OAI/Analysis/Nonexpansive/PointwiseWeakApproximation.lean
6 declarations, 0 contain sorry, 0 declared as axioms
Simultaneous approximation in the pointwise weak topology
A convex family of weak-valued maps can approximate each point in its closure in norm at any prescribed finite family of arguments. The finite family is handled as one vector in the finite product normed space. Consequently the chosen approximating map is the same at every argument.
No completeness, reflexivity, or nontriviality assumption is required here.
- def
OAI.ReflexiveFixedPoints.finiteNormEvaluationline 34Source documentation
The underlying normed tuple at a prescribed family of arguments. This is a linear map, without a claim of weak-to-norm continuity.
def finiteNormEvaluation (a : ι → α) : (α → WeakSpace ℝ X) →ₗ[ℝ] (ι → X) := … - theorem
OAI.ReflexiveFixedPoints.finiteNormEvaluation_applyline 39 - theorem
OAI.ReflexiveFixedPoints.continuous_finite_weak_evaluationline 47Source documentation
Finite evaluation is continuous when its target has the weak topology of the finite product normed space.
theorem continuous_finite_weak_evaluation (a : ι → α) : Continuous (fun f : α → WeakSpace ℝ X => toWeakSpace ℝ (ι → X) (finiteNormEvaluation a f)) := by … - theorem
OAI.ReflexiveFixedPoints.exists_mem_norm_approx_of_mem_pointwiseWeak_closureline 72Source documentation
One member of a convex family approximates a point of its pointwise weak closure simultaneously in norm at every argument in a finite indexed family. The index type is allowed to be empty, and arguments may repeat.
theorem exists_mem_norm_approx_of_mem_pointwiseWeak_closure {s : Set (α → WeakSpace ℝ X)} (hs : Convex ℝ s) {f : α → WeakSpace ℝ X} (hf : f ∈ closure s) (a : ι → α) {ε : ℝ} (hε : 0 < ε) : ∃ g ∈ s, ∀ i, ‖(toWeakSpace ℝ X).symm (g (a i)) - (toWeakSpace ℝ X).symm (f (a i))‖ < ε := by … - theorem
OAI.ReflexiveFixedPoints.exists_mem_norm_approx_finset_of_mem_pointwiseWeak_closureline 99Source documentation
Finset form of simultaneous norm approximation from pointwise weak closure. In particular, separate choices at different arguments are not used.
theorem exists_mem_norm_approx_finset_of_mem_pointwiseWeak_closure {s : Set (α → WeakSpace ℝ X)} (hs : Convex ℝ s) {f : α → WeakSpace ℝ X} (hf : f ∈ closure s) (t : Finset α) {ε : ℝ} (hε : 0 < ε) : ∃ g ∈ s, ∀ x ∈ t, ‖(toWeakSpace ℝ X).symm (g x) - (toWeakSpace ℝ X).symm (f x)‖ < ε := by … - theorem
OAI.ReflexiveFixedPoints.mem_closure_of_finite_dist_approxline 115Source documentation
Simultaneous approximation on every finite set of arguments implies membership in the product-topology closure. This uses only the codomain's pseudometric, so it applies directly to maps taking values in a norm subtype.
theorem mem_closure_of_finite_dist_approx {Y : Type u} [PseudoMetricSpace Y] {B : Set (α → Y)} {f : α → Y} (h : ∀ (s : Finset α) (ε : ℝ), 0 < ε → ∃ g ∈ B, ∀ a ∈ s, dist (g a) (f a) < ε) : f ∈ closure B := by …