text{ for all }i\in\mathbb{N}. (b) There exist \varepsilon>0 and an increasing sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that, for each k\in\mathbb{N} , there exists i\in\mathbb{N} with \operatorname{card}\{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(B_{i})}>k\}\geq\varepsilon N_{k}. Proof. ( \Rightarrow ): Let \varphi\in C_{0}(\Omega) be a distributionally irregular vector for C_{w,f} . There exist D,E\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=\operatorname{\overline{dens}}(E)=1 such that \lim_{n\in D}\|(C_{w,f})^{n}(\varphi)\|=0\ \ \text{ and }\ \ \lim_{n\in E}\|(C_{w,f})^{n}(\varphi)\|=\infty. Consider the relatively compact open sets B_{i}=\{x\in\Omega:|\varphi(x)|>i^{-1}\}\ \ \ \ \ (i\in\mathbb{N}). Since i^{-1}\|w^{(n)}\|_{f^{-n}(B_{i})}\leq\|(\varphi\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B_{i})}\leq\|(C_{w,f})^{n}(\varphi)\|, we have that property (a) holds. On the other hand, for each k\in\mathbb{N} , since \operatorname{\overline{dens}}\{n\in\mathbb{N}:\|(C_{w,f})^{n}(\varphi)\|>k\|\varphi\|\}\geq\operatorname{\overline{dens}}(E)=1, there exists N_{k}\in\mathbb{N} such that \operatorname{card}\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}(\varphi)\|>k\|\varphi\|\}\geq N_{k}\Big(1-\frac{1}{2k}\Big). Moreover, the N_{k} ’s can be chosen so that the sequence (N_{k})_{k\in\mathbb{N}} is increasing. Since \|(C_{w,f})^{n}(\varphi)\|=\sup_{i\in\mathbb{N}}\|(\varphi\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B_{i})}\leq\|\varphi\|\sup_{i\in\mathbb{N}}\|w^{(n)}\|_{f^{-n}(B_{i})}, we see that there exists i_{k}\in\mathbb{N} such that \operatorname{card}\{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(B_{i_{k}})}>k\}\geq N_{k}\Big(1-\frac{1}{2k}\Big). Thus, property (b) holds with \varepsilon=1/2 . ( \Leftarrow ): By property (b), for each k\in\mathbb{N} , there exists i_{k}\in\mathbb{N} such that \operatorname{card}F_{k}\geq\varepsilon N_{k},\ \text{ where }F_{k}=\{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(B_{i_{k}})}>k\}. We shall construct a sequence (A_{k})_{k\in\mathbb{N}} of relatively compact open sets in \Omega such that \overline{A_{k}}\subset B_{i_{1}}\cup\ldots\cup B_{i_{k}} , \overline{A_{k}}\subset A_{k+1} and \{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(A_{k})}>k\}\supset F_{k}\ \text{ for all }k\in\mathbb{N}. For this purpose, for each n\in F_{k} , take x_{k,n}\in f^{-n}(B_{i_{k}}) such that |w^{(n)}(x_{k,n})|>k . We begin by choosing an open set A_{1} in \Omega satisfying \{f^{n}(x_{1,n}):n\in F_{1}\}\subset A_{1}\subset\overline{A_{1}}\subset B_{i_{1}}. If k\geq 2 and A_{1},\ldots,A_{k-1} have already been chosen, then take an open set A_{k} in \Omega such that \overline{A_{k-1}}\cup\{f^{n}(x_{k,n}):n\in F_{k}\}\subset A_{k}\subset\overline{A_{k}}\subset B_{i_{1}}\cup\ldots\cup B_{i_{k}}. It is clear that the sequence (A_{k})_{k\in\mathbb{N}} constructed in this way has the desired properties. Now, for each k\in\mathbb{N} , take a continuous map \phi_{k}:\Omega\to[0,1] such that \phi_{k}=1\text{ on }\overline{A_{k}}\ \ \text{ and }\ \ \operatorname{supp}\phi_{k}\subset A_{k+1}. It follows from property (a) that \lim_{n\in D}\|(C_{w,f})^{n}(\phi_{k})\|\leq\lim_{n\in D}\|w^{(n)}\|_{f^{-n}(A_{k+1})}=0\ \text{ for all }k\in\mathbb{N}. Let \varphi_{k}=\frac{1}{k}\phi_{k} for each k\in\mathbb{N} . Obviously, \varphi_{k}\to 0 as k\to\infty . If n\in\{1,\ldots,N_{k}\} satisfies \|w^{(n)}\|_{f^{-n}(A_{k})}>k , then \|(C_{w,f})^{n}(\varphi_{k})\|\geq\|(\varphi_{k}\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(A_{k})}=\frac{1}{k}\,\|w^{(n)}\|_{f^{-n}(A_{k})}>1. Thus, \operatorname{card}\{1\leq n\leq N_{k}:\|(C_{w,f})^{n}(\varphi_{k})\|>1\}\geq\operatorname{card}F_{k}\geq\varepsilon N_{k}\ \text{ for all }k\in\mathbb{N}. Hence, by the Distributional Chaos Criterion, C_{w,f} is distributionally chaotic. ∎ Remark 43 . Note that the sequence (B_{i})_{i\in\mathbb{N}} constructed in the proof of Theorem 42 has the following additional property: \overline{B_{i}}\subset B_{i+1} for all i\in\mathbb{N} . Corollary 44 . A weighted shift B_{w} on c_{0}(\mathbb{N}) is distributionally chaotic if and only if \inf_{k\in\mathbb{N}}\Big(\sup_{N\in\mathbb{N}}\frac{\operatorname{card}\{1\leq n\leq N:|w_{i}\cdots w_{i+n-1}|>k\text{ for some }i\in\mathbb{N}\}}{N}\Big)>0. (21) Proof. ( \Leftarrow ): Since condition (a) in Theorem 42 is superfluous in the present case, it is enough to establish condition (b). The hypothesis implies the existence of an \varepsilon>0 and a sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that \operatorname{card}I_{k}>\varepsilon N_{k},\ \text{ where }I_{k}=\{1\leq n\leq N_{k}:|w_{i}\cdots w_{i+n-1}|>k\text{ for some }i\in\mathbb{N}\}. By definition, for each n\in I_{k} , there exists i_{k,n}\in\mathbb{N} such that |w_{i_{k,n}}\cdots w_{i_{k,n}+n-1}|>k . Let B_{k}=\{i_{k,n}+n:n\in I_{k}\}. Since \|w^{(n)}\|_{f^{-n}(B_{k})}\geq|w^{(n)}(i_{k,n})|=|w_{i_{k,n}}\cdots w_{i_{k,n}+n-1}|>k for every n\in I_{k} , we obtain \operatorname{card}\{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(B_{k})}>k\}\geq\operatorname{card}I_{k}>\varepsilon N_{k}. Now, by passing to a subsequence, if necessary, we may assume that the sequence (N_{k})_{k\in\mathbb{N}} is increasing. Thus, condition (b) in Theorem 42 holds and B_{w} is distributionally chaotic. ( \Rightarrow ): By Theorem 42 , for each k\in\mathbb{N} , there exists i_{k}\in\mathbb{N} such that \operatorname{card}\{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(B_{i_{k}})}>k\}\geq\varepsilon