SLIDE 1
Linear Dynamical Properties of Weighted Backward Shifts on Spaces of Real Analytic Functions
Can Deha Karıksız (joint work with late Pawe l Doma´ nski)
¨ Ozye˘ gin University, Istanbul deha.kariksiz@ozyegin.edu.tr
03.07.2018
SLIDE 2 Outline
1 Introduction 2 Conditions on linear dynamical properties using dynamical
transference principles
3 Conditions on linear dynamical properties via eigenvalues
SLIDE 3
Introduction
Linear Dynamical Properties
An operator T on a topological vector space X is called hypercyclic if there is some x ∈ X such that the set {x, Tx, T 2x, · · · , T nx, · · · }, called the orbit of x under T, is dense in X.
SLIDE 4
Introduction
Linear Dynamical Properties
An operator T on a topological vector space X is called hypercyclic if there is some x ∈ X such that the set {x, Tx, T 2x, · · · , T nx, · · · }, called the orbit of x under T, is dense in X. topologically transitive if for any pair of nonempty open subsets U, V of X, there exists some n ∈ N such that T n(U) ∩ V = ∅.
SLIDE 5
Introduction
Linear Dynamical Properties
An operator T on a topological vector space X is called mixing if for any pair of nonempty open subsets U, V of X, there exists some N ∈ N such that for every n ≥ N we have T n(U) ∩ V = ∅.
SLIDE 6
Introduction
Linear Dynamical Properties
An operator T on a topological vector space X is called mixing if for any pair of nonempty open subsets U, V of X, there exists some N ∈ N such that for every n ≥ N we have T n(U) ∩ V = ∅. chaotic if T is topologically transitive and has a dense set of periodic points.
SLIDE 7 Introduction
Weighted Backward Shifts on Fr´ echet Sequence Spaces
Let X be a Fr´ echet sequence space with canonical unit sequences
- en. For a sequence of nonzero scalars ω = (ωn)n∈N, the operator
Bω : X → X defined by Bωen = ωnen−1, n ≥ 1, e0 = 0, is called a weighted backward shift on X.
SLIDE 8 Introduction
Weighted Backward Shifts on Fr´ echet Sequence Spaces
Let X be a Fr´ echet sequence space with canonical unit sequences
- en. For a sequence of nonzero scalars ω = (ωn)n∈N, the operator
Bω : X → X defined by Bωen = ωnen−1, n ≥ 1, e0 = 0, is called a weighted backward shift on X. Theorem Let Bω : X → X be a weighted backward shift acting on a Fr´ echet sequence space X in which (en)n∈N is a basis. (i) Bω is hypercyclic ⇔ there is an increasing sequence (nk)k∈N
- f positive integers such that (nk
ν=1 ων)−1 enk → 0 in X as
k → ∞.
SLIDE 9 Introduction
Weighted Backward Shifts on Fr´ echet Sequence Spaces
Let X be a Fr´ echet sequence space with canonical unit sequences
- en. For a sequence of nonzero scalars ω = (ωn)n∈N, the operator
Bω : X → X defined by Bωen = ωnen−1, n ≥ 1, e0 = 0, is called a weighted backward shift on X. Theorem Let Bω : X → X be a weighted backward shift acting on a Fr´ echet sequence space X in which (en)n∈N is a basis. (i) Bω is hypercyclic ⇔ there is an increasing sequence (nk)k∈N
- f positive integers such that (nk
ν=1 ων)−1 enk → 0 in X as
k → ∞. (ii) Bω is mixing ⇔ (n
ν=1 ων)−1 en → 0 in X as n → ∞.
SLIDE 10 Introduction
Weighted Backward Shifts on Fr´ echet Sequence Spaces
Let X be a Fr´ echet sequence space with canonical unit sequences
- en. For a sequence of nonzero scalars ω = (ωn)n∈N, the operator
Bω : X → X defined by Bωen = ωnen−1, n ≥ 1, e0 = 0, is called a weighted backward shift on X. Theorem Let Bω : X → X be a weighted backward shift acting on a Fr´ echet sequence space X in which (en)n∈N is an unconditional basis. (i) Bω is hypercyclic ⇔ there is an increasing sequence (nk)k∈N
- f positive integers such that (nk
ν=1 ων)−1 enk → 0 in X as
k → ∞. (ii) Bω is mixing ⇔ (n
ν=1 ων)−1 en → 0 in X as n → ∞.
(iii) Bω is chaotic ⇔ ∞
n=1 (n ν=1 ων)−1 en converges in X.
