The structure of hyperbolic attractors on surfaces
Todd Fisher [email protected]
Department of Mathematics University of Maryland, College Park
The structure of hyperbolic attractors on surfaces – p. 1/21 Examples Previous Results Outline of Argument and open questions
Outline of Talk
Background
The structure of hyperbolic attractors on surfaces – p. 2/21 Previous Results Outline of Argument and open questions
Outline of Talk
Background Examples
The structure of hyperbolic attractors on surfaces – p. 2/21 Outline of Argument and open questions
Outline of Talk
Background Examples Previous Results
The structure of hyperbolic attractors on surfaces – p. 2/21 Outline of Talk
Background Examples Previous Results Outline of Argument and open questions
The structure of hyperbolic attractors on surfaces – p. 2/21 Standing Assumptions
M is a compact smooth boundaryless surface. f ∈ Di (M)
The structure of hyperbolic attractors on surfaces – p. 3/21 If M is a compact manifold and f a diffeomorphism of M, then equivalently we say that given two open sets U and V in X ∃ n ∈ N such that f n(U) ∩ V =6 ∅. Definition: A set X is mixing (topologically) if for any open sets U and V in X ∃ N ∈ N such that ∀ n N f n(U) ∩ V =6 ∅.
Transitivity and Mixing
Definition: A set X is transitive for a map f if there is a point x ∈ X with a dense forward orbit
The structure of hyperbolic attractors on surfaces – p. 4/21 Definition: A set X is mixing (topologically) if for any open sets U and V in X ∃ N ∈ N such that ∀ n N f n(U) ∩ V =6 ∅.
Transitivity and Mixing
Definition: A set X is transitive for a map f if there is a point x ∈ X with a dense forward orbit If M is a compact manifold and f a diffeomorphism of M, then equivalently we say that given two open sets U and V in X ∃ n ∈ N such that f n(U) ∩ V =6 ∅.
The structure of hyperbolic attractors on surfaces – p. 4/21 Transitivity and Mixing
Definition: A set X is transitive for a map f if there is a point x ∈ X with a dense forward orbit If M is a compact manifold and f a diffeomorphism of M, then equivalently we say that given two open sets U and V in X ∃ n ∈ N such that f n(U) ∩ V =6 ∅. Definition: A set X is mixing (topologically) if for any open sets U and V in X ∃ N ∈ N such that ∀ n N f n(U) ∩ V =6 ∅.
The structure of hyperbolic attractors on surfaces – p. 4/21 Definition: A set X is an attractor for a map f if ∃ neighborhood U (an attracting set) of X such n that X = n∈N f (U) and f(U) U. Definition: A set is a hyperbolic attractor if is transitive Tand has an attracting set U.
Hyperbolic Attractors
Definition: A compact invariant set for f ∈ Di (M) is hyperbolic if the tangent space has a continuous invariant splitting s u s T M = E E where E is uniformly contracting and Eu is uniformly expanding.
The structure of hyperbolic attractors on surfaces – p. 5/21 Definition: A set is a hyperbolic attractor if is transitive and has an attracting set U.
Hyperbolic Attractors
Definition: A compact invariant set for f ∈ Di (M) is hyperbolic if the tangent space has a continuous invariant splitting s u s T M = E E where E is uniformly contracting and Eu is uniformly expanding. Definition: A set X is an attractor for a map f if ∃ neighborhood U (an attracting set) of X such n that X = n∈N f (U) and f(U) U. T
The structure of hyperbolic attractors on surfaces – p. 5/21 Hyperbolic Attractors
Definition: A compact invariant set for f ∈ Di (M) is hyperbolic if the tangent space has a continuous invariant splitting s u s T M = E E where E is uniformly contracting and Eu is uniformly expanding. Definition: A set X is an attractor for a map f if ∃ neighborhood U (an attracting set) of X such n that X = n∈N f (U) and f(U) U. Definition: A set is a hyperbolic attractor if is transitive Tand has an attracting set U.
The structure of hyperbolic attractors on surfaces – p. 5/21 If is mixing we know W u(x) = and W s(x) is dense in W s( ) ∀ x ∈ . k = i=1 i where i’s are compact disjoint, k f( i) = i+1, and f ( i) is a mixing hyperbolic attractorS .
Properties of Hyperbolic Attractors
For each x ∈ we know W u(x)
The structure of hyperbolic attractors on surfaces – p. 6/21 k = i=1 i where i’s are compact disjoint, k f( i) = i+1, and f ( i) is a mixing hyperbolic attractorS .
