Attractors and basins of maps and flows
Definitionpalis_attractorLet be a topological space modelled on .
Discrete time. For a map and , the -limit set is the set of accumulation points of , namely . The basin of is . The set is transitive if there is with , and is an attractor of if
- (invariance),
- is transitive,
- is not Lebesgue-null.
has finitely many attractors covering almost all of if there are finitely many attractors of such that is Lebesgue-null.
Continuous time. For a flow the same notions are defined with the set of accumulation points of as , invariance meaning for every , and transitivity meaning for some .
These are exactly the notions in the Definition of the source: an attractor is an invariant, transitive set whose basin has positive Lebesgue measure.
Formalization Note -limits are Mathlib's omegaLimit atTop. Lebesgue-nullity is PalisConjecture.IsLebesgueNull. The finite family of attractors is a Finset (Set M).
import Mathlib
import Definitions.Def_palis_lebesgue_null
/-!
# Attractors and their basins (discrete and continuous time)
Following the definition in the source: a set `Λ ⊆ M` is an *attractor* if it is invariant,
transitive, and its basin `B(Λ) = {p ∈ M | ω(p) ⊆ Λ}` has positive Lebesgue measure.
-/
namespace PalisConjecture
open Filter
variable {M : Type*} [TopologicalSpace M]
/-! ### Discrete time: a map `T : M → M` -/
/-- The ω-limit set `ω(p)` of `p` under iteration of `T`. -/
def omegaLimitMap (T : M → M) (p : M) : Set M :=
omegaLimit atTop (fun (k : ℕ) (x : M) => T^[k] x) {p}
/-- The basin of attraction `B(Λ) = {p ∈ M | ω(p) ⊆ Λ}`. -/
def basin (T : M → M) (Λ : Set M) : Set M :=
{p | omegaLimitMap T p ⊆ Λ}
/-- `Λ` is transitive: some point `x ∈ Λ` has forward orbit `{T^k x | k ≥ 0}` dense in `Λ`
(the closure of the orbit equals `Λ`). -/
def IsTransitive (T : M → M) (Λ : Set M) : Prop :=
∃ x ∈ Λ, closure {y | ∃ k : ℕ, T^[k] x = y} = Λ
/-- `Λ` is an attractor of `T`: invariant (`T(Λ) = Λ`), transitive, and its basin is not
Lebesgue-null. -/
def IsAttractor (n : ℕ) {M : Type*} [TopologicalSpace M]
[ChartedSpace (EuclideanSpace ℝ (Fin n)) M] (T : M → M) (Λ : Set M) : Prop :=
T '' Λ = Λ ∧ IsTransitive T Λ ∧ ¬ IsLebesgueNull n (basin T Λ)
/-- `T` exhibits finitely many attractors whose basins cover Lebesgue-almost all of `M`. -/
def HasFiniteAttractorsCoveringAE (n : ℕ) {M : Type*} [TopologicalSpace M]
[ChartedSpace (EuclideanSpace ℝ (Fin n)) M] (T : M → M) : Prop :=
∃ A : Finset (Set M), (∀ Λ ∈ A, IsAttractor n T Λ) ∧
IsLebesgueNull n (⋃ Λ ∈ A, basin T Λ)ᶜ
/-! ### Continuous time: a flow `Φ : ℝ → M → M` -/
/-- The ω-limit set `ω(p)` of `p` under the flow `Φ` as `t → +∞`. -/
def omegaLimitFlow (Φ : ℝ → M → M) (p : M) : Set M :=
omegaLimit atTop Φ {p}
/-- The basin `B(Λ) = {p ∈ M | ω(p) ⊆ Λ}` for the flow `Φ`. -/
def flowBasin (Φ : ℝ → M → M) (Λ : Set M) : Set M :=
{p | omegaLimitFlow Φ p ⊆ Λ}
/-- `Λ` is transitive for the flow: some `x ∈ Λ` has forward orbit `{Φ t x | t ≥ 0}` dense
in `Λ`. -/
def IsFlowTransitive (Φ : ℝ → M → M) (Λ : Set M) : Prop :=
∃ x ∈ Λ, closure {y | ∃ t : ℝ, 0 ≤ t ∧ Φ t x = y} = Λ
/-- `Λ` is an attractor of the flow `Φ`: invariant (`Φ_t(Λ) = Λ` for all `t`), transitive,
and its basin is not Lebesgue-null. -/
def IsFlowAttractor (n : ℕ) {M : Type*} [TopologicalSpace M]
[ChartedSpace (EuclideanSpace ℝ (Fin n)) M] (Φ : ℝ → M → M) (Λ : Set M) : Prop :=
(∀ t : ℝ, Φ t '' Λ = Λ) ∧ IsFlowTransitive Φ Λ ∧ ¬ IsLebesgueNull n (flowBasin Φ Λ)
/-- The flow `Φ` exhibits finitely many attractors whose basins cover Lebesgue-almost
all of `M`. -/
def FlowHasFiniteAttractorsCoveringAE (n : ℕ) {M : Type*} [TopologicalSpace M]
[ChartedSpace (EuclideanSpace ℝ (Fin n)) M] (Φ : ℝ → M → M) : Prop :=
∃ A : Finset (Set M), (∀ Λ ∈ A, IsFlowAttractor n Φ Λ) ∧
IsLebesgueNull n (⋃ Λ ∈ A, flowBasin Φ Λ)ᶜ
end PalisConjecture
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.