@ Appl. Gen. Topol. 15, no. 2(2014), 175-181doi:10.4995/agt.2014.1841 c© AGT, UPV, 2014 Lifting dynamical properties to hyperspaces and hyperspace suspension Dania Masood and Pooja Singh Department of Mathematics, University of Allahabad, Allahabad - 211 002, India. (dpub@pphmj.com, poojasingh 07@rediffmail.com) Abstract For a dynamical system (X, f), the passage of various dynamical pro- perties such as transitivity, total transitivity, weakly mixing, mixing, topological exactness, topological conjugacy, to the hyperspace C(X) of X consisting of nonempty closed connected subsets of X, and also to the hyperspace suspension HS(X) of X, have been considered and studied. 2010 MSC: 54H20; 37B99. Keywords: transitivity; total transitivity; weakly mixing; mixing; topolog- ical exactness; topological conjugacy; Vietoris topology, hyper- space suspension. 1. Introduction Throughout the paper, (X,f) denotes a dynamical system, where X is a compact metric space and f is a selfmap on X. By a map, we mean a continuous map. The symbol N denotes the set of positive integers. Let (X,f) be a dynamical system. Then f is called (i) transitive if for a pair of nonempty open sets U, V of X, there exists an n ∈ N, such that fn(U)∩V 6= φ, (ii) totally transitive if the map fk is transitive, for each k ∈ N, (iii) weakly mixing if for pairs of nonempty open sets U1, U2 and V1, V2 of X, there exists an n ∈ N, such that fn(Ui) ∩ Vi 6= φ, for i = 1, 2, (iv) mixing if for a pair of nonempty open sets U, V of X, there exists an N ∈ N, such that fn(U) ∩ V 6= φ, n ≥ N, (v) topologically exact if for an open set U of X, there exists an n ∈ N, such that fn(U) = X. It is known that, if f is weakly Received 7 November 2013 – Accepted 15 March 2014 http://dx.doi.org/10.4995/agt.2014.1841 D. Masood and P. Singh mixing and U1,U2, . . . ,Un, V1,V2, . . . ,Vn are nonempty open sets of X, then there exists a k ≥ 1 such that fk(Ui) ∩ Vi 6= φ, for i = 1,2, . . . ,n [12]. The dynamical system (X,f) is said to possess a dynamical property P , if f has P . Two dynamical systems (X,f) and (Y,g) are said to be topologically conju- gate if there exists a homeomorphism ϕ from X to Y such that g ◦ ϕ = ϕ ◦ f. By 2X, we denote the set consisting of all nonempty closed subsets of the space X. On it, (i) the upper Vietoris topology τU , has BU ≡ {< U1, . . . ,Um >: U1, . . . ,Um are open sets of X}, where < U1, . . . ,Um > = {A ∈ 2 X : A ⊆ ⋃m i=1 Ui}, as a basis, (ii) the lower Vietoris topology τL, has BL ≡ {< U ′ 1, . . . ,U ′ m >: U ′ 1, . . . ,U ′ m are open sets of X}, where < U′1, . . . ,U ′ m > = {A ∈ 2 X : A ∩ U′i 6= φ, i = 1, . . . ,m}, as a basis, and (iii) the Vietoris topology τ, has B ≡ {< V1, . . . ,Vn >: V1, . . . ,Vn are open sets of X}, where < V1, . . . ,Vn >= {A ∈ 2 X : A ⊆ ⋃n i=1 Vi, A ∩ Vi 6= φ, i = 1, . . . ,n}, as a basis. The hyperspace C(X), denotes the subspace of 2X consisting of nonempty closed connected subsets of X. We shall write C(X) to mean (C(X),τ). The upper and lower Vietoris topologies on C(X) are specified by writing (C(X),τU ) and (C(X),τL), respectively. For nonempty open sets U1, . . . ,Un of X and U = Ui, for some i ≤ n, when m ≥ 1, we may write, as in [1], < U1, . . . ,Un > alternatively in the form < U1, . . . ,Un,U, . . . ,U ︸ ︷︷ ︸ m >. For a nonempty basic open set U = < U1, . . . ,Um > of C(X), by ⋃ U we mean ⋃m i=1 Ui. Observe that X is embedded into C(X) by the embedding x 7→ {x}. Also, every continuous selfmap f on X, induces in a natural way, a selfmap f′ on 2X [2]. The hyperspace C(X ) is invariant with respect to f′, and hence we obtain a dynamical system (C(X), f̄), where f̄ = f′|C(X). For k ∈ N, the space Fk(X), denotes the collection of those nonempty closed subsets of X that consist of atmost k elements. The set F1(X) consisting of all singletons in X, is a closed subspace of C(X), and the quotient space C(X)/F1(X), denoted by HS(X ), is known as the hyperspace suspension of X [10]. Further, for a dynamical system (X,f), the pair (HS(X ), HS(f )) constitutes a dynamical system with the evolution map HS(f ) sending [A] ∈ HS(X) to (qX ◦ f ◦ q −1 X )([A]), where qX denotes the cannonical projection from C(X ) to HS(X ) [5]. That HS(f ) is continuous follows from [4, 4.3, p. 146]. It is pertinent to mention that HS taking X to HS(X ) and f to HS(f ) describes a covariant functor from the category of topological spaces to itself. A brief exposition of our work is as follows. If (X,f) is transitive, then ( (C(X),τU ), f̄ ) is transitive. However, ( (C(X),τL), f̄ ) is transitive if X is pathconnected and f is weakly mixing. The same is the state of affairs so far as the total transitivity and weakly mixing are concerned. It is known that the transitivity and total transitivity do not necessarily lift to C(X) endowed c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 2 176 Lifting dynamical properties to hyperspaces and hyperspace suspension with the Vietoris topology. For a dynamical system (X,f) with f mixing, it is obtained that f̄ on (C(X),τL) continues to be mixing provided X is pathconnected. All these constitute the contents of Section 3. Section 4 is devoted to the passage of these dynamical properties to the dynamical system (HS(X),HS(f)) from the dynamical system (X,f). It has been obtained that all these properties get lifted smoothly. Finally, it is obtained that topological conjugacy is preserved under the hyperspace suspension. We begin with the preliminaries in Section 2, wherein, certain results scat- tered across various references are stated that we subsequently use. 2. Preliminaries A topological graph is a connected compact Hausdorff space G for which there exists a finite collection of subspaces Ii, i = 1,2, . . . ,s such that G = ⋃s i=1 Ii, where each Ii is homeomorphic to a compact interval, and each intersection Ii ∩Ij, for i 6= j is finite. Equivalently, a graph is a one-dimensional connected, compact polyhedron. Below we state a result related to the lift of transitivity to the hyperspace C(G) of a dynamical system (G,f), where G is a graph and f is transitive. Proposition 2.1 ([9, Cor. 29]). For a dynamical system (G,f), the map f̄ is never transitive. Remark 2.2. Note that the map f̄ has been denoted by f̃ in [9]. Denoting by f̂ the map induced on K(X) consisting of all nonempty compact subsets of X, we have the following: Proposition 2.3 ([7, Lemma 5]). For a dynamical system (X,f), the following are equivalent: (1) f is topologically exact. (2) f̂ is topologically exact. Remark 2.4. In [7], the map