Properties of Hyperbolic Attractors
For each x ∈ we know W u(x) If is mixing we know W u(x) = and W s(x) is dense in W s( ) ∀ x ∈ .
The structure of hyperbolic attractors on surfaces – p. 6/21 Properties of Hyperbolic Attractors
For each x ∈ we know W u(x) If is mixing we know W u(x) = and W s(x) is dense in W s( ) ∀ x ∈ . k = i=1 i where i’s are compact disjoint, k f( i) = i+1, and f ( i) is a mixing hyperbolic attractorS .
The structure of hyperbolic attractors on surfaces – p. 6/21 int(Ri) ∩ int(Rj) = ∅ if i =6 j for some sufficiently small Ri is u s (W (x) ∩ Ri) (W (x) ∩ Ri) x ∈ Ri, f(x) ∈ Rj, and i → j is an allowed transition, then s 1 u f(W (x, Ri)) Rj and f (W (f(x), Rj)) Ri.
Markov Partition
Decomposition of into finite number of dynamical rectangles R1, ..., Rn such that for each 1 i n
The structure of hyperbolic attractors on surfaces – p. 7/21 for some sufficiently small Ri is u s (W (x) ∩ Ri) (W (x) ∩ Ri) x ∈ Ri, f(x) ∈ Rj, and i → j is an allowed transition, then s 1 u f(W (x, Ri)) Rj and f (W (f(x), Rj)) Ri.
Markov Partition
Decomposition of into finite number of dynamical rectangles R1, ..., Rn such that for each 1 i n int(Ri) ∩ int(Rj) = ∅ if i =6 j
The structure of hyperbolic attractors on surfaces – p. 7/21 x ∈ Ri, f(x) ∈ Rj, and i → j is an allowed transition, then s 1 u f(W (x, Ri)) Rj and f (W (f(x), Rj)) Ri.
Markov Partition
Decomposition of into finite number of dynamical rectangles R1, ..., Rn such that for each 1 i n int(Ri) ∩ int(Rj) = ∅ if i =6 j
for some sufficiently small Ri is u s (W (x) ∩ Ri) (W (x) ∩ Ri)
The structure of hyperbolic attractors on surfaces – p. 7/21 Markov Partition
Decomposition of into finite number of dynamical rectangles R1, ..., Rn such that for each 1 i n int(Ri) ∩ int(Rj) = ∅ if i =6 j for some sufficiently small Ri is u s (W (x) ∩ Ri) (W (x) ∩ Ri) x ∈ Ri, f(x) ∈ Rj, and i → j is an allowed transition, then
s 1 u f(W (x, Ri)) Rj and f (W (f(x), Rj)) Ri.
The structure of hyperbolic attractors on surfaces – p. 7/21 In general these address the question where we assume we have an attractor, then what can be said about the set and when are two sets homeomorphic. Question: Suppose we know the topology of and is hyperbolic, what can be concluded about the set?
General Question
The structure of hyperbolic attractors on surfaces has been studied extensively by Plykin, Bonatti, Williams, Zhirov, Grines, F. and J. Rodriquez-Hertz, and others.
The structure of hyperbolic attractors on surfaces – p. 8/21 Question: Suppose we know the topology of and is hyperbolic, what can be concluded about the set?
General Question
The structure of hyperbolic attractors on surfaces has been studied extensively by Plykin, Bonatti, Williams, Zhirov, Grines, F. and J. Rodriquez-Hertz, and others. In general these address the question where we assume we have an attractor, then what can be said about the set and when are two sets homeomorphic.
The structure of hyperbolic attractors on surfaces – p. 8/21 General Question
The structure of hyperbolic attractors on surfaces has been studied extensively by Plykin, Bonatti, Williams, Zhirov, Grines, F. and J. Rodriquez-Hertz, and others. In general these address the question where we assume we have an attractor, then what can be said about the set and when are two sets homeomorphic. Question: Suppose we know the topology of and is hyperbolic, what can be concluded about the set?
The structure of hyperbolic attractors on surfaces – p. 8/21 A nontrivial attractor means not the orbit of a periodic sink Counterexamples in higher dimensions
Statement of Main Result
Theorem 1:(F.) If M is a compact surface and is a nontrivial mixing hyperbolic attractor for a diffeomorphism f of M, and is hyperbolic for a diffeomorphism g of M, then is either a nontrivial mixing hyperbolic attractor or a nontrivial mixing hyperbolic repeller for g.
