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 full-rank local Hessian does not capture both posterior modes
Statement refuted
A nondegenerate local Hessian at one mode does not by itself ensure that a one-mode Laplace Gaussian approximates the full posterior law in total variation, even when the full-data model identifies the parameter.
Facts & Assumptions
Given: Assume Countable Choice. Let with its Borel sigma-algebra, let be an integer, and set Give the data space its power-set sigma-algebra and counting measure, and define the th experiment by Use the uniform prior on Borel , observe in the th experiment, and use the likelihood version . Let be the exact target and let be its image on under the inclusion . Define the one-mode Laplace proxy on by the density
Countable Choice means every countable family of nonempty sets has a choice function. The Axiom of Countable Choice ()
A fixed measurable likelihood version and finite positive evidence define ; the likelihood version here is the density relative to the stated counting measure. Exact inferential targets and approximation chains Dominated statistical experiments A measurable function between measurable spaces
In the parametric model, observational equivalence means equality of sampling laws, and the identity estimand is identifiable when equivalent parameters agree. Statistical experiments Parametric statistical models Observational equivalence Estimands on a statistical model Identifiability of an estimand
The cited Laplace asymptotic theorem assumes a unique interior global maximizer in addition to a positive-definite negative Hessian and local and global domination conditions. Laplace asymptotic under a unique nondegenerate interior maximum
Under Countable Choice, the one-dimensional Gaussian integral gives , and the positive-definite Gaussian integral lemma includes its normalization. Gaussian integral for a positive-definite quadratic form
Under Countable Choice, an affine diffeomorphism changes a nonnegative Lebesgue integral by its absolute Jacobian. A C^1 diffeomorphism satisfies the change-of-variables formula for nonnegative Lebesgue measurable functions
For increasing measurable sets, measure of the union is the supremum of their measures. Continuity from below for measures
Total variation is the supremum of event-probability differences. Total variation distance for probability laws
Borel subsets of Euclidean space are Lebesgue measurable, bounded boxes have their usual volume, and the Borel sigma-algebra on is the trace of the one on . Assuming countable choice, every Borel subset of is Lebesgue measurable A box in with parameters is Lebesgue measurable of measure , whichever of its faces are included The Borel sigma-algebra of a subspace is the trace of the ambient Borel sigma-algebra
The real numbers form a complete ordered field, so their natural numbers are unbounded; and , and squaring is strictly increasing on nonnegative numbers. Uniqueness of nonnegative square roots therefore makes the square root order-preserving. The Cauchy-sequence reals have the least-upper-bound property The reals form a totally ordered field Every complete ordered field is Archimedean Basic properties of the absolute value Squaring is monotone on the nonnegatives Square roots exist: a unique with ; the positives are
Counterexample
Given: The hypotheses and explicit model in Facts & Assumptions.
On , , so . Also , hence ; the three entries are nonnegative and sum to one. Thus each is a probability measure on . Since is injective, implies , so the identity estimand is identifiable by [F2].
The Borel sets of are Lebesgue measurable, , and is continuous and satisfies . Therefore is a probability measure, is measurable, and . By [F1], is the exact posterior target with density proportional to relative to .
The phase has exactly two global maxima, at , because exactly there and is positive elsewhere. Direct differentiation gives , so ; each local negative Hessian is full rank, but [F3]'s unique-maximum hypothesis fails. Writing gives , so its quadratic term at is . The one-mode Laplace density is therefore proportional to , exactly the stated , and integrates to one by [A1, F4].
Under the affine change , [F5] sends to the standard normal probability of . These sets are increasing. Given any , choose by [F9]; positivity and monotonicity of the square root then give , so their union is all of . Continuity from below [F6] therefore gives .
Define on ; it is a nonnegative even Borel function. Apply [F5] to the reflection diffeomorphism on , using the function : the unnormalized integrals of over and are equal. The singleton has Lebesgue measure zero by [F8], so these two integrals split the prior-weighted posterior normalizer equally. Thus .
Using event in [F7], , whose right side tends to by steps 2.1 and 2.2. Thus , proving that the local curvature at one mode misses persistent posterior mass at the other.
Depends on
- The Axiom of Countable Choice ($\mathrm{AC}_\omega$)
- Dominated statistical experiments
- Estimands on a statistical model
- Exact inferential targets and approximation chains
- Identifiability of an estimand
- Lebesgue measurable sets, the family $\mathcal{L}(\mathbb{R}^n)$, and the restricted set function $\lambda_n$
- A measurable function between measurable spaces
- The nonnegative Lebesgue integral
- Observational equivalence
- Parametric statistical models
- Probability measures and probability spaces
- Statistical experiments
- Total variation distance for probability laws
- Basic properties of the absolute value
- Squaring is monotone on the nonnegatives
- Gaussian integral for a positive-definite quadratic form
- Assuming countable choice, every Borel subset of $\mathbb{R}^n$ is Lebesgue measurable
- The Borel sigma-algebra of a subspace is the trace of the ambient Borel sigma-algebra
- A C^1 diffeomorphism satisfies the change-of-variables formula for nonnegative Lebesgue measurable functions
- The Cauchy-sequence reals have the least-upper-bound property
- Continuity from below for measures
- Laplace asymptotic under a unique nondegenerate interior maximum
- A box in $\mathbb{R}^n$ with parameters $a_i\le b_i$ is Lebesgue measurable of measure $\prod_{i<n}(b_i-a_i)$, whichever of its faces are included
- Every complete ordered field is Archimedean
- Square roots exist: a unique $\sqrt{a} \ge 0$ with $(\sqrt{a})^2 = a$; the positives are $\{x^2 : x \neq 0\}$
- The reals form a totally ordered field
Used by
Nothing in the library uses this result yet.
Dependency tree · two levels
134 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
- Laplace method for integrals, PHYS 2400 – Mathematical methods for the physical sciences, University of Connecticut (standard reference, not scraped)
- NIST Digital Library of Mathematical Functions, §2.3(iii), Laplace's Method (standard reference, not scraped)
- Francis Bach, Approximating Integrals with Laplace's Method (standard reference, not scraped)