N_{k}. If n\in\{1,\ldots,N_{k}\} satisfies \|w^{(n)}\|_{f^{-n}(B_{i_{k}})}>k , then |w_{i}\cdots w_{i+n-1}|=|w^{(n)}(i)|>k for some i\in f^{-n}(B_{i_{k}}) . Therefore, \frac{\operatorname{card}\{1\leq n\leq N_{k}:|w_{i}\cdots w_{i+n-1}|>k\text{ for some }i\in\mathbb{N}\}}{N_{k}}\geq\varepsilon. This clearly implies ( 21 ). ∎ Corollary 45 . A weighted shift B_{w} on c_{0}(\mathbb{Z}) is distributionally chaotic if and only if there exist a set S\subset\mathbb{Z} and a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that \lim_{n\in D}|w_{i-n}\cdots w_{i-1}|=0\ \text{ for all }i\in S (22) and \inf_{k\in\mathbb{N}}\Big(\sup_{N\in\mathbb{N}}\frac{\operatorname{card}\{1\leq n\leq N:|w_{i-n}\cdots w_{i-1}|>k\text{ for some }i\in S\}}{N}\Big)>0. (23) Proof. ( \Leftarrow ): Let (B_{i})_{i\in\mathbb{N}} be an enumeration of the collection of all nonempty finite subsets of S . For every i\in\mathbb{N} , \lim_{n\in D}\|w^{(n)}\|_{f^{-n}(B_{i})}=\lim_{n\in D}\Big(\max_{j\in B_{i}}|w_{j-n}\cdots w_{j-1}|\Big)=0, because of ( 22 ). Hence, condition (a) in Theorem 42 holds. By ( 23 ), there exist \varepsilon>0 and a sequence (N_{k})_{k\in\mathbb{N}} of positive integers such that \operatorname{card}I_{k}>\varepsilon N_{k},\ \text{ where }I_{k}=\{1\leq n\leq N_{k}:|w_{i-n}\cdots w_{i-1}|>k\text{ for some }i\in S\}. For each n\in I_{k} , take i_{k,n}\in S such that |w_{i_{k,n}-n}\cdots w_{i_{k,n}-1}|>k . Then, \{i_{k,n}:n\in I_{k}\}=B_{i_{k}} for some i_{k}\in\mathbb{N} . Since \operatorname{card}\{1\leq n\leq N_{k}:\|w^{(n)}\|_{f^{-n}(B_{i_{k}})}>k\}\geq\operatorname{card}I_{k}>\varepsilon N_{k}, we see that (by passing to a subsequence, if necessary) condition (b) in Theorem 42 also holds. Thus, B_{w} is distributionally chaotic. ( \Rightarrow ): Let (B_{i})_{i\in\mathbb{N}} be the sequence given by Theorem 42 . Define S=\bigcup_{i\in\mathbb{N}}B_{i}\subset\mathbb{Z} . It is straightforward to check that properties (a) and (b) in Theorem 42 imply ( 22 ) and ( 23 ), respectively. ∎ Remark 46 . If the weights w_{n} are nonzero and \lim_{n\to\infty}\prod_{j=-n+1}^{0}w_{j}=0, then ( 22 ) holds for S=\mathbb{Z} and D=\mathbb{N} . Hence, in this case, the weighted shift B_{w} on c_{0}(\mathbb{Z}) is distributionally chaotic if and only if \inf_{k\in\mathbb{N}}\Big(\sup_{N\in\mathbb{N}}\frac{\operatorname{card}\{1\leq n\leq N:|w_{i}\cdots w_{i+n-1}|>k\text{ for some }i\in\mathbb{Z}\}}{N}\Big)>0. The corresponding results for weighted translation operators are stated below, but their proofs are left to the reader. Corollary 47 . A weighted translation operator T_{w} on C_{0}[1,\infty) is distributionally chaotic if and only if \inf_{k\in\mathbb{N}}\Big(\sup_{N\in\mathbb{N}}\frac{\operatorname{card}\{1\leq n\leq N:|w(x)\cdots w(x+n-1)|>k\text{ for some }x\in[1,\infty)\}}{N}\Big)>0. Corollary 48 . Consider a weighted translation operator T_{w} on C_{0}(\mathbb{R}) with a zero-free weight function satisfying \lim_{n\to\infty}\sup_{x\in(0,1)}|w(x-n)\cdots w(x-1)|=0. Then, T_{w} is distributionally chaotic if and only if \inf_{k\in\mathbb{N}}\Big(\sup_{N\in\mathbb{N}}\frac{\operatorname{card}\{1\leq n\leq N:|w(x)\cdots w(x+n-1)|>k\text{ for some }x\in\mathbb{R}\}}{N}\Big)>0. Theorem 49 . Consider a weighted composition operator C_{w,f} on C_{0}(\Omega) with positive weight function w:\Omega\to(0,\infty) . Suppose that there is a sequence (A_{k})_{k\in\mathbb{N}} of relatively compact open sets in \Omega such that the following properties hold: (a) There exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 such that \lim_{n\in D}\|w^{(n)}\|_{f^{-n}(A_{k})}=0\ \text{ for all }k\in\mathbb{N}. (b) There exists a set E\subset\mathbb{N} with \operatorname{\overline{dens}}(E)>0 such that \sum_{n\in E}\frac{1}{\|w^{(n)}\|_{f^{-n}(A_{n})}}<\infty. Then, C_{w,f} is distributionally chaotic. Proof. For each k\in E , choose a point x_{k}\in f^{-k}(A_{k}) with w^{(k)}(x_{k})>\frac{1}{2}\|w^{(k)}\|_{f^{-k}(A_{k})} , an open set B_{k} in \Omega with f^{k}(x_{k})\in B_{k}\subset\overline{B_{k}}\subset A_{k} , and a continuous map \phi_{k}:\Omega\to[0,1] with \phi_{k}=1\text{ on }\overline{B_{k}}\ \ \text{ and }\ \ \operatorname{supp}\phi_{k}\subset A_{k}. By property (a), \lim_{n\in D}\|(C_{w,f})^{n}(\phi_{k})\|\leq\lim_{n\in D}\|w^{(n)}\|_{f^{-n}(A_{k})}=0\ \text{ for all }k\in E. By property (b), \sum_{n\in E}\frac{1}{\|w^{(n)}\|_{f^{-n}(B_{n})}}<2\sum_{n\in E}\frac{1}{\|w^{(n)}\|_{f^{-n}(A_{n})}}<\infty. Let us now prove that there exists \varphi\in\overline{\operatorname{span}\{\phi_{k}:k\in E\}} such that \lim_{n\in E}\|(C_{w,f})^{n}(\varphi)\|=\infty. In view of [ 7 , Proposition 8] , this will allow us to apply the Distributional Chaos Criterion and conclude that C_{w,f} is distributionally chaotic. Let \displaystyle a_{n} \displaystyle=\left\{\begin{array}[]{cl}\frac{1}{\|w^{(n)}\|_{f^{-n}(B_{n})}}&\text{ if }n\in E\\ 0&\text{ if }n\notin E,\end{array}\right. \displaystyle r_{n} \displaystyle=\displaystyle\sum_{i\geq n}a_{i}\,, \displaystyle c_{n} \displaystyle=\left\{\begin{array}[]{cl}\frac{1}{\sqrt{r_{n}}\,\|w^{(n)}\|_{f^{-n}(B_{n})}}&\text{ if }n\in E\\ 0&\text{ if }n\notin E.