The structure of hyperbolic attractors on surfaces – p. 9/21 Counterexamples in higher dimensions
Statement of Main Result
Theorem 1:(F.) If M is a compact surface and is a nontrivial mixing hyperbolic attractor for a diffeomorphism f of M, and is hyperbolic for a diffeomorphism g of M, then is either a nontrivial mixing hyperbolic attractor or a nontrivial mixing hyperbolic repeller for g. A nontrivial attractor means not the orbit of a periodic sink
The structure of hyperbolic attractors on surfaces – p. 9/21 Statement of Main Result
Theorem 1:(F.) If M is a compact surface and is a nontrivial mixing hyperbolic attractor for a diffeomorphism f of M, and is hyperbolic for a diffeomorphism g of M, then is either a nontrivial mixing hyperbolic attractor or a nontrivial mixing hyperbolic repeller for g. A nontrivial attractor means not the orbit of a periodic sink Counterexamples in higher dimensions
The structure of hyperbolic attractors on surfaces – p. 9/21 Non-mixing Case
Theorem 2:(F.) If M is a compact surface and is a nontrivial hyperbolic attractor for a diffeomorphism f of M, and is hyperbolic for a diffeomorphism g of M, then there exists an N N n ∈ and sets 1, ..., N where = i=1 i, i ∩ j = ∅ if i =6 j, and each i is a mixing hyperbolic attractor or repeller for gn.S
The structure of hyperbolic attractors on surfaces – p. 10/21 Commuting Diffeo. r 2
Theorem 3:(F.) Let M be a compact surface and is a nontrivial hyperbolic attractor for f ∈ Di r(M), r 2. Then there exists a neighborhood U of f in Di r(M) and an open and dense set U 0 U such that for all f 0 ∈ U if g ∈ Di 1(M) where f 0g = gf 0 (g in the centralizer 0 j of f), then g|W s( ) = (f ) for some j ∈ Z. This is an extension of a result from the work of Palis and Yoccoz.
The structure of hyperbolic attractors on surfaces – p. 11/21 p 1
V p p 2 n T2 Let = n∈N f ( V ), then is a hyperbolic attractor. T
DA example
Let f : T2 → T2 Anosov map from map 2 1 A = , p a hyperbolic fixed point of f. "1 1#
The structure of hyperbolic attractors on surfaces – p. 12/21 p 1
V p p 2 n T2 Let = n∈N f ( V ), then is a hyperbolic attractor. T
DA example
Let f : T2 → T2 Anosov map from map 2 1 A = , p a hyperbolic fixed point of f. "1 1#
p
The structure of hyperbolic attractors on surfaces – p. 12/21 n T2 Let = n∈N f ( V ), then is a hyperbolic attractor. T
DA example
Let f : T2 → T2 Anosov map from map 2 1 A = , p a hyperbolic fixed point of f. "1 1#
p 1
V p
p 2
The structure of hyperbolic attractors on surfaces – p. 12/21 DA example
Let f : T2 → T2 Anosov map from map 2 1 A = , p a hyperbolic fixed point of f. "1 1#
p 1
V p
p 2 n T2 Let = n∈N f ( V ), then is a hyperbolic attractor. T
The structure of hyperbolic attractors on surfaces – p. 12/21 Plykin Attractor
Another example due to Plykin can be built on the disk, so embedded into any surface. Note: we need three holes in our domain.
The structure of hyperbolic attractors on surfaces – p. 13/21 Plykin Attractor
Another example due to Plykin can be built on the disk, so embedded into any surface. Note: we need three holes in our domain. V
The structure of hyperbolic attractors on surfaces – p. 13/21 Plykin Attractor
Another example due to Plykin can be built on the disk, so embedded into any surface. Note: we need three holes in our domain. V
f(V)
The structure of hyperbolic attractors on surfaces – p. 13/21 Plykin: Shows trapping region is homeomorphic to disk with holes, then must have at least 3 holes.
Results of Williams and Plykin
Williams: Shows that if x is in an attractor on a surface, then a neighborhood of x in the attractor is [0, 1] C for a Cantor set C. Also, uses symbolic dynamics to classify when two diffeomorphisms restricted to two attractors are conjugate.
The structure of hyperbolic attractors on surfaces – p. 14/21 Results of Williams and Plykin
Williams: Shows that if x is in an attractor on a surface, then a neighborhood of x in the attractor is [0, 1] C for a Cantor set C. Also, uses symbolic dynamics to classify when two diffeomorphisms restricted to two attractors are conjugate. Plykin: Shows trapping region is homeomorphic to disk with holes, then must have at least 3 holes.
The structure of hyperbolic attractors on surfaces – p. 14/21 F. and J. Rodriquez-Hertz: Obtain a local dynamical and topological picture of expansive attractors on surfaces.
Zhirov and F. and J. Rodriquez-Hertz
Zhirov: Classifies when 2 diffeomorphisms are conjugate in neighborhood of attractor. Gives algorithm for the classification.
The structure of hyperbolic attractors on surfaces – p. 15/21 Zhirov and F. and J. Rodriquez-Hertz
Zhirov: Classifies when 2 diffeomorphisms are conjugate in neighborhood of attractor. Gives algorithm for the classification. F. and J. Rodriquez-Hertz: Obtain a local dynamical and topological picture of expansive attractors on surfaces.
The structure of hyperbolic attractors on surfaces – p. 15/21 W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f) for some x ∈
Outline of argument - 2 cases
W u(x, g) = W u(x, f) or W s(x, g) = W u(x, f) for all x ∈
The structure of hyperbolic attractors on surfaces – p. 16/21 Outline of argument - 2 cases
W u(x, g) = W u(x, f) or W s(x, g) = W u(x, f) for all x ∈ W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f) for some x ∈
The structure of hyperbolic attractors on surfaces – p. 16/21 Since cl(W u(x, f)) = then can show mixing hyperbolic attractor.
Case 1
Assume W u(x, f) = W u(x, g) We show is attractor: Take > 0 sufficiently small and s V = x∈ W (x) is attracting set. i.e. - = gn(V ). Tn 0 T
The structure of hyperbolic attractors on surfaces – p. 17/21 Case 1
Assume W u(x, f) = W u(x, g) We show is attractor: Take > 0 sufficiently small and s V = x∈ W (x) is attracting set. i.e. - = gn(V ). Tn 0 SinceTcl(W u(x, f)) = then can show mixing hyperbolic attractor.
The structure of hyperbolic attractors on surfaces – p. 17/21 ˜ s u For x, y ∈ in same rectangle W (x) ∩ W (y) is one point in ˜. So product structure for points in same rectangle.
Markov Partition Theorem
Theorem 4:(F.) If is a hyperbolic set and V is a neighborhood of , then there exists a hyperbolic set ˜ with a Markov partition such that ˜ V .
The structure of hyperbolic attractors on surfaces – p. 18/21 Markov Partition Theorem
Theorem 4:(F.) If is a hyperbolic set and V is a neighborhood of , then there exists a hyperbolic set ˜ with a Markov partition such that ˜ V . ˜ s u For x, y ∈ in same rectangle W (x) ∩ W (y) is one point in ˜. So product structure for points in same rectangle.
The structure of hyperbolic attractors on surfaces – p. 18/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f).
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
x W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
y
x W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
y
x x' W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
x W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
y
x W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
x W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Case 2
Take U neighborhood of and ˜ in U containing with Markov partition. Show there exists x ∈ such that x in interior of rectangle and W u(x, g) =6 W u(x, f) or W s(x, g) =6 W u(x, f). W s (x,g)
x W u (x,g)
W u (x,f)
The structure of hyperbolic attractors on surfaces – p. 19/21 Since U arbitrarily small we show this implies has attractor and repeller
Using density of W u(x, f) in we see contradiction.
Contradiction
So ˜ has interior. Then contains hyperbolic attractor and repeller.
The structure of hyperbolic attractors on surfaces – p. 20/21 Using density of W u(x, f) in we see contradiction.
Contradiction
So ˜ has interior. Then contains hyperbolic attractor and repeller.
Since U arbitrarily small we show this implies has attractor and repeller
The structure of hyperbolic attractors on surfaces – p. 20/21 Contradiction
So ˜ has interior. Then contains hyperbolic attractor and repeller.
Since U arbitrarily small we show this implies has attractor and repeller
Using density of W u(x, f) in we see contradiction.
The structure of hyperbolic attractors on surfaces – p. 20/21 If a hyperbolic set is locally maximal (isolated) for f and hyperbolic for g is it necessarily locally maximal for g?
Open Questions
Does the main Theorem hold for codimension one attractors?
The structure of hyperbolic attractors on surfaces – p. 21/21 Open Questions
Does the main Theorem hold for codimension one attractors? If a hyperbolic set is locally maximal (isolated) for f and hyperbolic for g is it necessarily locally maximal for g?
The structure of hyperbolic attractors on surfaces – p. 21/21