SLIDE 11
Introduction
Spaces of Real Analytic Functions
Let A(Ω) denote the space of all complex-valued real analytic functions on an open set Ω in R.
SLIDE 12
Introduction
Spaces of Real Analytic Functions
Let A(Ω) denote the space of all complex-valued real analytic functions on an open set Ω in R. Equivalent Topologies on A(Ω) (Martineau 1966) Projective limit topology: A(Ω) = projN∈N H(KN) = projN∈N indn∈N H∞(UN,n), where (KN)N∈N is a compact increasing exhaustion of Ω, and (UN,n)n∈N are fundamental sequences of neighborhoods of KN for each N.
SLIDE 13
Introduction
Spaces of Real Analytic Functions
Let A(Ω) denote the space of all complex-valued real analytic functions on an open set Ω in R. Equivalent Topologies on A(Ω) (Martineau 1966) Projective limit topology: A(Ω) = projN∈N H(KN) = projN∈N indn∈N H∞(UN,n), where (KN)N∈N is a compact increasing exhaustion of Ω, and (UN,n)n∈N are fundamental sequences of neighborhoods of KN for each N. Inductive limit topology: A(Ω) = ind H(U) where the inductive limit is taken over all complex neighborhoods of Ω.
SLIDE 14
Introduction
Spaces of Real Analytic Functions
Main difficulties These locally convex topologies on A(Ω) are not metrizable, hence A(Ω) is not a Fr´ echet space. (Doma´ nski, Vogt 2000) A(Ω) has no Schauder basis.
SLIDE 15
Introduction
Weighted Backward Shifts on A(Ω)
Definition Given a sequence of nonzero scalars ω = (ωn)n∈N, a continuous linear operator Bω : A(Ω) → A(Ω), that sends the monomials xn to ωn−1xn−1 for all n ≥ 1, the unit function to the zero function, is called a weighted backward shift with the weight sequence ω.
SLIDE 16
Introduction
Weighted Backward Shifts on A(Ω)
Definition Given a sequence of nonzero scalars ω = (ωn)n∈N, a continuous linear operator Bω : A(Ω) → A(Ω), that sends the monomials xn to ωn−1xn−1 for all n ≥ 1, the unit function to the zero function, is called a weighted backward shift with the weight sequence ω. Problem: How to characterize well-defined weighted backward shifts on A(Ω)?
SLIDE 17
Introduction
Hadamard Multipliers on A(Ω)
A linear continuous operator M : A(Ω) → A(Ω) is called a (Hadamard) multiplier whenever every monomial is an eigenvector.
SLIDE 18
Introduction
Hadamard Multipliers on A(Ω)
A linear continuous operator M : A(Ω) → A(Ω) is called a (Hadamard) multiplier whenever every monomial is an eigenvector. Observation There is a one-to-one correspondence between the weighted backward shifts and the multipliers on A(Ω) whenever 0 ∈ Ω.
SLIDE 19
Introduction
Hadamard Multipliers on A(Ω)
A linear continuous operator M : A(Ω) → A(Ω) is called a (Hadamard) multiplier whenever every monomial is an eigenvector. Observation There is a one-to-one correspondence between the weighted backward shifts and the multipliers on A(Ω) whenever 0 ∈ Ω. In this case, we can use the representation theorems for Hadamard multipliers on A(Ω) (Doma´ nski, Langenbruch 2012). Proposition Let Ω ⊂ R with 0 ∈ Ω be an open set. Then, TFAE: (i) Bω is a w.b.s. with the weight sequence ω = (ωn). (ii) Bω maps a function ∞
n=0 fnzn around zero into a real
analytic function on Ω represented around zero by the series ∞
n=0 fnωn−1zn−1.
SLIDE 20
Conditions Using Dynamical Transference Principles
An operator T on X is called quasiconjugate to an operator S on Y via a continuous map φ : Y → X with dense range if T ◦ φ = φ ◦ S. Linear dynamical properties like hypercyclicity, mixing, and chaos are preserved under quasiconjugacy.
SLIDE 21
Conditions Using Dynamical Transference Principles
An operator T on X is called quasiconjugate to an operator S on Y via a continuous map φ : Y → X with dense range if T ◦ φ = φ ◦ S. Linear dynamical properties like hypercyclicity, mixing, and chaos are preserved under quasiconjugacy. By considering Bω as an operator acting on the space H(C) of entire functions, and the space H({0}) of germs of holomorphic functions at zero, we obtain the following two quasiconjugacies. H(C)
Bω
− − → H(C) ↓ ↓ A(Ω)
Bω
− − → A(Ω) A(Ω)
Bω
− − → A(Ω) ↓ ↓ H({0})
Bω
− − → H({0})
SLIDE 22
Conditions Using Dynamical Transference Principles
Using the quasiconjugacy involving H(C), we obtain the following sufficient conditions. Proposition (Doma´ nski, K. 2018) Let Ω ⊂ R be an open set with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms.
