Pipeline-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.

✓ 13 results · all verified · 10 also independently AI-judged

Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 3 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Volume integral scattering and partial wave error

1 · Prerequisites

2 · Summary

This page defines the three-dimensional outgoing volume integral model on $L^{2} (D)$ , proves boundedness and compactness with the singular diagonal treated explicitly, and separates the small-norm Born branch from the injective Fredholm branch. Its cell-volume product rule integrates complete cell pairs and obtains eventual solution and far-field convergence from a locally verified collective-compactness argument. The radial section derives conditional spherical-harmonic mode equations with explicit regular-origin, outgoing, and matching hypotheses, and proves a uniform cutoff estimate under an explicit exponential coefficient-decay assumption. The PC-SFUN page supplies certified scaled and unscaled spherical Bessel and outgoing Hankel values and first argument derivatives on positive real argument interfaces. The radial matching discussion constructs the $k R$ interface from certified parameter enclosures; the square-well example also constructs $\sqrt{k^{2} + V_{0}} R$ and propagates absolute error through its nonzero matching denominator. These numerical representation conditions leave the exact real-parameter scattering statements unchanged. No unconditional scattering uniqueness or resonance existence is asserted.

The following schematic records the radial regions and the conditions at their interface. It is not to scale, its horizontal coordinate is radius rather than a graph of $w_{ℓ m}$ , and the radial theorem is proved from the differential equation and continuity hypotheses rather than from the drawing.

\begin{tikzpicture}[x=1cm,y=1cm,>=stealth]
  \draw[->] (0,0) -- (8.1,0) node[right] {$r$};
  \draw[very thick,blue] (0.5,0) -- (4,0);
  \draw[very thick,red] (4,0) -- (7.5,0);
  \draw[dashed] (4,-0.7) -- (4,1.15);
  \fill (0.5,0) circle (1.5pt);
  \fill (4,0) circle (1.5pt);
  \node[above,blue,align=center] at (2.15,0.28) {Interior: radial contrast};
  \node[above,red,align=center] at (6.05,0.28) {Exterior: incident plus outgoing mode};
  \node[below,align=center] at (0.5,-0.18) {Origin: regular mode};
  \node[below,align=center] at (4,-0.75) {Radius R: match value and radial derivative};
  \draw[->,red] (6.75,0.78) -- (7.7,0.78) node[right] {outgoing};
\end{tikzpicture}

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-luna)Open item page →

The three-dimensional volume integral scattering model

Definition

Assume Countable Choice. Let $D \subset R^{3}$ be a bounded Lebesgue-measurable set, let $M : = ∣ D ∣$ , let $q \in L^{\infty} (D; C)$ , and let $k \in R$ . Here $L^{2}$ and $L^{\infty}$ are equivalence classes modulo null functions (The space $L^{p} (μ)$ as the quotient by null functions), and $L^{\infty}$ means essentially bounded with respect to Lebesgue measure (The essential supremum of a measurable function with respect to a measure). We regard $q$ as zero off $D$ . The convention is the volume equation $u = f + K_{k} u, K_{k} u (x) : = \int_{D} G_{k} (x - y) q (y) u (y) d y, G_{k} (z) : = \frac{e^{i k ∣ z ∣}}{4 π ∣ z ∣} (z \neq 0) .$ Set $G_{k} (0) : = 0$ ; when $M > 0$ the diagonal in $D \times D$ is product-null, so this value is only a representative convention. The input and unknown are classes in complex $L^{2} (D)$ , and the displayed equation is interpreted as an equality of those classes. Its formula is well defined by The volume integral operator is well defined and bounded on L2 ↗.

The associated far-field map is $F_{k} [v] (θ) : = \frac{1}{4 π} \int_{D} e^{- i k θ \cdot y} q (y) v (y) d y, θ \in S^{2} .$ Here $S^{2}$ has the surface measure of The polar surface set function on the unit sphere. The map is well defined as a continuous function of $θ$ by The volume far-field map is bounded and is the uniform far-field limit ↗.

For $x \notin \overline{D}$ , the exterior volume potential is $V_{k} v (x) : = \int_{D} G_{k} (x - y) q (y) v (y) d y .$ For each $v \in L^{2} (D; C)$ , the essential-supremum result gives $∣ q ∣ \leq ∥ q ∥_{\infty}$ almost everywhere, and Cauchy--Schwarz gives $∥ q v ∥_{L^{1} (D)} \leq ∥ q ∥_{\infty} \sqrt{M} ∥ v ∥_{2}$ . Thus, for $x \notin \overline{D}$ , the exterior volume-potential integral is absolutely convergent: $\overline{D}$ is closed, so its complement contains a ball about $x$ disjoint from $D$ .

