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.
The Matérn native space has its weighted Fourier norm and strict Gram positivity
Statement
Assume the Axiom of Choice conservatively for the declared Hilbert-space dependency. Fix , , , and , and let be the normalized Matérn kernel and positive spectral density of The normalized Matérn kernel in the 2π Fourier convention. Let consist of the inverse Fourier representatives with , with inner product Then this is a Hilbert space, each kernel section reproduces evaluation at , and its span is dense. There are constants , depending only on , such that For every finite list of distinct sites , the Matérn Gram matrix is strictly positive definite.
Facts & Assumptions
Given: The Axiom of Choice, fixed , the definitions of , and distinct sites .
is positive and integrable with integral one; the inverse Fourier representation of is injective. (The normalized Matérn kernel in the 2π Fourier convention, The Matérn spectral integral, normalization, and radial formula, The weighted Fourier model is an injective Hilbert representation)
The Fourier norm is , and the inverse Fourier representative is injective. (Fourier Bessel-potential and restriction-quotient Sobolev spaces, The weighted Fourier model is an injective Hilbert representation)
A closed subspace of a Hilbert space has an orthogonal complement, and a subspace whose orthogonal complement is zero is dense. (Orthogonal decomposition by a closed subspace)
A nonzero univariate polynomial has at most its degree many roots. The real numbers form an ordered field, hence are infinite. (A nonzero polynomial of degree over an integral domain has at most distinct roots, The reals form a totally ordered field, The reals form a field)
The complex exponential has derivative itself, the chain and product rules apply to its finite linear combinations, Cauchy–Schwarz holds in , and a box of positive side lengths has positive Lebesgue measure. (The complex exponential is entire and its complex derivative is itself, The chain rule, in one line from Carathéodory: if is differentiable at and is differentiable at , then is differentiable at with , Sums, scalar multiples, products and quotients: , , , and when , Cauchy-Schwarz inequality for , A box in with parameters is Lebesgue measurable of measure , whichever of its faces are included)
A nonnegative measurable function with integral zero vanishes almost everywhere. (A nonnegative measurable function has integral exactly when it vanishes almost everywhere)
For , the function is increasing on by its positive derivative and the mean value theorem. (Continuity and derivatives of positive-base real powers, The mean value theorem, as the case of Cauchy's: for continuous on with and differentiable on there is with )
Full AC is carried only to conservatively cover the legacy declared dependency of the published completeness result; the Fourier calculations and finite-site selection use no arbitrary-index choice. (The Axiom of Choice, The Axiom of Countable Choice ())
Proof technique: direct.
Proof
Given: The data above; all Fourier transforms use the negative-sign convention.
Since , the two positive functions and have ratio bounded above and below by positive constants: compare with separately using the minimum and maximum of and , and use monotonicity of the positive power [F7]. Multiplying by and taking square roots gives the displayed norm equivalence. Thus the weighted norms defining and are equivalent and have the same representatives by injectivity [F1,F2]; completeness of implies completeness of .
For a complex coefficient vector , the Hermitian quadratic form of the Gram matrix is If it is zero, [F6] says times the squared modulus of the continuous exponential sum vanishes almost everywhere. Since , the sum is zero almost everywhere. A continuous function zero almost everywhere is zero everywhere, because any point of nonzero value would have a neighborhood on which its modulus is bounded below, and such a neighborhood contains a positive-measure box [F5].
The spectral representative of is . Its norm squared is , and for , The integral exists because Cauchy--Schwarz bounds by . Hence point evaluation is continuous and the stated kernel sections reproduce it.
If is orthogonal to every section centered at , step 2.1 gives for every . Fourier injectivity [F1] then makes in . By [F3] the span of all sections is dense.
Choose so that the numbers are pairwise distinct. For each pair , the polynomial is nonzero because ; [F4] says its zero set is finite. The ordered field has , so the natural multiples are strictly increasing and is infinite. Thus a real exists outside the finite union of root sets. Restrict the vanishing exponential sum to and differentiate at for orders . This gives . For each , the Lagrange polynomial has degree at most and satisfies if and otherwise. Taking the same linear combination of the moment equations as the coefficients of yields . Thus all , and the real Gram matrix is strictly positive definite.
Depends on
- Radial kernels, polynomial augmentation, and site geometry
- The normalized Matérn kernel in the 2π Fourier convention
- Fourier Bessel-potential and restriction-quotient Sobolev spaces
- The Matérn spectral integral, normalization, and radial formula
- The weighted Fourier model is an injective Hilbert representation
- Schwartz functions are dense in the Fourier Bessel-potential norm
- Orthogonal decomposition by a closed subspace
- A nonzero polynomial of degree $n$ over an integral domain has at most $n$ distinct roots
- The reals form a totally ordered field
- The reals form a field
- Continuity and derivatives of positive-base real powers
- 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 $f(b) - f(a) = f'(c)(b-a)$
- 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
- The complex exponential is entire and its complex derivative is itself
- Cauchy-Schwarz inequality for $L^2$
- A nonnegative measurable function has integral $0$ exactly when it vanishes almost everywhere
- 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 chain rule, in one line from Carathéodory: if $g$ is differentiable at $c$ and $f$ is differentiable at $g(c)$, then $f \circ g$ is differentiable at $c$ with $(f \circ g)'(c) = f'(g(c))\,g'(c)$
- The Axiom of Choice
- The Axiom of Countable Choice ($\mathrm{AC}_\omega$)
Used by
Dependency tree · two levels
99 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
- Rasmussen and Williams, Gaussian Processes for Machine Learning, Chapter 4, equations (4.5)–(4.8) (standard reference, not scraped)
- Gregory E. Fasshauer, MATH 590: Meshfree Methods, Chapter 13, pp. 18–25 (standard reference, not scraped)