\end{array}\right. Calculations similar to those made in the proof of Theorem 33 show that \sum_{n\in E}c_{n}<\infty\ \ \text{ and }\ \ \lim_{n\in E}c_{n}\,\|w^{(n)}\|_{f^{-n}(B_{n})}=\infty. Hence, we can define \varphi=\sum_{n\in E}c_{n}\phi_{n}\in C_{0}(\Omega), since this series is absolutely convergent. Moreover, \varphi\in\overline{\operatorname{span}\{\phi_{k}:k\in E\}} . Finally, \displaystyle\lim_{n\in E}\|(C_{w,f})^{n}(\varphi)\| \displaystyle=\lim_{n\in E}\Big\|\sum_{k\in E}c_{k}\,(C_{w,f})^{n}(\phi_{k})\Big\| \displaystyle\geq\lim_{n\in E}\|c_{n}\,(C_{w,f})^{n}(\phi_{n})\| \displaystyle\geq\lim_{n\in E}c_{n}\,\|w^{(n)}\|_{f^{-n}(B_{n})}=\infty, which completes the proof. ∎ Corollary 50 . Consider a weighted composition operator C_{w,f} on C_{0}(\Omega) with positive weight function w:\Omega\to(0,\infty) . If there is a relatively compact open set A in \Omega such that • there exists D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)=1 and \displaystyle\lim_{n\in D}\|w^{(n)}\|_{f^{-n}(A)}=0 , • there exists E\subset\mathbb{N} with \operatorname{\overline{dens}}(E)>0 and \displaystyle\sum_{n\in E}\frac{1}{\|w^{(n)}\|_{f^{-n}(A)}}<\infty , then C_{w,f} is distributionally chaotic. Theorem 51 . Consider a weighted composition operator C_{w,f} on C_{0}(\Omega) . If the space C_{0}(\Omega) is separable and \lim_{n\to\infty}\|w^{(n)}\|_{f^{-n}(B)}=0 for every relatively compact open set B in \Omega , then the following assertions are equivalent: (i) C_{w,f} is distributionally chaotic; (ii) C_{w,f} is densely distributionally chaotic; (iii) C_{w,f} admits a dense distributionally irregular manifold; (iv) There exist \phi\in C_{0}(\Omega) and \delta>0 such that \operatorname{\overline{dens}}\{n\in\mathbb{N}:\|(C_{w,f})^{n}(\phi)\|\geq\delta\}>0. Proof. (i) \Rightarrow (iv): Take a distributionally irregular vector \phi\in C_{0}(\Omega) for C_{w,f} . (iv) \Rightarrow (iii): The assumption implies that \lim_{n\to\infty}(C_{w,f})^{n}(\varphi)=0\ \text{ for all }\varphi\in C_{c}(\Omega). Since C_{c}(\Omega) is dense in C_{0}(\Omega) , (iii) follows from (iv) and [ 9 , Theorem 33] . (iii) \Rightarrow (ii) \Rightarrow (i): Obvious. ∎ Corollary 52 . Assume the hypotheses of the previous theorem. If there exist a constant C>0 and a relatively compact open set A in \Omega such that \operatorname{\overline{dens}}\{n\in\mathbb{N}:\|w^{(n)}\|_{f^{-n}(A)}\geq C\}>0, then C_{w,f} is densely distributionally chaotic. Proof. Take any \phi\in C_{0}(\Omega) with \phi=1 on A , and put \delta=C . Then, property (iv) in the previous theorem holds. ∎ Theorem 53 . Consider a weighted composition operator C_{w,f} on C_{0}(\Omega) with positive weight function w:\Omega\to(0,\infty) . Suppose that the space C_{0}(\Omega) is separable and the following properties hold: (a) For every relatively compact open set B in \Omega , \lim_{n\to\infty}\|w^{(n)}\|_{f^{-n}(B)}=0. (b) There exist a relatively compact open set A in \Omega and a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)>0 such that \sum_{n\in D}\frac{1}{\|w^{(n)}\|_{A}}<\infty. Then, C_{w,f} is densely distributionally chaotic. Proof. For each n\in\mathbb{N} , let A_{n} be a relatively compact open set in \Omega such that f^{n}(A)\subset A_{n} . Since \sum_{n\in D}\frac{1}{\|w^{(n)}\|_{f^{-n}(A_{n})}}\leq\sum_{n\in D}\frac{1}{\|w^{(n)}\|_{A}}<\infty, it follows from Theorem 49 that C_{w,f} is distributionally chaotic. Hence, by Theorem 51 , C_{w,f} is densely distributionally chaotic. ∎ The following results on weighted shifts follow easily from the previous theorem (we omit the details), but they can also be derived from Corollaries 44 and 45 . Corollary 54 . Consider a weighted shift B_{w} on c_{0}(\mathbb{N}) with positive weights. If there exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)>0 such that \sum_{n\in D}\left(\prod_{j=1}^{n}w_{j}\right)^{-1}<\infty, then B_{w} is densely distributionally chaotic. Corollary 55 . Consider a weighted shift B_{w} on c_{0}(\mathbb{Z}) with positive weights. If \lim_{n\to\infty}\prod_{j=-n+1}^{0}w_{j}=0 and there exists a set D\subset\mathbb{N} with \operatorname{\overline{dens}}(D)>0 such that \sum_{n\in D}\left(\prod_{j=1}^{n}w_{j}\right)^{-1}<\infty, then B_{w} is densely distributionally chaotic. 5 Absolutely Cesàro bounded weighted composition operators Recall that an operator T on a Banach space Y is said to be p -absolutely Cesàro bounded if there exists C\in(0,\infty) such that \sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\|T^{n}y\|^{p}\leq C\|y\|^{p}\ \text{ for all }y\in Y. More generally, given a subspace Z of Y , we say that T is p -absolutely Cesàro bounded in Z if the above inequality holds for every y\in Z . By defining the extended real number N_{p}(T)=\sup_{\|y\|=1}\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\|T^{n}y\|^{p}, we have that T\text{ is $p$-absolutely Ces\`{a}ro bounded }\ \Longleftrightarrow\ N_{p}(T)<\infty. It is usual to say absolutely Cesàro bounded instead of 1 -absolutely Cesàro bounded. Recall also that an operator T on a Banach space Y is said to be mean ergodic if the sequence (M_{n}(T))_{n\in\mathbb{N}} of Cesàro means of T , defined by M_{n}(T)y=\frac{1}{n+1}\sum_{k=0}^{n}T^{k}y\ \ \ \ \ (n\in\mathbb{N}), converges in the strong operator topology of the space of all operators on Y . 