If $M = 0$ , define $L^{2} (D) = {0}$ and $K_{k} = 0$ ; all solution and far-field statements on that space are the zero-space statements. The substantive operator results below assume $0 < M < \infty$ .

This is a stipulated integral-equation model with the sign convention $(- Δ - k^{2}) (u - f) = q u$ when the corresponding distributional Helmholtz equation is used. In the acoustic notation of Scherzer–Mindrinos, their equation $u = u_{i} - k^{2} Φ * ((1 - n) u)$ corresponds to $q = k^{2} (n - 1)$ . No existence or uniqueness for all contrasts is asserted by this definition.

LemmaStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-6-luna)Open item page →

Rescaling finite-measure kernel operators to probability spaces

Statement

Assume AC. Let $(X, μ)$ and $(Y, ν)$ be finite measure spaces with masses $0 < M_{X} : = μ (X) < \infty$ and $0 < M_{Y} : = ν (Y) < \infty$ . Put $d \hat{μ} = d μ / M_{X}$ and $d \hat{ν} = d ν / M_{Y}$ . For $a \in L^{2} (X \times Y, μ \times ν)$ , let $T_{a} : L^{2} (Y, ν) \to L^{2} (X, μ)$ be its square-integrable-kernel operator and let $\hat{T}$ be the operator with kernel $\hat{a} : = \sqrt{M_{X} M_{Y}} a$ against $\hat{ν}$ . The maps $U_{X} g = \sqrt{M_{X}} g$ and $U_{Y} h = \sqrt{M_{Y}} h$ are unitary and $U_{X} T_{a} U_{Y}^{- 1} = \hat{T} .$ Moreover $∥ \hat{a} ∥_{L^{2} (\hat{μ} \times \hat{ν})} = ∥ a ∥_{L^{2} (μ \times ν)}$ , so the operator norms and kernel $L^{2}$ norms agree. If either mass is zero, the corresponding $L^{2}$ space and kernel operator are zero.

For $X = Y = D$ with Lebesgue mass $M > 0$ , the normalized kernel is $\hat{a} = M a$ and the measure is $d \hat{x} = d x / M$ .

Facts & Assumptions

Given: AC, finite measure spaces $(X, μ)$ and $(Y, ν)$ , their positive masses, and a square-integrable kernel class $a$ .

[F1]

The complex $L^{2}$ norm is the quotient norm induced by $\int ∣ f ∣^{2}$ ; Cauchy–Schwarz and representative independence hold for this norm (The space $L^{p} (μ)$ as the quotient by null functions, Complex Lp classes and Euclidean test-function conventions, The complex $L^{2}$ pairing is well-defined and satisfies Cauchy–Schwarz).

[F2]

Product integration and section integrals are valid almost everywhere on the completed product under Countable Choice, which AC supplies (The completed product measure, Tonelli and Fubini for the completed product, with only almost-everywhere section measurability, The Axiom of Choice).

Proof

Given: AC, the spaces, $M_{X}, M_{Y} > 0$ , and $a \in L^{2} (μ \times ν)$ .

1.1givenF1algebra

For every $g \in L^{2} (X, μ)$ , scaling the integral first for nonnegative simple functions and then by the definition of the nonnegative integral gives $∥ U_{X} g ∥_{L^{2} (\hat{μ})}^{2} = \int M_{X} ∣ g ∣^{2} d \hat{μ} = \int ∣ g ∣^{2} d μ = ∥ g ∥_{L^{2} (μ)}^{2}$ . Its inverse is $h \mapsto h / \sqrt{M_{X}}$ , so $U_{X}$ is unitary; the same calculation proves the assertion for $U_{Y}$ .

2.1givenF1F2step 1.1algebra

For $g \in L^{2} (Y, \hat{ν})$ , $U_{Y}^{- 1} g = g / \sqrt{M_{Y}}$ . Using $d ν = M_{Y} d \hat{ν}$ in the integral defining $T_{a}$ and then multiplying the output by $\sqrt{M_{X}}$ gives $U_{X} T_{a} U_{Y}^{- 1} g = \int_{Y} \sqrt{M_{X} M_{Y}} a (x, y) g (y) d \hat{ν} (y)$ . The equality holds almost everywhere by [F2], and is independent of representatives by [F1].

