The spaces , , and their unit balls
DefinitionPlasticity_SeqSpaceFor an arbitrary index set , is the space of bounded real functions on with the supremum norm . Inside it, is the subspace of functions that vanish at infinity (for every only finitely many have ), and the subspace of finitely supported functions. A space between and is a linear subspace with ; its closed unit ball is . A subspace of realizes all sign patterns if for every countable and every choice of signs , , some has for all .
import Mathlib.Analysis.Normed.Lp.lpSpace
import Mathlib.Order.Filter.Cofinite
/-!
# Plasticity: sequence spaces over an arbitrary index set
`Linf Γ` is ℓ∞(Γ), the real bounded functions on `Γ` with the supremum norm (Mathlib's
`lp (fun _ => ℝ) ∞`). Inside it, `c0 Γ` is the subspace of functions vanishing at
infinity (tending to `0` along the cofinite filter) and `c00 Γ` the subspace of finitely
supported functions. A space "between c₀₀(Γ) and c₀(Γ)" is a submodule `X` with
`c00 Γ ≤ X ≤ c0 Γ`; its closed unit ball inside `Linf Γ` is `ballIn X`.
-/
namespace Plasticity
open Filter Topology
/-- ℓ∞(Γ): bounded real functions on `Γ` with the supremum norm. -/
abbrev Linf (Γ : Type*) : Type _ := lp (fun _ : Γ => ℝ) ⊤
/-- c₀(Γ): the functions in ℓ∞(Γ) that tend to `0` along the cofinite filter. -/
def c0 (Γ : Type*) : Submodule ℝ (Linf Γ) where
carrier := {x | Tendsto (⇑x) cofinite (𝓝 0)}
zero_mem' := by
simp only [Set.mem_ofPred_eq, lp.coeFn_zero]
exact tendsto_const_nhds
add_mem' := by
intro x y hx hy
simp only [Set.mem_ofPred_eq, lp.coeFn_add] at *
have h := hx.add hy
rw [add_zero] at h
exact h
smul_mem' := by
intro c x hx
simp only [Set.mem_ofPred_eq, lp.coeFn_smul] at *
have h := hx.const_smul c
rw [smul_zero] at h
exact h
/-- c₀₀(Γ): the finitely supported functions in ℓ∞(Γ). -/
def c00 (Γ : Type*) : Submodule ℝ (Linf Γ) where
carrier := {x | (Function.support (⇑x)).Finite}
zero_mem' := by
show (Function.support (⇑(0 : Linf Γ))).Finite
rw [lp.coeFn_zero, Function.support_zero]
exact Set.finite_empty
add_mem' := by
intro x y hx hy
simp only [Set.mem_ofPred_eq, lp.coeFn_add] at *
exact (hx.union hy).subset (Function.support_add _ _)
smul_mem' := by
intro c x hx
simp only [Set.mem_ofPred_eq, lp.coeFn_smul] at *
exact hx.subset (Function.support_const_smul_subset c _)
/-- The closed unit ball of a subspace `X` of ℓ∞(Γ), as a subset of ℓ∞(Γ). -/
def ballIn {Γ : Type*} (X : Submodule ℝ (Linf Γ)) : Set (Linf Γ) :=
(X : Set (Linf Γ)) ∩ Metric.closedBall 0 1
/-- `Y` realizes all sign patterns: for every countable `C ⊆ Δ` and every choice of signs
`ε d = ±1` on `C` there is `v ∈ Y` with `v d` of sign `ε d` for every `d ∈ C`. -/
def RealizesSignPatterns {Δ : Type*} (Y : Submodule ℝ (Linf Δ)) : Prop :=
∀ C : Set Δ, C.Countable → ∀ ε : Δ → ℝ, (∀ d ∈ C, ε d = 1 ∨ ε d = -1) →
∃ v : Linf Δ, v ∈ Y ∧ ∀ d ∈ C, 0 < ε d * v d
end Plasticity
Read-back
What the Lean code literally says, in plain math · claude-opus-5-5
This file introduces five definitions in the namespace Plasticity. Throughout, (and ) denote arbitrary types used as index sets: they may be empty, finite, countably infinite or uncountable.
Linf — the space . is the set of all functions whose absolute values are bounded, i.e. is bounded above, with pointwise addition and scalar multiplication and the norm
with the convention when is empty (then ). The metric is . An element is identified with its coordinate function, and denotes its -th coordinate. (This is Mathlib's space of real-valued functions with exponent ; it is a real Banach space.)
c0 — the subspace . is the set of such that along the cofinite filter on ; concretely,
The definition packages this set as a real linear subspace of (the attached proofs only check that it contains and is closed under addition and real scalar multiplication). Edge case: when is finite every subset of is finite (the cofinite filter is the degenerate bottom filter), so the condition holds for every and with the max norm; when is empty this is . When regarded as a normed space in its own right, carries the restriction of .
c00 — the subspace . is the set of whose support is a finite set; i.e. all finitely supported real functions on (these are automatically bounded). Again it is packaged as a real linear subspace of (the attached proofs only check closure under , and scalar multiplication), and as a normed space it carries the restricted sup norm. No closedness or completeness is part of the definition. When is finite, .
ballIn — closed unit ball of a subspace. For any real linear subspace of (no other conditions on ),
regarded as a subset of (not as a subset of a separate type). It always contains .
RealizesSignPatterns. For any real linear subspace of (no other conditions on ), " realizes sign patterns" means:
for every subset that is countable (finite, including empty, or countably infinite), and every function with for every (values of off are unrestricted and irrelevant), there exists with
i.e. and has the sign at every .
No condition is imposed on outside and no bound on is required. Edge cases: the requirement for is met by ; if is empty the property holds trivially; for and nonempty it fails (take a singleton). Whether it holds for a given depends on : for instance, if is infinite, it fails for (a countably infinite would require with infinite support), whereas it holds for (enumerate and take , off ).