f̂ has been denoted by f̄, which we have used for the map induced on the hyperspace C(X ). Proposition 2.5 ([9, Lemma 4]). If dynamical systems (X,f) and (Y,g) are topologically conjugate, then the same holds for the induced dynamical systems (2X,f′) and (2Y ,g′), and also for (C(X), f̄) and (C(Y), ḡ). Remark 2.6. In [9], the maps f′,g′, f̄ and ḡ have been denoted by f̄, ḡ, f̃ and g̃, respectively. Proposition 2.7 ([8]). Topological transitivity is preserved under topological conjugacy. c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 2 177 D. Masood and P. Singh 3. Dynamical properties and C(X) Theorem 3.1. Let (X,f) be a dynamical system with f transitive. Then the induced map f̄ on (C(X),τU ) is transitive. Proof. Consider a pair of nonempty basic open sets U and V of C(X). Then ⋃ U and ⋃ V constitute a pair of nonempty open sets in X. Since f is transitive, there is a natural number l such that for some y ∈ ⋃ U, fl(y) ∈ ⋃ V. That f̄l(U) intersects V follows by noting that {y} ∈ U and f̄l({y}) ∈ V. � Proposition 2.1 shows that there is no dearth of dynamical systems whose evolution maps are transitive possessing induced evolution maps which fail to be transitive. Below we provide an example for the sake of illustration. Example 3.2. The irrational rotation Rα : S 1 → S1 defined by Rα(θ) = θ + α, θ ∈ [0,2π), where α is an irrational number, and the multiplicative group S1 is identified with the additive group [0,2π) mod 2π, is a transitive map [3, 11]. Let Rα be the map induced on C(S1), which is homeomorphic to D2 ≡ {(r,θ) | r ∈ [0,1] and θ ∈ [0,2π)} [2, 6], via the homeomorphism ψ, sending A ∈ C(S1) to (1 − lA/2π,mid A) ∈ D 2, where lA and mid A, denote the arclength and the middle point of A, respectively. The map g : D2 → D2 defined by g(r,θ) = (r,θ + α), r ∈ [0,1], and θ ∈ [0,2π), is such that g ◦ ψ = ψ ◦ Rα, making the systems (C(S 1),Rα) and (D 2,g) topologically conjugate. Since the orbit of no element of D2 under g is dense in it, the map g is not transitive [8], and hence, by Proposition 2.7, Rα is also not transitive. Theorem 3.3. Let (X,f) be a dynamical system with f totally transitive. Then the induced map f̄ on (C(X),τU ) is totally transitive. Proof. Since for a natural number n, fn = f̄n, the result follows from Theorem 3.1. � Theorem 3.4. Let (X,f) be a dynamical system with f weakly mixing. Then the induced map f̄ on (C(X),τU ) is weakly mixing. Proof. Consider pairs of nonempty basic open sets U1,V1 and U2,V2 of C(X). Then ⋃ U1, ⋃ V1 and ⋃ U2, ⋃ V2 are pairs of nonempty open sets of X. Since f is weakly mixing, there are n ∈ N, y ∈ ⋃ U1 and y ′ ∈ ⋃ U2 such that f n(y) ∈ ⋃ V1, and f n(y′) ∈ ⋃ V2. That f̄ n(U1) ∩ V1 6= φ and f̄ n(U2) ∩ V2 6= φ, follows by observing that {y} ∈ U1, f̄ n({y}) ∈ V1, {y ′} ∈ U2 and f̄ n({y′}) ∈ V2. � Theorem 3.5. Let (X,f) be a dynamical system with f weakly mixing. If X is pathconnected, then the induced map f̄ on (C(X),τL) is weakly mixing. c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 2 178 Lifting dynamical properties to hyperspaces and hyperspace suspension Proof. Consider pairs < U1, . . . ,Um >,< V1, . . . ,Vm >, and < U ′ 1, . . . ,U ′ n >,< V ′1, . . . ,V ′ n > of nonempty basic open sets of C(X). Then the pairs (Ui,Vi), i = 1, . . . ,m, and (U′j,V ′ j ), j = 1, . . . ,n, consist of nonempty open sets of X. Since f is weakly mixing, there is a k ∈ N, such that fk(Ui) ∩ Vi 6= φ, i = 1, . . . ,m, and fk(U′j) ∩ V ′ j 6= φ, j = 1, . . . ,n. Thus there are elements xi ∈ Ui, x ′ j ∈ U ′ j such that fk(xi) ∈ Vi, and f k(x′j) ∈ V ′ j , i = 1, . . . ,m, j = 1, . . . ,n. Let A and B be paths containing all xi’s, and all x ′ j’s, respectively. Then A ∈ < U1, . . . ,Um >, B ∈ < U ′ 1, . . . ,U ′ n >, and f̄ k(A) ∈ < V1, . . . ,Vm >, f̄ k(B) ∈ < V ′1, . . . ,V ′ n >. Hence, the result. � Because, for a given dynamical system (X,f), the weakly mixing of f implies that f is transitive and also totally transitive, we have the following: Corollary 3.6. Let (X,f) be a dynamical system with f weakly mixing. If X is pathconnected, then the induced map f̄ on (C(X),τL) is transitive and also totally transitive. Theorem 3.7. Let (X,f) be a dynamical system with f mixing. Then the induced map f̄ on (C(X),τU ) is mixing. Proof. For a pair U,V of nonempty basic open sets of C(X), ⋃ U, ⋃ V constitute a pair of nonempty open sets of X. Since f is mixing, there is an N ∈ N, such that for n ≥ N, fn( ⋃ U) intersects ⋃ V, which implies that f̄n(U) intersects V, for n ≥ N. Hence, the result. � Theorem 3.8. Let (X,f) be a dynamical system with f mixing. If X is pathconnected, then the induced map f̄ on (C(X),τL) is mixing. Proof. Consider a pair < U1, . . . ,Um >,< V1, . . . ,Vm > of nonempty basic open sets of C(X). Since f is mixing, for each pair (Ui,Vi), i = 1, . . . ,m, of open sets of X, there exist Ni ∈ N such that f k(Ui)∩Vi 6= φ, whenever k ≥ Ni. Set N = max {Ni : i = 1, . . . ,m}. Then f k(Ui) ∩ Vi 6= φ, for i = 1, . . . ,m, and k ≥ N. Thus for i = 1, . . . ,m, k ≥ N, there are xik ∈ Ui such that fk(xik ) ∈ Vi. Let Ak be a path containing all xik ’s. Since f̄ k(Ak) lies in f̄k(< U1, . . . ,Um >) as well as in < V1, . . . ,Vm >, the result follows. � 4. Dynamical properties and HS(X) We begin with the following: Lemma 4.1. For n ∈ N, (HS(f))n ≡ (qX ◦ f n ◦ q−1 X ). Proof. The result follows by induction on n. � Theorem 4.2. Let (X,f) be a dynamical system such that (C(X), f̄) is tran- sitive. Then the dynamical system (HS(X),HS(f)) is also transitive. Proof. For a pair U,V of nonempty open sets of HS(X), q−1 X (U) and q−1 X (V ) constitute a pair of nonempty open sets of C(X). Since f̄ is transitive, there is an n ∈ N such that fn(q−1 X (U)) ∩ q−1 X (V ) 6= φ. Applying qX and using Lemma 4.1, the result follows. � c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 2 179 D. Masood and P. Singh Theorem 4.3. Let (X,f) be a dynamical system such that (C(X), f̄) is to- tally transitive. Then the dynamical system (HS(X),HS(f)) is also totally transitive. Proof. For n ∈ N, Theorem 4.2 and Lemma 4.1 provide the transitivity of (HS(f))n. Thus, HS(f) is totally transitive. � Theorem 4.4. Let (X,f) be a dynamical system such that (C(X), f̄) is weakly mixing. Then the dynamical system (HS(X),HS(f)) is also weakly mixing. Proof. Consider pairs U1,U2 and V1,V2 of nonempty open sets of HS(X). Then q−1 X (U1),q −1 X (U2) and q −1 X (V1),q −1 X (V2) are pairs of nonempty open sets of C(X). The weakly mixing property of the map f̄ implies the existence of an n ∈ N, such that fn(q−1 X (Ui)) ∩q −1 X (Vi) 6= φ, for i = 1,2. Now arguing as in Theorem 4.2, we have the result. � Theorem 4.5. Let (X,f) be a dynamical system such that (C(X), f̄) is mixing. Then the dynamical system (HS(X),HS(f)) is also mixing. Proof. For a pair U,V of nonempty open sets of HS(X), q−1 X (U) and q−1 X (V ) constitute a pair of nonempty open sets of C(X). The mixing property of f̄ on C(X) provides an N ∈ N, such that fn(q−1 X (U)) ∩ q−1 X (V ) 6= φ, for n ≥ N. Now proceeding as in the proof of Theorem 4.2, we have the result. � Theorem 4.6. Let (X,f) be a dynamical system such that (C(X), f̄) is topo- logically exact. Then the dynamical system (HS(X),HS(f)) is also topologi- cally exact. Proof. Let U be a nonempty open set of HS(X). Then q−1 X (U) is a nonempty open set of C(X). The topological exactness of f̄ on C(X) yields an n ∈ N, such that fn(q−1 X (U)) = C(X). Applying qX and using Lemma 4.1, the result follows. � Theorem 4.7. If the dynamical systems (X,f) and (Y,g) are topologically conjugate, then the dynamical systems (HS(X),HS(f)) and (HS(Y ),HS(g)) are also topologically conjugate. Proof. Let ϕ be a conjugacy between the pairs (X,f) and (Y,g). Then by Proposition 2.5, the dynamical systems (C(X), f̄) and (C(Y ), ḡ) are topolog- ically conjugate, the induced map ϕ̄ being the topological conjugacy. Since ϕ̄ maps F1(X) homeomorphically onto F1(Y ), HS(ϕ̄) : HS(X) → HS(Y ) is the required topological conjugacy between the dynamical systems (HS(X),HS(f)) and (HS(Y ),HS(g)). � Acknowledgements. The authors are grateful to their supervisor Professor K. K. Azad for his valuable guidance and encouragement. The first author thanks the Department of Science and Technology, New Delhi, for providing financial support. c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 2 180 Lifting dynamical properties to hyperspaces and hyperspace suspension References [1] J. Banks, Chaos for induced hyperspace maps, Chaos Solitons Fractals 25 (2005), 681– 685. [2] J. J. Charatonik, Recent Research in Hyperspace Theory, Extracta Math. 18 (2003), 235–262. [3] R. L. Devaney, An introduction to chaotic dynamical systems, Redwood City, Addison- Wesley, 1989. [4] J. Dugundji, Topology, Allyn and Bacon, Boston, 1966. [5] R. Escobedo, M. López and S. Macias, On the hyperspace suspension of a continuum, Topology Appl. 138 (2004), 109–124. [6] N. Esty, The hyperspaces Cn(X) for finite ray-graphs, arXiv: 1012. 1578 (2010). [7] J. L. G. Guirao, D. Kwietniak, M. Lampart, P. Oprocha and A. Peris, Chaos on hyper- spaces, Nonlinear Anal. 71 (2009), 1–8. [8] S. Kolyada and L’. Snoha, Some aspects of topological transitivity - a survey, Grazer Math. Ber. 334 (1997), 3–35. [9] D. Kwietniak and P. Oprocha, Topological entropy and chaos for maps induced on hyperspaces, Chaos Solitons Fractals 33 (2007), 76–86. [10] S. B. Nadler, Hyperspaces of sets, M. Dekker, New York and Basel, 1978. [11] H. Roman-Flores, A note on transitivity in set-valued discrete systems, Chaos Solitons Fractals 17 (2003), 99–104. [12] S. Ruette, Chaos for continuous interval maps, Université Paris-Sud, 91405 Orsay, Cedex - France (2003). [13] P. Sharma and A. Nagar, Topological dynamics on hyperspaces, Appl. Gen. Topol. 11 (2010), 1–19. c© AGT, UPV, 2014 Appl. Gen. Topol. 15, no. 2 181