import Mathlib.GroupTheory.Subgroup.Basic import Mathlib.GroupTheory.GroupAction.Basic import Mathlib.Topology.Basic import Mathlib.Topology.Homeomorph import Mathlib.Data.Set.Basic import Rubin.MulActionExt namespace Rubin /-- Specificies that a group action is continuous, that is, for every group element `g`, `x ↦ g • x` is continuous. Note that this is a weaker statement than `ContinuousSMul`, as the group `G` is not required to be a topology. --/ class ContinuousMulAction (G α : Type _) [Group G] [TopologicalSpace α] [MulAction G α] where continuous : ∀ g : G, Continuous (fun x: α => g • x) #align continuous_mul_action Rubin.ContinuousMulAction def ContinuousMulAction.toHomeomorph {G : Type _} (α : Type _) [Group G] [TopologicalSpace α] [MulAction G α] [hc : ContinuousMulAction G α] (g : G) : Homeomorph α α where toFun := fun x => g • x invFun := fun x => g⁻¹ • x left_inv := by intro y simp right_inv := by intro y simp continuous_toFun := by simp exact hc.continuous _ continuous_invFun := by simp exact hc.continuous _ theorem ContinuousMulAction.toHomeomorph_toFun {G : Type _} (α : Type _) [Group G] [TopologicalSpace α] [MulAction G α] [ContinuousMulAction G α] (g : G) : (ContinuousMulAction.toHomeomorph α g).toFun = fun x => g • x := rfl theorem ContinuousMulAction.toHomeomorph_invFun {G : Type _} (α : Type _) [Group G] [TopologicalSpace α] [MulAction G α] [ContinuousMulAction G α] (g : G) : (ContinuousMulAction.toHomeomorph α g).invFun = fun x => g⁻¹ • x := rfl -- TODO: give this a notation? -- TODO: coe to / extend MulActionHom structure EquivariantHomeomorph (G α β : Type _) [Group G] [TopologicalSpace α] [TopologicalSpace β] [MulAction G α] [MulAction G β] extends Homeomorph α β where equivariant : is_equivariant G toFun #align equivariant_homeomorph Rubin.EquivariantHomeomorph variable {G α β : Type _} variable [Group G] variable [TopologicalSpace α] [TopologicalSpace β] theorem equivariant_fun [MulAction G α] [MulAction G β] (h : EquivariantHomeomorph G α β) : is_equivariant G h.toFun := h.equivariant #align equivariant_fun Rubin.equivariant_fun theorem equivariant_inv [MulAction G α] [MulAction G β] (h : EquivariantHomeomorph G α β) : is_equivariant G h.invFun := by intro g x symm let e := congr_arg h.invFun (h.equivariant g (h.invFun x)) rw [h.left_inv _, h.right_inv _] at e exact e #align equivariant_inv Rubin.equivariant_inv open Topology -- Note: this sounds like a general enough theorem that it should already be in mathlib lemma inter_of_open_subset_of_closure {α : Type _} [TopologicalSpace α] {U V : Set α} (U_open : IsOpen U) (U_nonempty : Set.Nonempty U) (V_nonempty : Set.Nonempty V) (U_ss_clV : U ⊆ closure V) : Set.Nonempty (U ∩ V) := by by_contra empty rw [Set.not_nonempty_iff_eq_empty] at empty rw [Set.nonempty_iff_ne_empty] at U_nonempty apply U_nonempty have clV_diff_U_ss_V : V ⊆ closure V \ U := by rw [Set.subset_diff] constructor exact subset_closure symm rw [Set.disjoint_iff_inter_eq_empty] exact empty have clV_diff_U_closed : IsClosed (closure V \ U) := by apply IsClosed.sdiff exact isClosed_closure assumption unfold closure at U_ss_clV simp at U_ss_clV specialize U_ss_clV (closure V \ U) clV_diff_U_closed clV_diff_U_ss_V rw [Set.subset_diff] at U_ss_clV rw [Set.disjoint_iff_inter_eq_empty] at U_ss_clV simp at U_ss_clV exact U_ss_clV.right /-- Note: `𝓝[≠] x` is notation for `nhdsWithin x {[x]}ᶜ`, ie. the neighborhood of x not containing itself. --/ class HasNoIsolatedPoints (α : Type _) [TopologicalSpace α] := -- TODO: rename to nhdsWithin_ne_bot nhbd_ne_bot : ∀ x : α, 𝓝[≠] x ≠ ⊥ #align has_no_isolated_points Rubin.HasNoIsolatedPoints instance has_no_isolated_points_neBot₁ {α : Type _} [TopologicalSpace α] [h_nip: HasNoIsolatedPoints α] (x: α) : Filter.NeBot (𝓝[≠] x) where ne' := h_nip.nhbd_ne_bot x theorem Filter.NeBot.choose {α : Type _} (F : Filter α) [Filter.NeBot F] : ∃ S : Set α, S ∈ F := by have res := (Filter.inhabitedMem (α := α) (f := F)).default exact ⟨res.val, res.prop⟩ theorem TopologicalSpace.IsTopologicalBasis.contains_point {α : Type _} [TopologicalSpace α] {B : Set (Set α)} (B_basis : TopologicalSpace.IsTopologicalBasis B) (p : α) : ∃ S : Set α, S ∈ B ∧ p ∈ S := by have nhds_basis := B_basis.nhds_hasBasis (a := p) rw [Filter.hasBasis_iff] at nhds_basis let ⟨S₁, S₁_in_nhds⟩ := Filter.NeBot.choose (𝓝 p) let ⟨S, ⟨⟨S_in_B, p_in_S⟩, _⟩⟩ := (nhds_basis S₁).mp S₁_in_nhds exact ⟨S, S_in_B, p_in_S⟩ -- The collection of all the sets in `B` (a topological basis of `α`), such that `p` is in them. def TopologicalBasisContaining {α : Type _} [TopologicalSpace α] {B : Set (Set α)} (B_basis : TopologicalSpace.IsTopologicalBasis B) (p : α) : FilterBasis α where sets := {b ∈ B | p ∈ b} nonempty := by let ⟨S, S_in_B, p_in_S⟩ := TopologicalSpace.IsTopologicalBasis.contains_point B_basis p use S simp tauto inter_sets := by intro S T ⟨S_in_B, p_in_S⟩ ⟨T_in_B, p_in_T⟩ have S_in_nhds := B_basis.mem_nhds_iff.mpr ⟨S, S_in_B, ⟨p_in_S, Eq.subset rfl⟩⟩ have T_in_nhds := B_basis.mem_nhds_iff.mpr ⟨T, T_in_B, ⟨p_in_T, Eq.subset rfl⟩⟩ have ST_in_nhds : S ∩ T ∈ 𝓝 p := Filter.inter_mem S_in_nhds T_in_nhds rw [B_basis.mem_nhds_iff] at ST_in_nhds let ⟨U, props⟩ := ST_in_nhds use U simp simp at props tauto theorem TopologicalBasisContaining.mem_iff {α : Type _} [TopologicalSpace α] {B : Set (Set α)} (B_basis : TopologicalSpace.IsTopologicalBasis B) (p : α) (S : Set α) : S ∈ TopologicalBasisContaining B_basis p ↔ S ∈ B ∧ p ∈ S := by rw [<-FilterBasis.mem_sets] rfl theorem TopologicalBasisContaining.mem_nhds {α : Type _} [TopologicalSpace α] {B : Set (Set α)} (B_basis : TopologicalSpace.IsTopologicalBasis B) (p : α) (S : Set α) : S ∈ TopologicalBasisContaining B_basis p → S ∈ 𝓝 p := by rw [TopologicalBasisContaining.mem_iff] rw [B_basis.mem_nhds_iff] intro ⟨S_in_B, p_in_S⟩ use S instance TopologicalBasisContaining.neBot {α : Type _} [TopologicalSpace α] {B : Set (Set α)} (B_basis : TopologicalSpace.IsTopologicalBasis B) (p : α) : Filter.NeBot (TopologicalBasisContaining B_basis p).filter where ne' := by intro empty_in rw [<-Filter.empty_mem_iff_bot, FilterBasis.mem_filter_iff] at empty_in let ⟨S, ⟨S_in_basis, S_ss_empty⟩⟩ := empty_in rw [TopologicalBasisContaining.mem_iff] at S_in_basis exact S_ss_empty S_in_basis.right -- Note: the definition of "convergence" in the article doesn't quite match with the definition of ClusterPt -- Instead, `F ≤ nhds p` should be used. -- Note: Filter.HasBasis is a stronger statement than just FilterBasis - it defines a two-way relationship between a filter and a property; if the property is true for a set, then any superset of it is part of the filter, and vice-versa. -- With this, it's impossible for there to be a finer filter satisfying the property, -- as is evidenced by `filter_eq`: stripping away the `Filter` allows us to uniquely reconstruct it from the property itself. -- Proposition 3.3.1 trivially follows from `TopologicalSpace.IsTopologicalBasis.nhds_hasBasis` and `disjoint_nhds_nhds`: if `F.HasBasis (S → S ∈ B ∧ p ∈ S)` and `F.HasBasis (S → S ∈ B ∧ q ∈ S)`, -- then one can prove that `F ≤ nhds x` and `F ≤ nhds y` ~> `F = ⊥` -- Proposition 3.3.2 becomes simply `TopologicalSpace.IsTopologicalBasis.nhds_hasBasis` -- Proposition 3.3.3 is a consequence of the structure of `HasBasis` -- Proposition 3.4.1 can maybe be proven with `TopologicalSpace.IsTopologicalBasis.mem_closure_iff`? -- The tricky part here though is that "F is an ultra(pre)filter on B" can't easily be expressed. -- I should maybe define a Prop for it, and show that "F is an ultrafilter on B" + "F tends to a point p" -- is equivalent to `TopologicalSpace.IsTopologicalBasis.nhds_hasBasis`. -- The alternative is to only work with `Filter`, and state conditions with `Filter.HasBasis`, -- since that will force the filter to be an ultraprefilter on B. end Rubin