SLIDE 23 Conditions Using Dynamical Transference Principles
Using the quasiconjugacy involving H(C), we obtain the following sufficient conditions. Proposition (Doma´ nski, K. 2018) Let Ω ⊂ R be an open set with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) if there is a sequence (nk) such that for every R > 0, nk
ων−1 −1 Rnk → 0 then Bω is hypercyclic.
SLIDE 24 Conditions Using Dynamical Transference Principles
Using the quasiconjugacy involving H(C), we obtain the following sufficient conditions. Proposition (Doma´ nski, K. 2018) Let Ω ⊂ R be an open set with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) if there is a sequence (nk) such that for every R > 0, nk
ων−1 −1 Rnk → 0 then Bω is hypercyclic. (b) if for every R > 0,
ων−1 −1 Rn → 0 then Bω is mixing and chaotic.
SLIDE 25
Conditions Using Dynamical Transference Principles
Using the quasiconjugacy involving H({0}) and a construction due to Bonet, we obtain the following necessary conditions. Proposition (Doma´ nski, K. 2018) Let Ω ⊂ R be an open set with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms.
SLIDE 26 Conditions Using Dynamical Transference Principles
Using the quasiconjugacy involving H({0}) and a construction due to Bonet, we obtain the following necessary conditions. Proposition (Doma´ nski, K. 2018) Let Ω ⊂ R be an open set with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) if Bω is topologically transitive then there is a sequence (nk) and there is R > 0 such that nk
ων−1 −1 Rnk → 0.
SLIDE 27 Conditions Using Dynamical Transference Principles
Using the quasiconjugacy involving H({0}) and a construction due to Bonet, we obtain the following necessary conditions. Proposition (Doma´ nski, K. 2018) Let Ω ⊂ R be an open set with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) if Bω is topologically transitive then there is a sequence (nk) and there is R > 0 such that nk
ων−1 −1 Rnk → 0. (b) if Bω is mixing then there is R > 0 such that
ων−1 −1 Rn → 0.
SLIDE 28 Conditions Using Dynamical Transference Principles
Some Problems
Observation There are weight sequences satisfying the conditions nk
ων−1 −1 Rnk → 0
ων−1 −1 Rn → 0 for some R > 0, but not all R > 0.
SLIDE 29 Conditions Using Dynamical Transference Principles
Some Problems
Observation There are weight sequences satisfying the conditions nk
ων−1 −1 Rnk → 0
ων−1 −1 Rn → 0 for some R > 0, but not all R > 0. Example If we take the unweighted backward shift on A(Ω), that is, ω = (ωn) where ωn = 1 for all n ∈ N, then
ων−1 −1 Rn = Rn → 0 only for 0 < R < 1.
SLIDE 30
Conditions using Eigenvalues
Godefroy-Shapiro criterion
Godefroy-Shapiro criterion Let T be an operator on a topological vector space X. If the subspaces X0 := span{x ∈ X : Tx = λx for some λ with |λ| < 1}, Y0 := span{x ∈ X : Tx = λx for some λ with |λ| > 1} are dense in X, then T is mixing. If, moreover, X is a complex space and the subspace Z0 := span{x ∈ X : Tx = eαπi for some rational α} is dense in X, then T is chaotic.
SLIDE 31
Conditions using Eigenvalues
Description of Eigenvalues of A(Ω)
Definition A set U is called the star of holomorphy of a germ f ∈ H({0}) if it is a maximal star-like set in C around zero such that f extends holomorphically to V . A set U is star-like if for any z ∈ U we have {tz : t ∈ [0, 1]} ⊂ U.
SLIDE 32 Conditions using Eigenvalues
Description of Eigenvalues of A(Ω)
Proposition Let Ω ⊆ R be an interval with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. If sup
n
ων−1 −1 Rn < ∞ for some R > 0, then (i) zero is an eigenvalue with eigenspace of dimension one,
SLIDE 33 Conditions using Eigenvalues
Description of Eigenvalues of A(Ω)
Proposition Let Ω ⊆ R be an interval with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. If sup
n
ων−1 −1 Rn < ∞ for some R > 0, then (i) zero is an eigenvalue with eigenspace of dimension one, (ii) λ = 0 is an eigenvalue of Bω if and only if λΩ is contained in the star of holomorphy of the universal eigenfunction Eω(z) := zn+1 +
∞
ωk −1 zj.
