Sequence relations, measure equivalence, and Borel reductions
DefinitionMeasureEquivalenceSequencesLet and carry their product Borel structures. For a real exponent , define by summability of , and define by . For probability measures on , write when and . For a relation on , define
The probability space uses the sigma-algebra generated by evaluations on Borel sets. The exponent restriction is imposed by the target theorems. These definitions fix the common interface of the two main results.
import Mathlib.MeasureTheory.Measure.ProbabilityMeasure
import Mathlib.Analysis.SpecialFunctions.Pow.Real
/-! Source relations and actual mutual absolute continuity, using the canonical
product Borel structures and the evaluation sigma-algebra on probabilities. -/
open MeasureTheory Filter
open scoped Topology
namespace MeasureEquivalenceMission
abbrev RealSequences := ℕ → ℝ
abbrev CantorSpace := ℕ → Bool
def LpDifference (p : ℝ) (x y : RealSequences) : Prop :=
Summable (fun n => |x n - y n| ^ p)
def NullDifference (x y : RealSequences) : Prop :=
Tendsto (fun n => x n - y n) atTop (𝓝 0)
def MeasureEquivalent (μ ν : ProbabilityMeasure CantorSpace) : Prop :=
(μ : Measure CantorSpace) ≪ ν ∧ (ν : Measure CantorSpace) ≪ μ
def BorelReducesToMeasureEquivalence (E : RealSequences → RealSequences → Prop) : Prop :=
∃ f : RealSequences → ProbabilityMeasure CantorSpace,
Measurable f ∧ ∀ x y, E x y ↔ MeasureEquivalent (f x) (f y)
end MeasureEquivalenceMission
Read-back
What the Lean code literally says, in plain math · Fresh-context independent OpenAI model agent; exact model/deployment identifier unavailable.
is the type of all functions , where ; thus its elements are arbitrary infinite real sequences, including unbounded sequences, with no summability or convergence requirement. When used as a measurable space here, it carries the product of the usual Borel -algebras on , equivalently the Borel -algebra of the product topology.
is the type of all functions from to the two Boolean values. Its measurable structure here is the product of the full -algebras on the two-point factors, equivalently the Borel -algebra of their discrete product topology.
For every real number and every pair of arbitrary real sequences , means that the real-valued sequence is summable, that is, converges to a finite real number. The exponent is unrestricted in this definition, including zero and negative values. For a positive base , the real power is ; at a zero base the conventions are and for every nonzero real , including negative . Consequently gives the nonsummable constant sequence , even when , whereas gives the summable zero sequence when . The summation includes coordinate .
For every pair of arbitrary real sequences , means that tends to as tends to infinity: for every real there exists such that every with satisfies . Here includes , and neither sequence is individually required to converge, be bounded, or be summable; in particular satisfies the condition.
For every pair of probability measures on with its product Borel -algebra, so that each measure has total mass , means the conjunction of the following two statements: for every subset , implies , and for every subset , implies . Thus the underlying measures have exactly the same null subsets. The subset quantifiers include the empty set and nonmeasurable subsets, using the measures' outer-measure values on arbitrary subsets.
For every binary predicate on the space of all real sequences, means that there exists a function , where carries its product Borel -algebra and consists of all probability measures of total mass on that space, such that is measurable and, for every , holds if and only if both of the following hold: for every subset , implies , and for every subset , implies . The source carries its product Borel -algebra, and carries the -algebra generated by the evaluation maps for all Borel subsets , with their values in the nonnegative extended real numbers; consequently the measurability condition requires every map , for Borel , to be Borel measurable. The null-set conditions quantify over all subsets, including nonmeasurable subsets via outer-measure values and the empty set. The predicate is arbitrary, with no assumed equivalence-relation or measurability hypothesis, and range over all real sequences, including , with coordinates indexed from .
Confirmed by the mission captain (proposal self-audit).