5.1 The case of the space L^{p}(\mu) Theorem 56 . For any weighted composition operator C_{w,f} on L^{p}(\mu) , N_{p}(C_{w,f})=\sup_{0<\mu(B)<\infty}\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B))}{\mu(B)}\cdot (24) In particular, C_{w,f} is p -absolutely Cesàro bounded if and only if there exists a constant C\in(0,\infty) such that, for each measurable set B of finite positive \mu -measure, \frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))\leq C\,\mu(B)\ \text{ for all }N\in\mathbb{N}. (25) Proof. Denote the right-hand side of ( 24 ) by r . Given a measurable set B of finite positive \mu -measure, define \phi=\frac{1}{\mu(B)^{1/p}}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B} . Since \|\phi\|_{p}=1 , we obtain \sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B))}{\mu(B)}=\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\phi)\|_{p}^{p}\leq N_{p}(C_{w,f}). This shows that r\leq N_{p}(C_{w,f}) . Conversely, fix t>1 . Given \varphi\in L^{p}(\mu) with \|\varphi\|_{p}=1 , consider the measurable sets B_{i}=\{x\in X:t^{i-1}\leq|\varphi(x)|1 is arbitrary, we obtain N_{p}(C_{w,f})\leq r . The last assertion follows from ( 24 ). ∎ Corollary 57 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . If p>1 and there exists C\in(0,\infty) such that, for each measurable set B of finite positive \mu -measure, \frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))\leq C\,\mu(B)\ \text{for all }N\in\mathbb{N}, then C_{w,f} is mean ergodic. Proof. Every p -absolutely Cesàro bounded operator in a reflexive Banach space is mean ergodic [ 6 , Corollary 2.7] . ∎ Corollary 58 . For any weighted shift B_{w} on \ell^{p}(\mathbb{N}) , N_{p}(B_{w})=\sup_{i\in\mathbb{N},N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{\min\{N,\;i-1\}}|w_{i-n}\cdots w_{i-1}|^{p}. (26) In particular, B_{w} is p -absolutely Cesàro bounded if and only if the above supremum is finite. Proof. Formula ( 26 ) is a special case of formula ( 24 ). Indeed, regard B_{w} as C_{w,f} by considering X=\mathbb{N} , \mathfrak{M}=P(\mathbb{N}) , \mu the counting measure on \mathfrak{M} and f:n\in\mathbb{N}\mapsto n+1\in\mathbb{N} . Denote the right-hand side of ( 26 ) by r . By ( 24 ), N_{p}(B_{w})\geq\sup_{i\in\mathbb{N}}\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(\{i\}))}{\mu(\{i\})}=r. On the other hand, for any nonempty finite set B=\{i_{1},\ldots,i_{k}\}\subset\mathbb{N} , \displaystyle\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B))}{\mu(B)} \displaystyle=\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\sum_{j=1}^{k}\frac{\mu_{n}(f^{-n}(\{i_{j}\}))}{k} \displaystyle\leq\sum_{j=1}^{k}\frac{1}{k}\Big(\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{\min\{N,\,i_{j}-1\}}|w_{i_{j}-n}\cdots w_{i_{j}-1}|^{p}\Big)\leq r. Thus, by ( 24 ), N_{p}(B_{w})\leq r . ∎ Corollary 59 . For any weighted shift B_{w} on \ell^{p}(\mathbb{Z}) , N_{p}(B_{w})=\sup_{i\in\mathbb{Z},N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}|w_{i-n}\cdots w_{i-1}|^{p}. (27) In particular, B_{w} is p -absolutely Cesàro bounded if and only if the above supremum is finite. Proof. Formula ( 27 ) is also a special case of formula ( 24 ). The proof is similar to that of Corollary 58 and is left to the reader. ∎ A special case of Jensen’s inequality (see [ 30 , Theorem 3.3] ) asserts that \varphi\Big(\frac{a_{1}+\cdots+a_{N}}{N}\Big)\leq\frac{\varphi(a_{1})+\cdots+\varphi(a_{N})}{N}\,, whenever \varphi:[0,\infty)\to\mathbb{R} is a convex function, N\in\mathbb{N} and a_{1},\ldots,a_{N}\in[0,\infty) . As a consequence, for any operator T on a Banach space Y , T\text{ $p$-absolutely Ces\`{a}ro bounded }\ \Longrightarrow\ T\text{ $q$-absolutely Ces\`{a}ro bounded }\forall q\in[1,p] (for a generalization of this fact see [ 1 , Theorem 7.12] ). Example 60 . Given any real number q>p , consider the weighted shift B_{w} on \ell^{p}(\mathbb{N}) whose weight sequence w=(w_{n})_{n\in\mathbb{N}} is given by w_{n}=\left(\frac{n+1}{n}\right)^{\frac{1}{q}}\ \text{ for all }n\in\mathbb{N}. Then, B_{w} is p -absolutely Cesàro bounded, but it is not q -absolutely Cesàro bounded. In particular, B_{w} is not power-bounded. Proof. Indeed, let \alpha=\frac{p}{q}\in(0,1) . For any i\geq 2 and N\in\mathbb{N} , \displaystyle\sum_{n=1}^{\min\{N,\;i-1\}}|w_{i-n}\cdots w_{i-1}|^{p} \displaystyle= \displaystyle\sum_{n=1}^{\min\{N,\;i-1\}}\Big(\frac{i}{i-n}\Big)^{\alpha} \displaystyle= \displaystyle\left\{\begin{array}[]{cl}i^{\alpha}\sum_{n=1}^{\min\{N,\;i-1\}}(i-n)^{-\alpha}&\text{if }i\leq 2N\\ \sum_{n=1}^{N}\Big(\frac{i}{i-n}\Big)^{\alpha}&\text{if }i>2N\end{array}\right. \displaystyle\leq \displaystyle\left\{\begin{array}[]{cl}i^{\alpha}\sum_{n=1}^{i-1}n^{-\alpha}&\text{if }i\leq 2N\\ \sum_{n=1}^{N}\Big(\frac{i}{i-N}\Big)^{\alpha}&\text{if }i>2N\end{array}\right. \displaystyle\leq \displaystyle\left\{\begin{array}[]{cl}i^{\alpha}\big(1+\int_{1}^{i-1}t^{-\alpha}dt\big)&\text{if }i\leq 2N\\ \sum_{n=1}^{N}2^{\alpha}&\text{if }i>2N\end{array}\right. \displaystyle\leq \displaystyle\left\{\begin{array}[]{ll}\frac{i}{1-\alpha}&\text{if }i\leq 2N\\ 2N&\text{if }i>2N\end{array}\right. \displaystyle\leq \displaystyle\frac{2}{1-\alpha}\,N. Thus, by Corollary 58 , B_{w} is p -absolutely Cesàro bounded. On the other hand, suppose that B_{w} is q -absolutely Cesàro bounded. Then, there exists a constant C\in(0,\infty) such that \sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\|(B_{w})^{n}x\|_{p}^{q}\leq C\|x\|_{p}^{q}\ \text{ for all }x\in\ell_{p}(\mathbb{N}). In particular, for each i\geq 2 , C\geq\frac{1}{i-1}\sum_{n=1}^{i-1}\|(B_{w})^{n}e_{i}\|_{p}^{q}=\frac{1}{i-1}\Big(\frac{i}{i-1}+\frac{i}{i-2}+\cdots+\frac{i}{1}\Big)=\frac{i}{i-1}\sum_{n=1}^{i-1}\frac{1}{n}\cdot Since the harmonic series diverges, we have a contradiction. ∎ The above example comes from [ 6 , Theorem 2.1] , where it was proved that B_{w} is p -absolutely Cesàro bounded but not power-bounded. Here we obtain the stronger conclusion that B_{w} is not q -absolutely Cesàro bounded. Moreover, the proof that B_{w} is p -absolutely Cesàro bounded, as an application of Corollary 58 , is slightly shorter (but it uses similar estimates). Example 