Alphabeta Math
False statementConstruction: Literature-sourcedVerification: AI-adaptedPipeline-generatedjudge pass (gpt-5.6-terra)audited 2026-09-14
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Birkhoff's theorem requires integrability

Statement

Assume the Axiom of Countable Choice. False claim: Birkhoff's finite almost-everywhere convergence conclusion holds for every finite-valued measurable observable, without integrability.

Facts & Assumptions

Given: Countable choice, Lebesgue probability on [0,1), its doubling map D2, and

f(0)=0,f(x)=1/x(0<x<1).

[F1]

It suffices to test strict superlevel sets for extended-real measurability (Extended-real-valued measurable functions), and Borel sets are Lebesgue measurable under countable choice (Assuming countable choice, every Borel subset of Rn is Lebesgue measurable).

[F3]

The harmonic series diverges (For rational p>0, ∑1/kp converges iff p>1), and monotone convergence applies to increasing nonnegative measurable functions (Monotone convergence for the integral).

[F4]

The doubling map is ergodic (Doubling is ergodic for Lebesgue measure); Birkhoff gives invariant finite limits for integrable truncations (Birkhoff pointwise ergodic theorem), and ergodicity makes those limits constant (Equivalent invariant-set and invariant-function criteria for ergodicity).

[F5]

Dominated convergence and invariance of integrals identify the constant limits (Dominated convergence, Integral invariance under measure-preserving maps).

[F6]

A countable union of null sets is null (Finite and countable subadditivity of measures).

Refutation

technique · constructive truncation argument
1.1F1constructalgebra

For a<0, {f>a}=[0,1); for a=0, it is (0,1); and for a>0 it is (0,min⁡{1,1/a}), with the right endpoint omitted if it equals 1. These are Borel. Hence [F1] makes the everywhere-finite function f Lebesgue measurable.

1.2F2algebra

On Ik=[1/(k+1),1/k) one has f≥k. For every N, the simple function sN=∑k=1Nk1Ik therefore satisfies sN≤f, and ∫sN dλ=∑k=1Nk(1k−1k+1)=∑k=1N1k+1.

2.1F2F3step 1.2

By [F2]–[F3], the right side is unbounded, so ∫f dλ=+∞. Thus f is not integrable under the integrability convention of Integrable real and complex functions, and their integrals.

3.1F3step 1.1step 2.1

For each integer m≥1, let fm=min⁡{f,m}. It is measurable, bounded, and hence integrable on this probability space; moreover fm↑f. Monotone convergence and step 2.1 give cm:=∫fm dλ↑+∞.

4.1F4F5step 3.1

By [F4], Anfm converges almost everywhere to an invariant function and that function equals a constant almost everywhere. Since 0≤Anfm≤m, [F5] identifies this constant as lim⁡n∫Anfm dλ=∫fm dλ=cm.

5.1F6step 3.1step 4.1

Remove the countable union of the null exceptional sets in step 4.1; it is null by [F6]. At every remaining x, for every m and n, Anf(x)≥Anfm(x), whence lim inf⁡nAnf(x)≥cm. Because cm↑+∞, this says Anf(x)→+∞.

6.1step 2.1step 5.1discharge-construct∎

Thus a finite-valued measurable observable can have divergent-to-infinity ergodic averages almost everywhere. This refutes the finite-limit claim and shows exactly why the L1 hypothesis cannot be omitted. Countable choice is inherited from the Lebesgue and doubling-map suppliers; all truncations are explicit.

Depends on

Used by

Nothing in the library uses this result yet.

Dependency tree · two levels

88 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources