Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-adaptedPipeline-generatedjudge pass (gpt-6-luna)
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.

Dependent-input Shapley effects are nonnegative and sum to one

Statement

Assume AC and the positive-variance hypotheses of Shapley effects for dependent inputs. Then v is monotone with v(∅)=0, v([d])=V, and 0≤v(S)≤V for every S⊆[d]. Every ϕi≥0, and ∑i=1dϕi=1. Moreover, on the exact-oracle input for this game, the procedure Exact finite-permutation Shapley effects terminates after at most 2d d! coalition-value calls and returns the Shapley effects defined above.

Facts & Assumptions

Given: AC, Y=f(X)∈L2, V=Var⁡(Y)>0, and the variance game v and effects ϕi from the definition.

[A1]

The Axiom of Choice: AC is assumed for the conditional-expectation supplier.

[F1]

Conditional expectation is the l2 orthogonal projection: under AC, conditional expectation is the orthogonal projection onto the corresponding L2 subspace and fixes measurable inputs.

[F2]

Shapley effects for dependent inputs: v(S)=Var⁡(E[Y∣GS]) and ϕi is the average normalized marginal contribution over all permutations.

[F3]

Exact finite-permutation Shapley effects: the procedure initializes the accumulators, starts each permutation with S=∅, adds v(S∪{i})−v(S) at entry i, updates S to S∪{i}, and returns the accumulated values divided by d!V.

Proof

Given: The objects in the Statement and facts [A1], [F1]–[F3].

1.1A1F1givenalgebra

Fix S⊆T⊆[d], and write mS=E[Y∣GS] and mT=E[Y∣GT]. The subspaces of GS- and GT-measurable L2 functions are nested. For every h in the smaller subspace, ⟨mT−Y,h⟩=0 because h also belongs to the larger subspace. Thus mT and Y have the same orthogonal projection onto the smaller subspace, namely mS, so mT−mS⊥mS−EY. Pythagoras gives v(T)−v(S)=∥mT−mS∥22≥0.

2.1F1F2step 1.1algebra

For any permutation π, sum the successive increments v(Piπ∪{i})−v(Piπ) in the order of that permutation. Every intermediate coalition value appears once positively and once negatively, leaving v([d])−v(∅). The projection in [F1] fixes Y, so v([d])=V; the empty-set projection is constant EY, so v(∅)=0. Step 1.1 gives monotonicity, so these endpoint values imply 0≤v(S)≤V for every coalition. Thus the increments sum to V.

3.1F2step 1.1step 2.1givenalgebra

Each increment in step 2.1 is nonnegative by step 1.1. Average over the finite set of d! permutations and divide by V>0. The resulting effects are nonnegative and their sum is V/V=1.

4.1F2F3givenalgebra∎

In the procedure [F3], before each entry i of a permutation π, the current set S is exactly Piπ: it starts empty and each preceding entry is added once. Thus the increment added to Ai is v(Piπ∪{i})−v(Piπ). There are d! permutations with d entries each, so the loops perform finitely many iterations and at most 2d d! exact coalition-value calls; every queried value is available by the stated oracle input. Summing the increments gives Ai=∑π∈Sd(v(Piπ∪{i})−v(Piπ)), and division by d!V returns exactly ϕi by [F2].

Depends on

Used by

Dependency tree · two levels

12 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