61 . The weighted translation operator T_{w} on L^{p}[1,\infty) with weight function w(x)=\Big(\frac{x+1}{x}\Big)^{\frac{1-\varepsilon}{p}}, where \varepsilon>0 is fixed, is p -absolutely Cesàro bounded. Proof. In view of Theorem 56 , it is enough to show that there exists a constant C\in(0,\infty) such that, for each Lebesgue measurable set B\subset[1,\infty) with finite positive Lebesgue measure, \frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))\leq C\,\mu(B)\ \text{ for all }N\in\mathbb{N}. But, in fact, \displaystyle\sum_{n=1}^{N}\mu_{n}(f^{-n}(B)) \displaystyle=\sum_{n=1}^{N}\int_{(B-n)\cap[1,\infty)}\Big(\frac{x+n}{x}\Big)^{1-\varepsilon}dx \displaystyle=\sum_{n=1}^{N}\int_{B\cap[n+1,\infty)}\Big(\frac{x}{x-n}\Big)^{1-\varepsilon}dx=\int_{B}\sum_{n=1}^{\min\{N,\;[x]-1\}}\Big(\frac{x}{x-n}\Big)^{1-\varepsilon}dx \displaystyle\leq\int_{B\cap[1,1+2N]}x^{1-\varepsilon}\sum_{n=1}^{[x]-1}\Big(\frac{1}{x-n}\Big)^{1-\varepsilon}dx+\int_{B\cap[1+2N,\infty)}\sum_{n=1}^{N}\Big(\frac{x}{x-n}\Big)^{1-\varepsilon}dx \displaystyle<\frac{1+2N}{\varepsilon}\mu(B)+2N\mu(B)\leq(2+\frac{3}{\varepsilon})N\mu(B), as it was to be shown. ∎ 5.2 The case of the space C_{0}(\Omega) Theorem 62 . For any weighted composition operator C_{w,f} on C_{0}(\Omega) , N_{p}(C_{w,f})=\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|^{p}. (32) In particular, C_{w,f} is p -absolutely Cesàro bounded if and only if there exists a constant C\in(0,\infty) such that \frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|^{p}\leq C\ \text{ for all }N\in\mathbb{N}. (33) Proof. Denote the right-hand side of ( 32 ) by r . For every \varphi\in C_{0}(\Omega) with \|\varphi\|=1 , \frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|^{p}=\frac{1}{N}\sum_{n=1}^{N}\|(\varphi\circ f^{n})\cdot w^{(n)}\|^{p}\leq\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|^{p}. This shows that N_{p}(C_{w,f})\leq r . Conversely, fix N\in\mathbb{N} and \varepsilon>0 . For each n\in\{1,\ldots,N\} , there is x_{n}\in\Omega with |w^{(n)}(x_{n})|\geq\|w^{(n)}\|-\varepsilon . Let V be a relatively compact open set in \Omega such that \{f^{n}(x_{1}),\ldots,f^{n}(x_{N})\}\subset V . By Urysohn’s lemma, there exists a continuous map \phi:\Omega\to[0,1] such that \operatorname{supp}\phi\subset V and \phi(f^{n}(x_{n}))=1 for all n\in\{1,\ldots,N\} . Hence, N_{p}(C_{w,f})\geq\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\phi)\|^{p}\geq\frac{1}{N}\sum_{n=1}^{N}|\phi(f^{n}(x_{n}))w^{(n)}(x_{n})|^{p}\geq\frac{1}{N}\sum_{n=1}^{N}(\|w^{(n)}\|-\varepsilon)^{p}. Since N\in\mathbb{N} and \varepsilon>0 are arbitrary, we obtain N_{p}(C_{w,f})\geq r . The last assertion follows from ( 32 ). ∎ The results below follow immediately from the previous theorem. Corollary 63 . Let \Omega=\mathbb{N} or \mathbb{Z} . For any weighted shift B_{w} on c_{0}(\Omega) , N_{p}(B_{w})=\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\Big(\sup_{i\in\Omega}|w_{i}\cdots w_{i+n-1}|\Big)^{p}. (34) In particular, B_{w} is p -absolutely Cesàro bounded if and only if the above supremum is finite. Corollary 64 . Let \Omega=[1,\infty) or \mathbb{R} . For any weighted translation operator T_{w} on C_{0}(\Omega) , N_{p}(T_{w})=\sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\Big(\sup_{x\in\Omega}|w(x)\cdots w(x+n-1)|\Big)^{p}. (35) In particular, T_{w} is p -absolutely Cesàro bounded if and only if the above supremum is finite. Remark 65 . Formula ( 26 ) and formula ( 34 ) for \Omega=\mathbb{N} do not give the same value in general. For instance, consider the weight sequence w=(w_{n})_{n\in\mathbb{N}} obtained by concatenating the blocks z^{(n)}=\big(\underbrace{\sqrt[n]{e},\ldots,\sqrt[n]{e}}_{n\text{ times}},\frac{1}{e}\big)\ \ \text{ for }n\geq 2. Since \sup_{i\in\mathbb{N}}|w_{i}\cdots w_{i+n-1}|=\left\{\begin{array}[]{cl}\sqrt{e}&\text{if }n=1\\ e&\text{if }n\geq 2,\end{array}\right. we have that \sup_{N\in\mathbb{N}}\frac{1}{N}\sum_{n=1}^{N}\Big(\sup_{i\in\mathbb{N}}|w_{i}\cdots w_{i+n-1}|\Big)^{p}=\sup_{N\in\mathbb{N}}\frac{(\sqrt{e})^{p}+(N-1)e^{p}}{N}=e^{p}. On the other hand, the largest possible value V_{N} for \frac{1}{N}\sum_{n=1}^{\min\{N,\;i-1\}}|w_{i-n}\cdots w_{i-1}|^{p} is given by V_{N}=\left\{\begin{array}[]{cl}(\sqrt{e})^{p}&\text{if }N=1\\ \displaystyle\frac{(\sqrt[N]{e}\,)^{p}+(\sqrt[N]{e}\,)^{2p}+\cdots+(\sqrt[N]{e}\,)^{Np}}{N}&\text{if }N\geq 2.\end{array}\right. In particular, V_{N}0. If the set S can be chosen to be dense in M , then f is densely mean Li-Yorke chaotic . An extensive study of the concept of mean Li-Yorke chaos in the setting of linear dynamics was developed in [ 10 ] . In particular, the following useful characterizations were obtained: For any operator T on any Banach space Y , the following assertions are equivalent: (i) T is mean Li-Yorke chaotic; (ii) T admits an absolutely mean semi-irregular vector , that is, a vector y\in Y such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|T^{n}y\|=0\ \ \text{ and }\ \ \limsup_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|T^{n}y\|>0. (iii) T admits an absolutely mean irregular vector , that is, a vector y\in Y such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|T^{n}y\|=0\ \ \text{ and }\ \ \limsup_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|T^{n}y\|=\infty. Let us also recall that by an absolutely mean irregular manifold for T we mean a vector subspace of Y consisting, except for the zero vector, of absolutely mean irregular vectors for T . 6.1 The case of the space L^{p}(\mu) Theorem 66 (Necessary Condition) . If a weighted composition operator C_{w,f} on L^{p}(\mu) is mean Li-Yorke chaotic, then there exists a nonempty countable family (B_{i})_{i\in I} of measurables sets of finite positive \mu -measure such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B_{i}))^{\frac{1}{p}}=0\ \ \text{ for all }i\in I (36) and \sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}:i\in I,N\in\mathbb{N}\Big\}=\infty. (37) Proof. Let \varphi\in L^{p}(\mu) be an absolutely mean irregular vector for C_{w,f} . Consider the measurable sets B_{i}=\{x\in X:2^{i-1}\leq|\varphi(x)|<2^{i}\}\ \ \ (i\in\mathbb{Z}) and let I be the nonempty subset of \mathbb{Z} given by I=\{i\in\mathbb{Z}:\mu(B_{i})>0\} . We have that 0<\mu(B_{i})<\infty for all i\in I , because 2^{(i-1)p}\mu(B_{i})\leq\int_{X}|\varphi|^{p}d\mu<\infty\ \ \ (i\in\mathbb{Z}). Since \displaystyle 2^{i-1}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B_{i}))^{\frac{1}{p}} \displaystyle\leq\frac{1}{N}\sum_{n=1}^{N}\Big(\int_{f^{-n}(B_{i})}|\varphi\circ f^{n}|^{p}|w^{(n)}|^{p}d\mu\Big)^{\frac{1}{p}} \displaystyle\leq\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|_{p} and \varphi is an absolutely mean irregular vector for C_{w,f} , it follows that ( 36 ) holds. On the other hand, if ( 37 ) fails, that is, C=\sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}:i\in I,N\in\mathbb{N}\Big\}<\infty, then \displaystyle\Big(\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|_{p}\Big)^{p} \displaystyle\leq\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|_{p}^{p} \displaystyle=\frac{1}{N}\sum_{n=1}^{N}\sum_{i\in\mathbb{Z}}\int_{f^{-n}(B_{i})}|\varphi\circ f^{n}|^{p}|w^{(n)}|^{p}d\mu \displaystyle\leq\frac{1}{N}\sum_{n=1}^{N}\sum_{i\in\mathbb{Z}}2^{ip}\mu_{n}(f^{-n}(B_{i})) \displaystyle=\sum_{i\in\mathbb{Z}}2^{ip}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B_{i})) \displaystyle\leq\sum_{i\in\mathbb{Z}}2^{ip}C\mu(B_{i})=2^{p}C\sum_{i\in\mathbb{Z}}2^{(i-1)p}\mu(B_{i})\leq 2^{p}C\|\varphi\|_{p}^{p}, contradicting the fact that \varphi is an absolutely mean irregular vector for C_{w,f} . ∎ Remark 67 . Note that the countable family (B_{i})_{i\in I} constructed in the proof of Theorem 66 has the following additional properties: the sets B_{i} are pairwise disjoint and there exists an increasing sequence (N_{j})_{j\in\mathbb{N}} of positive integers such that \lim_{j\to\infty}\frac{1}{N_{j}}\sum_{n=1}^{N_{j}}\mu_{n}(f^{-n}(B_{i}))^{\frac{1}{p}}=0\ \ \text{ for all }i\in I. Theorem 68 (Sufficient Condition) . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . If there exists a nonempty countable family (B_{i})_{i\in I} of measurables sets of finite positive \mu -measure such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B_{i}))^{\frac{1}{p}}=0\ \ \text{ for all }i\in I (38) and \sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B_{i}))^{\frac{1}{p}}}{\mu(B_{i})}:i\in I,N\in\mathbb{N}\Big\}=\infty, (39) then C_{w,f} is mean Li-Yorke chaotic Proof. Let \mathfrak{X} be the closed subspace of L^{p}(\mu) generated by the characteristic functions {\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B} such that B is a measurable set of finite \mu -measure satisfying \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))^{\frac{1}{p}}=0. (40) For each i\in I , let \phi_{i}=\frac{1}{\mu(B_{i})}\,{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{i}} . By ( 38 ), \phi_{i}\in\mathfrak{X} for all i\in I . Since \|\phi_{i}\|_{p}=1 and \frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\phi_{i})\|_{p}=\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B_{i}))^{\frac{1}{p}}}{\mu(B_{i})}, it follows from ( 39 ) that C_{w,f} is not absolutely Cesàro bounded in \mathfrak{X} . Suppose, by contradiction, that C_{w,f} is not mean Li-Yorke chaotic. Then, C_{w,f} does not admit an absolutely mean semi-irregular vector, and so ( 40 ) is equivalent to \lim_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))^{\frac{1}{p}}=0. This implies that the set \mathcal{R}_{1} of all \varphi\in\mathfrak{X} such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|_{p}=0 is residual in \mathfrak{X} . Since C_{w,f} is not absolutely Cesàro bounded in \mathfrak{X} , [ 10 , Theorem 4] implies that the set \mathcal{R}_{2} of all \varphi\in\mathfrak{X} such that \limsup_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|_{p}=\infty is also residual in \mathfrak{X} . Since each \varphi\in\mathcal{R}_{1}\cap\mathcal{R}_{2} is an absolutely mean irregular vector for C_{w,f} , we conclude that C_{w,f} is mean Li-Yorke chaotic, a contradiction. ∎ In the case p=1 , conditions ( 37 ) and ( 39 ) coincide, and so Theorems 66 and 68 gives us a characterization of the mean Li-Yorke chaotic weighted composition operators on L^{1}(\mu) . More precisely, the following result holds. Corollary 69 . Assume p=1 . Given a weighted composition operator C_{w,f} on L^{1}(\mu) , let \mathfrak{X} be the closed subspace of L^{1}(\mu) generated by the characteristic functions {\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B} such that B is a measurable set of finite \mu -measure satisfying \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))=0. Then, the following assertions are equivalent: (i) C_{w,f} is mean Li-Yorke chaotic; (ii) There exists a nonempty countable family (B_{i})_{i\in I} of measurables sets of finite positive \mu -measure such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B_{i}))=0\ \ \text{ for all }i\in I and \sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\frac{\mu_{n}(f^{-n}(B_{i}))}{\mu(B_{i})}:i\in