SLIDE 34 Conditions using Eigenvalues
Description of Eigenvalues of A(Ω)
Proposition Let Ω ⊆ R be an interval with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. If sup
n
ων−1 −1 Rn < ∞ for some R > 0, then (i) zero is an eigenvalue with eigenspace of dimension one, (ii) λ = 0 is an eigenvalue of Bω if and only if λΩ is contained in the star of holomorphy of the universal eigenfunction Eω(z) := zn+1 +
∞
ωk −1 zj. If λ = 0 is an eigenvalue of Bω, then its eigenspace is
- ne-dimensional, spanned by the function fλ(z) := Eω(λz).
SLIDE 35
Conditions using Eigenvalues
Application of Godefroy-Shapiro Criterion
Theorem (Doma´ nski, K. 2018) Let Ω ⊆ R be an interval with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) If Bω has a universal eigenfunction such that its star of holomorphy contains at least one interval λΩ for some λ ∈ C with |λ| > 1, then Bω is mixing.
SLIDE 36
Conditions using Eigenvalues
Application of Godefroy-Shapiro Criterion
Theorem (Doma´ nski, K. 2018) Let Ω ⊆ R be an interval with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) If Bω has a universal eigenfunction such that its star of holomorphy contains at least one interval λΩ for some λ ∈ C with |λ| > 1, then Bω is mixing. (b) If additionally to (a) the star of holomorphy of the universal eigenfunction contains a strip {z ∈ C : |z − tλ| < ε for some t ∈ R} for some ε > 0, then Bω is also hypercyclic.
SLIDE 37
Conditions using Eigenvalues
Application of Godefroy-Shapiro Criterion
Theorem (Doma´ nski, K. 2018) Let Ω ⊆ R be an interval with 0 ∈ Ω, and Bω : A(Ω) → A(Ω) be a weighted backward shift such that ω has no zero terms. (a) If Bω has a universal eigenfunction such that its star of holomorphy contains at least one interval λΩ for some λ ∈ C with |λ| > 1, then Bω is mixing. (b) If additionally to (a) the star of holomorphy of the universal eigenfunction contains a strip {z ∈ C : |z − tλ| < ε for some t ∈ R} for some ε > 0, then Bω is also hypercyclic. (c) If additionally to (a) the star of holomorphy of the universal eigenfunction contains a set of the form {teπai : t ∈ Ω, a ∈ (a1, a2)} for some −∞ < a1 < a2 < ∞ then Bω is also chaotic.
SLIDE 38
Conditions using Eigenvalues
Some Examples
Examples Let Ω be an interval containing zero. The following weighted backward shifts are chaotic, mixing, and hypercyclic. (i) the unweighted shift B1,
SLIDE 39
Conditions using Eigenvalues
Some Examples
Examples Let Ω be an interval containing zero. The following weighted backward shifts are chaotic, mixing, and hypercyclic. (i) the unweighted shift B1, (ii) the differentiation operator D : A(Ω) → A(Ω) which corresponds to the weighted shift with ωn = n + 1,
SLIDE 40
Conditions using Eigenvalues
Some Examples
Examples Let Ω be an interval containing zero. The following weighted backward shifts are chaotic, mixing, and hypercyclic. (i) the unweighted shift B1, (ii) the differentiation operator D : A(Ω) → A(Ω) which corresponds to the weighted shift with ωn = n + 1, (iii) the weighted backward shift Qk : A(Ω) → A(Ω), k > 0, Qk(f )(z) = 1
0 f ′(zt)tk dt which corresponds to the weighted
backward shift with ωn =
n+1 n+1+k .
SLIDE 41 References I
nski, C.D. Karıksız, Eigenvalues and dynamical properties of weighted backward shifts on the space of real analytic functions, Studia Math. 242 (2018), 57–78.
nski, M. Langenbruch, Representation of multipliers
- n spaces of real analytic functions, Analysis 32 (2012),
137–162.
nski, D. Vogt, The space of real-analytic functions has no basis, Studia Math. 142 (2) (2000), 187–200.
- A. Martineau, Sur la topologie des espaces de fonctions
holomorphes, Math. Ann. 163 (1966), 62–88.
SLIDE 42
The End