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.
A weighted total-variation bound from relative entropy
Statement
Assume AC. Let be Borel probability measures on , and let be Borel measurable with . Then where are probability measures in the sense of Probability measures and probability spaces, is defined in Weighted total variation for probability measures and is the directional relative entropy of Kullback–Leibler divergence. If , the inequality is understood in the extended-real sense and is immediate.
Facts & Assumptions
Given: AC; probability measures ; a Borel weight with finite second moments under both measures.
AC supplies the Radon–Nikodym representative used in the definition of directional relative entropy. The Axiom of Choice The Radon-Nikodym derivative as an almost-everywhere equivalence class
If , then , the density is nonnegative, and with . Kullback–Leibler divergence Integrating against a Radon-Nikodym derivative recovers integration against the measure
The directional relative entropy has finite negative part and . Kullback–Leibler divergence Integrating against a Radon-Nikodym derivative recovers integration against the measure
Cauchy–Schwarz gives for square-integrable . Cauchy-Schwarz inequality for
For , and , because the integral logarithm agrees with the natural logarithm of The natural logarithm as the inverse of the exponential function. The integral logarithm satisfies for and The integral logarithm is the published natural logarithm The natural logarithm as the inverse of the exponential function
Differentiation obeys the sum, scalar-multiple and product rules on ; squares are products of two identical factors. The mean value theorem applies to the resulting continuously differentiable scalar functions. Sums, scalar multiples, products and quotients: , , , and when The integral logarithm satisfies for and The mean value theorem, as the case of Cauchy's: for continuous on with and differentiable on there is with
Integrals of integrable real functions are linear. The Lebesgue integral is linear on
Proof
Given: The hypotheses in the Statement.
Define and for . For , the function satisfies and . The mean value theorem on gives , so . Now satisfies and by [F4]–[F5]; another application of the mean value theorem gives . At the inequality holds because .
For , put . By [F5], and using [F4] and for . The mean value theorem applied first to and then to shows and for . Combining with step 1.1 gives
It remains to consider finite . Let as in [F1]. For every Borel with , the Radon–Nikodym identity gives Applying step 2.1 pointwise and [F3] with factors and bounds this by The moment hypotheses make the first factor finite, and [F1], [F2], and [F6] identify the second integral with the directional relative entropy, using . Taking the supremum over proves the claim.
Depends on
- Cauchy-Schwarz inequality for $L^2$
- The integral logarithm $L$ is the published natural logarithm
- The mean value theorem, as the case $g(x) = x$ of Cauchy's: for $f$ continuous on $[a,b]$ with $a < b$ and differentiable on $(a,b)$ there is $c \in (a,b)$ with $f(b) - f(a) = f'(c)(b-a)$
- The Axiom of Choice
- Kullback–Leibler divergence
- The natural logarithm as the inverse of the exponential function
- Probability measures and probability spaces
- The Radon-Nikodym derivative as an almost-everywhere equivalence class
- Weighted total variation for probability measures
- Sums, scalar multiples, products and quotients: $(f+g)'(c) = f'(c) + g'(c)$, $(\alpha f)'(c) = \alpha f'(c)$, $(fg)'(c) = f'(c)g(c) + f(c)g'(c)$, and $(f/g)'(c) = \bigl(f'(c)g(c) - f(c)g'(c)\bigr)/g(c)^{2}$ when $g(c) \ne 0$
- The integral logarithm satisfies $L'(x)=1/x$ for $x>0$ and $L(1)=0$
- Integrating against a Radon-Nikodym derivative recovers integration against the measure
- The Lebesgue integral is linear on $L^1(\mu)$
Used by
Dependency tree · two levels
58 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
- Alain Durmus and Éric Moulines, Non-asymptotic convergence analysis for the Unadjusted Langevin Algorithm, Lemma 24 and proof §4.6 (standard reference, not scraped)