I,N\in\mathbb{N}\Big\}=\infty. (iii) C_{w,f} is not absolutely Cesàro bounded in \mathfrak{X} . Theorem 70 . Consider a weighted composition operator C_{w,f} on L^{p}(\mu) . If for every measurable set A of finite \mu -measure and for every \varepsilon>0 , there is a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \ \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))^{\frac{1}{p}}=0, then the following assertions are equivalent: (i) C_{w,f} is mean Li-Yorke chaotic; (ii) C_{w,f} has a residual set of absolutely mean irregular vectors; (iii) C_{w,f} is not absolutely Cesàro bounded. Proof. Let \mathfrak{X}_{0} be the set of all simple functions of the form \sum_{k=1}^{m}b_{k}{\mathchoice{\raisebox{0.0pt}{$\displaystyle\chi$}}{\raisebox{0.0pt}{$\textstyle\chi$}}{\raisebox{0.0pt}{$\scriptstyle\chi$}}{\raisebox{0.0pt}{$\scriptscriptstyle\chi$}}}_{B_{k}} , where b_{1},\ldots,b_{m} are scalars and B_{1},\ldots,B_{m} are measurable sets of finite \mu -measure satisfying \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B_{1}\cup\ldots\cup B_{m}))^{\frac{1}{p}}=0. (41) By the assumption, \mathfrak{X}_{0} is dense in L^{p}(\mu) . Moreover, by ( 41 ), \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|_{p}=0\ \text{ for all }\varphi\in\mathfrak{X}_{0}. Hence, the equivalences between properties (i), (ii) and (iii) follow from [ 10 , Theorem 22] . ∎ Remark 71 . (a) In the case L^{p}(\mu) is separable, [ 10 , Theorem 17] says that (ii) is equivalent to (ii’) C_{w,f} is densely mean Li-Yorke chaotic. (b) If the space L^{p}(\mu) is separable and for every measurable set A of finite \mu -measure and for every \varepsilon>0 , there exists a measurable set B\subset A with \mu(A\backslash B)<\varepsilon\ \ \text{ and }\ \ \ \lim_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))^{\frac{1}{p}}=0, then [ 10 , Theorem 29] implies that (i)–(iii) are also equivalent to (iv) C_{w,f} admits a dense absolutely mean irregular manifold. Let us now present some applications to weighted shifts and weighted translation operators. Corollary 72 . For weighted shifts B_{w} on \ell^{p}(\mathbb{N}) and weighted translation operators T_{w} on L^{p}[1,\infty) , properties (i)–(iv) above are always equivalent to each other. Proof. The conditions in Remark 71 (b) are satisfied. ∎ Corollary 73 . A weighted shift B_{w} on \ell^{p}(\mathbb{Z}) with nonzero weights is mean Li-Yorke chaotic if and only if it is not absolutely Cesàro bounded and \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}|w_{-n}\cdots w_{-1}|=0. (42) Proof. ( \Rightarrow ): Since B_{w} is mean Li-Yorke chaotic, it is not absolutely Cesàro bounded. Let (B_{i})_{i\in I} be the family given by Theorem 66 . Choose i\in I and j\in B_{i} . By ( 36 ), \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}|w_{-n+j}\cdots w_{-1+j}|=\liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(\{j\}))^{\frac{1}{p}}=0, which gives ( 42 ). ( \Leftarrow ): If \limsup_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}|w_{-n}\cdots w_{-1}|>0, then e_{0} is an absolutely mean irregular vector for B_{w} , and so B_{w} is mean Li-Yorke chaotic. If \lim_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}|w_{-n}\cdots w_{-1}|=0, then \lim_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\mu_{n}(f^{-n}(B))^{\frac{1}{p}}=0\ \text{ for all }B\subset\mathbb{Z}\text{ finite}. Since B_{w} is not absolutely Cesàro bounded, we can apply Theorem 70 and conclude that B_{w} is mean Li-Yorke chaotic. ∎ 6.2 The case of the space C_{0}(\Omega) Theorem 74 . Given a weighted composition operator C_{w,f} on C_{0}(\Omega) , let \mathfrak{X} be the closed subspace of C_{0}(\Omega) generated by the functions \varphi\in C_{0}(\Omega) whose support is contained in a relatively compact open set B in \Omega satisfying \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B)}=0. (43) Then, the following assertions are equivalent: (i) C_{w,f} is mean Li-Yorke chaotic; (ii) There exists a sequence (B_{i})_{i\in\mathbb{N}} of relatively compact open sets in \Omega such that \overline{B_{i}}\subset B_{i+1}\ \text{ for all }i\in\mathbb{N}, (44) \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B_{i})}=0\ \text{ for all }i\in\mathbb{N} (45) and \sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B_{i})}:i,N\in\mathbb{N}\Big\}=\infty. (46) (iii) C_{w,f} is not absolutely Cesàro bounded in \mathfrak{X} . Proof. (i) \Rightarrow (ii): Let \varphi\in C_{0}(\Omega) be an absolutely mean irregular vector for C_{w,f} . Then, there is an increasing sequence (N_{j})_{j\in\mathbb{N}} of positive integers such that \lim_{j\to\infty}\frac{1}{N_{j}}\sum_{n=1}^{N_{j}}\|(C_{w,f})^{n}(\varphi)\|=0. For each i\in\mathbb{N} , consider the relatively compact open subset B_{i} of \Omega given by B_{i}=\{x\in\Omega:|\varphi(x)|>i^{-1}\}. Clearly, ( 44 ) holds. Since, for each i\in\mathbb{N} , \frac{1}{N_{j}}\sum_{n=1}^{N_{j}}\|(C_{w,f})^{n}(\varphi)\|\geq\frac{1}{N_{j}}\sum_{n=1}^{N_{j}}\|(\varphi\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B_{i})}\geq\frac{1}{i}\frac{1}{N_{j}}\sum_{n=1}^{N_{j}}\|w^{(n)}\|_{f^{-n}(B_{i})}, we obtain ( 45 ). Now, suppose that ( 46 ) is false, that is, C=\sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B_{i})}:i,N\in\mathbb{N}\Big\}<\infty. For each n\in\mathbb{N} , take x_{n}\in\Omega such that \|(C_{w,f})^{n}(\varphi)\|=|\varphi(f^{n}(x_{n}))w^{(n)}(x_{n})| . Given N\in\mathbb{N} , there exists i_{N}\in\mathbb{N} such that x_{n}\in f^{-n}(B_{i_{N}}) for all