Chapter LinftyMaximalAbelian
DefinitionChapterLinftyMaximalAbeliandefinitionsformalizationtimepiece
Formal definitions for the timepiece Lean 4 formalization (source chapter BookProof/ChapterLinftyMaximalAbelian.lean): generated def bundle for ChapterLinftyMaximalAbelian. See BookProof/ChapterLinftyMaximalAbelian.lean for full context.
Definition code
import Definitions.Def_ChapterLinftyMultiplication
import Mathlib
/-!
# `L∞(μ)` is *maximal* abelian on `L²(μ)` — the diffuse half of the classification
`ChapterLinftyMultiplication` builds the diffuse model of the abelian
classification: for essentially bounded `φ : α → ℂ` the multiplication operators
`multOp φ : L²(μ) →L[ℂ] L²(μ)` form a unital, abelian, star-closed and faithful
algebra. `ChapterAbelianAtomicCondensation` proves the *atomic* condensation:
a purely atomic maximal abelian algebra **is** the diagonal algebra `ℓ∞`.
This module proves the matching statement for the diffuse model, which is the
structural fact the classification actually rests on:
**the multiplication algebra is its own commutant.**
Concretely (`commutant_eq_multOp`), on a finite measure space every bounded
operator `T` on `L²(μ)` that commutes with *every* multiplication operator is
itself a multiplication operator `T = multOp ψ`, and its symbol is
`ψ = T(1)` with `‖ψ‖_∞ ≤ ‖T‖` (`symbol_ae_norm_le`,
`memLp_top_symbol`). Hence the algebra is **maximal abelian**
(`multOp_algebra_maximal_abelian`): no bounded operator can be adjoined to it
without breaking commutativity. Specialized to Lebesgue measure on `[0,1]`
(`unitInterval_multOp_maximal_abelian`) this is the diffuse companion of the
atomic condensation — the two ends of the classification list.
The proof is the classical one:
* `symbol T := T(1)` makes sense because `1 ∈ L²(μ)` for a finite measure;
* `symbol_mul` — commutation gives `T(φ) = φ · ψ` for every bounded `φ`;
* `symbol_ae_norm_le` — testing on the indicator of `{‖ψ‖ ≥ ‖T‖ + ε}` and
comparing the two `L²` norms forces that set to be null, so `ψ ∈ L∞(μ)`;
* `commutant_eq_multOp` — `T` and `multOp ψ` are continuous and agree on the
indicator functions, hence everywhere, by `Lp.induction`.
Everything is `sorry`-free and `axiom`-free (only `propext`, `Classical.choice`,
`Quot.sound`).
-/
noncomputable section
open MeasureTheory ENNReal Complex
namespace BookProof.ChapterLinftyMaximalAbelian
open BookProof.ChapterLinftyMultiplication
variable {α : Type*} [MeasurableSpace α] {μ : Measure α} [IsFiniteMeasure μ]
/-- The constant function `1`, as an element of `L²(μ)` (available because `μ`
is finite). It is the cyclic vector of the multiplication algebra. -/
def oneLp (μ : Measure α) [IsFiniteMeasure μ] : Lp ℂ 2 μ :=
MemLp.toLp (fun _ : α => (1 : ℂ)) (memLp_const 1)
/-- The **symbol** of an operator: a strongly measurable representative of
`T(1)`. For a `T` commuting with all multiplications this is the essentially
bounded function that `T` multiplies by. -/
def symbol (T : Lp ℂ 2 μ →L[ℂ] Lp ℂ 2 μ) : α → ℂ :=
(Lp.aestronglyMeasurable (T (oneLp μ))).mk _
/-- An operator commuting with every multiplication operator. -/
def CommutesWithMultOps (T : Lp ℂ 2 μ →L[ℂ] Lp ℂ 2 μ) : Prop :=
∀ (φ : α → ℂ) (hφ : MemLp φ ⊤ μ), T.comp (multOp φ hφ) = (multOp φ hφ).comp T
end BookProof.ChapterLinftyMaximalAbelian
end
Source