Alphabeta Math
LemmaStatement: AI-adaptedProof: AI-generatedPipeline-generated
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.

Descendants of a youngest child are conditionally independent of its parent

Statement

Assume Countable Choice, or suppose the specified noise densities are Borel measurable. Let a finite causally sufficient additive-noise model have graph G, and let y→z be an edge such that there is no directed path from z to any other child of y. Write D=de⁡G(z) for the strict descendants of z and S=V∖(D∪{y,z}). Use the supplied Borel noise densities, or fix positive finite Borel versions under Countable Choice, and let p be their shifted-noise product density. Then there are positive measurable functions A(y,s,z) and B(d,s,z) such that p(d,y,s,z)=A(y,s,z)B(d,s,z), where A does not depend on d and B does not depend on y. Use marginal density versions obtained by iterated integration. Then pXD,Xy,XS,Xz(d,y,s,z) pXS,Xz(s,z)=pXD,XS,Xz(d,s,z) pXy,XS,Xz(y,s,z) for almost every (d,y,s,z). This is the density conditional-independence factorization of XD and Xy given (XS,Xz). Moreover, for every s with 0<pXS(s)<∞, the conditional law of (XD,Xy) given XS=s satisfies the density conditional-independence factorization given Xz.

Facts & Assumptions

Given: A finite causally sufficient additive-noise model, an edge y→z satisfying the youngest-child condition in the Statement, and its positive joint density.

[F1]

In a topological order, Xv=fv(Xpa⁡(v))+Nv recursively, and the noises are mutually independent. Continuous additive-noise models on finite DAGs

[F2]

For supplied Borel noise densities, or positive finite Borel versions chosen under Countable Choice, the selected Borel joint density is p(x)=∏v∈Vqv(xv−fv(xpa⁡(v))), with each factor depending only on that vertex and its parents. Positive joint density of a finite additive-noise model

[F3]

Nonnegative product-measurable functions can be integrated in any finite coordinate order by iterated sections. Tonelli's theorem for nonnegative measurable functions on a sigma-finite product

Proof

Given: The hypotheses in the Statement.

1.1givenF1construct

The Borel noise densities specified in the Statement give the Borel product [F2]; when necessary their versions are supplied under Countable Choice by that same result. Every shifted factor and every finite regrouping is Borel measurable. If y were a parent of any d∈D, then d would be another child of y and also a descendant of z, contrary to the hypothesis. Thus no vertex in D has y as a parent. Every parent of a vertex in D therefore lies in D∪{z}∪S.

2.1F2step 1.1algebra

No vertex in D can be a parent of a vertex outside D: that would make the latter a descendant of z. Hence each density factor in [F2] for a vertex outside D depends only on (y,s,z), while each factor for a vertex in D depends only on (d,s,z). Regrouping the finite product gives p(d,y,s,z)=A(y,s,z)B(d,s,z) for nonnegative measurable functions A,B.

3.1step 2.1F3algebra∎

By [F3], the marginal densities obtained from the full product are pXS,Xz(s,z)=(∫A(y,s,z) dy)(∫B(d,s,z) dd), pXy,XS,Xz(y,s,z)=A(y,s,z)∫B(d,s,z) dd,pXD,XS,Xz(d,s,z)=B(d,s,z)∫A(y,s,z) dy. These identities hold almost everywhere where the marginal densities are finite; Tonelli also permits infinite values on exceptional null sets. Multiplying the last two identities and comparing with the first proves the factorization in the Statement almost everywhere. For any fixed s with 0<pXS(s)<∞, put as(z)=∫A(y,s,z) dy and bs(z)=∫B(d,s,z) dd. Tonelli gives pXS(s)=∫as(z)bs(z) dz<∞, so for almost every z under that conditional law both as(z) and bs(z) are finite and positive. The conditional densities of Xy and XD given (XS=s,Xz=z) are respectively A(y,s,z)/as(z) and B(d,s,z)/bs(z); their product is the joint conditional density. This proves the stated conditional-independence factorization at each such s.

Depends on

Used by

Dependency tree · two levels

29 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