n\in\{1,\ldots,N\} . Hence, \frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|=\frac{1}{N}\sum_{n=1}^{N}|\varphi(f^{n}(x_{n}))w^{(n)}(x_{n})|\leq\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B_{i_{N}})}\|\varphi\|\leq C\|\varphi\|, which contradicts the fact that \varphi is an absolutely mean irregular vector for C_{w,f} . (ii) \Rightarrow (iii): Without loss of generality, we may assume that B_{i}\neq\emptyset for all i\in\mathbb{N} . By ( 44 ), for each i\in\mathbb{N} , there exists a continuous map \phi_{i}:\Omega\to[0,1] such that \operatorname{supp}\phi_{i}\subset B_{i+1} and \phi_{i}=1 on \overline{B_{i}} . By ( 45 ), \phi_{i}\in\mathfrak{X} for all i\in\mathbb{N} . Since \|\phi_{i}\|=1 and \frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\phi_{i})\|\geq\frac{1}{N}\sum_{n=1}^{N}\|(\phi_{i}\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B_{i})}=\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B_{i})}, it follows from ( 46 ) that C_{w,f} is not absolutely Cesàro bounded in \mathfrak{X} . (iii) \Rightarrow (i): Suppose, by contradiction, that C_{w,f} is not mean Li-Yorke chaotic. If \phi\in C_{0}(\Omega) has support contained in a relatively compact open set B in \Omega satisfying ( 43 ), then \displaystyle\liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\phi)\| \displaystyle=\liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(\phi\circ f^{n})\cdot w^{(n)}\|_{f^{-n}(B)} \displaystyle\leq\|\phi\|\liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B)}=0. Since C_{w,f} does not admit an absolutely mean semi-irregular vector, we conclude that \lim_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\phi)\|=0. This implies that the set \mathcal{R}_{1} of all \varphi\in\mathfrak{X} such that \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|=0 is residual in \mathfrak{X} . Now, by arguing as at the end of the proof of Theorem 69 , we get a contradiction. ∎ Theorem 75 . Consider a weighted composition operator C_{w,f} on C_{0}(\Omega) . If \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B)}=0 for every relatively compact open set B in \Omega , then the following assertions are equivalent: (i) C_{w,f} is mean Li-Yorke chaotic; (ii) C_{w,f} has a residual set of absolutely mean irregular vectors; (iv) C_{w,f} is not absolutely Cesàro bounded. Proof. Given \varphi\in C_{c}(\Omega) , take a relatively compact open set B in \Omega with \operatorname{supp}\varphi\subset B . By the hypothesis, \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|(C_{w,f})^{n}(\varphi)\|\leq\liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B)}\|\varphi\|=0. Since C_{c}(\Omega) is dense in C_{0}(\Omega) , the equivalences between conditions (i), (ii) and (iii) follow from [ 10 , Theorem 22] . ∎ Remark 76 . (a) In the case C_{0}(\Omega) is separable, [ 10 , Theorem 17] says that (ii) is equivalent to (ii’) C_{w,f} is densely mean Li-Yorke chaotic. (b) If the space C_{0}(\Omega) is separable and \lim_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\|w^{(n)}\|_{f^{-n}(B)}=0 for every relatively compact open set B in \Omega , then [ 10 , Theorem 29] implies that (i)–(iii) are also equivalent to (iv) C_{w,f} admits a dense absolutely mean irregular manifold. Let us now present some applications to weighted shifts and weighted translation operators. Corollary 77 . For weighted shifts B_{w} on c_{0}(\mathbb{N}) and weighted translation operators T_{w} on C_{0}[1,\infty) , properties (i)–(iv) above are always equivalent to each other. Proof. The conditions in Remark 76 (b) are satisfied. ∎ Corollary 78 . A weighted shift B_{w} on c_{0}(\mathbb{Z}) with nonzero weights is mean Li-Yorke chaotic if and only if it is not absolutely Cesàro bounded and \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}|w_{-n}\cdots w_{-1}|=0. (47) Proof. The proof is analogous to that of Corollary 73 , but we have to use Theorems 74 and 75 instead of Theorems 69 and 70 . ∎ Corollary 79 . Given a weighted translation operator T_{w} on C_{0}(\mathbb{R}) , let \mathfrak{X} be the closed subspace of C_{0}(\mathbb{R}) generated by the functions \varphi\in C_{0}(\mathbb{R}) whose support is contained in a bounded open set B in \mathbb{R} satisfying \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\Big(\sup_{x\in B}|w(x-n)\cdots w(x-1)|\Big)=0. Then, the following assertions are equivalent: (i) T_{w} is mean Li-Yorke chaotic; (ii) There exists a sequence (B_{i})_{i\in\mathbb{N}} of bounded open sets in \mathbb{R} such that \overline{B_{i}}\subset B_{i+1}\ \text{ for all }i\in\mathbb{N}, \liminf_{N\to\infty}\frac{1}{N}\sum_{n=1}^{N}\Big(\sup_{x\in B_{i}}|w(x-n)\cdots w(x-1)|\Big)=0\ \text{ for all }i\in\mathbb{N} and \sup\Big\{\frac{1}{N}\sum_{n=1}^{N}\Big(\sup_{x\in B_{i}}|w(x-n)\cdots w(x-1)|\Big):i,N\in\mathbb{N}\Big\}=\infty. (iii) T_{w} is not absolutely Cesàro bounded in \mathfrak{X} . Proof. This is a particular case of Theorem 74 . ∎ Nilson C. Bernardes Jr. Departamento de Matemática Aplicada, Instituto de Matemática, Universidade Federal do Rio de Janeiro, Caixa Postal 68530, RJ 21941-909, Brazil. e-mail address : ncbernardesjr@gmail.com Antonio Bonilla Departamento de Análisis Matemático and Instituto de Matemáticas y Aplicaciones (IMAULL), Universidad de La Laguna, C/Astrofísico Francisco Sánchez, s/n, 38721 La Laguna, Tenerife, Spain email address: abonilla@ull.edu.es João V. A. Pinto Departamento de Matemática Aplicada, Instituto de Matemática, Universidade Federal do Rio de Janeiro, Caixa Postal 68530, RJ 21941-909, Brazil. e-mail address : joao.pinto@ufu.br