3.1F1F2step 2.1algebra∎

Scaling both factors of the product measure gives $∥ \hat{a} ∥_{2}^{2} = M_{X} M_{Y} \int ∣ a ∣^{2} d (\hat{μ} \times \hat{ν}) = \int ∣ a ∣^{2} d (μ \times ν)$ . Unitary equivalence in step 2.1 therefore preserves the operator norm and the kernel $L^{2}$ norm. If either mass is zero, every $L^{2}$ class on that factor is zero, so the integral operator is zero.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

Dyadic cell-average projections converge strongly in L2 of a bounded set

Statement

Assume Countable Choice. Let $D \subset R^{3}$ be bounded and Lebesgue measurable, and choose a half-open cube $Q$ of positive side length containing $D$ . Subdivide $Q$ into its nested dyadic cubes at mesh $h_{N} = 2^{- N} side (Q)$ . Let $E_{N, j} = D \cap Q_{N, j}$ range over the finitely many cells of positive measure, let $V_{N} \subset L^{2} (D; C)$ be the functions constant on each such cell, and define $P_{N}$ by replacing a function on each $E_{N, j}$ by its cell average. Then $P_{N}$ is the orthogonal projection onto $V_{N}$ , $∥ P_{N} ∥ \leq 1$ , the spaces $V_{N}$ are increasing, and $P_{N} v \to v$ in $L^{2} (D; C)$ for every $v \in L^{2} (D; C)$ .

Facts & Assumptions

Given: Countable Choice, a bounded measurable $D$ , a positive-side containing cube $Q$ , its nested dyadic partitions, and $v \in L^{2} (D; C)$ .

[F1]

Finite linear combinations of dyadic box indicators are dense in real $L^{2}$ of the Borel cube (Base- $b$ box step functions are dense in $L^{2}$ of the cube); the complex conclusion follows by applying the real result to real and imaginary parts.

[F2]

Complex $L^{2}$ Cauchy--Schwarz holds, and Lebesgue measure is invariant under translations and scales under invertible linear changes of variables (The complex $L^{2}$ pairing is well-defined and satisfies Cauchy–Schwarz, Lebesgue outer measure, Lebesgue measurability and Lebesgue measure are unchanged by translation, A linear map $T$ of $R^{n}$ sends Lebesgue measurable sets to Lebesgue measurable sets, with $λ_{n} (T [E]) = ∣ \det T ∣ λ_{n} (E)$ when $T$ is invertible and $T [E]$ Lebesgue null when it is not, Lebesgue measurable sets, the family $L (R^{n})$ , and the restricted set function $λ_{n}$ ).

[F3]

Under Countable Choice, a function measurable for the completion is almost everywhere equal to a Borel-measurable function when the base sigma-algebra is Borel (A function measurable for a completion is almost everywhere equal to one measurable for the original sigma-algebra).

Proof

Given: Countable Choice, a bounded measurable $D$ , a positive-side containing cube $Q$ , its nested dyadic partitions, and $v \in L^{2} (D; C)$ .

1.1givenF2constructalgebra

For each positive-measure cell $E_{N, j}$ the average is finite by Cauchy–Schwarz, since $\int_{E_{N, j}} ∣ v ∣ \leq ∣ E_{N, j} ∣^{1 / 2} ∥ v ∥_{L^{2} (E_{N, j})}$ . For every function constant on the cells, the integral of $v - P_{N} v$ against that function is zero cell by cell. Thus $v - P_{N} v ⊥ V_{N}$ and $P_{N} v \in V_{N}$ , so $P_{N}$ is the orthogonal projection. Pythagoras gives $∥ P_{N} v ∥_{2} \leq ∥ v ∥_{2}$ .

1.2F1F2F3givenconstruct

Extend $v$ by zero from $D$ to $Q$ . The extension is measurable for completed Lebesgue measure; rescale and translate $Q$ to $[0, 1)^{3}$ , which preserves completed measurability by [F2]. Apply [F3] separately to the real and imaginary parts to obtain a Borel-measurable representative of the rescaled extension equal to it almost everywhere. This representative is in Borel $L^{2}$ because the original extension is square integrable and the affine change multiplies its squared integral by the fixed positive factor $side (Q)^{- 3}$ . Applying [F1] to its real and imaginary parts gives dyadic box step functions converging in Borel $L^{2} ([0, 1)^{3})$ , hence also in completed $L^{2}$ . Pulling them back to $Q$ and restricting to $D$ preserves convergence up to the fixed affine measure factor. Every finite family of base-2 boxes is constant on the cells of a sufficiently fine common dyadic partition, so the restrictions belong to some $V_{N}$ . Thus $⋃_{N} V_{N}$ is dense in $L^{2} (D; C)$ ; the partitions are nested, so $V_{N} \subseteq V_{N + 1}$ .

2.1givenstep 1.1step 1.2algebra∎

Fix $ε > 0$ . By step 1.2 choose $w \in V_{m}$ with $∥ v - w ∥_{2} < ε / 2$ . For every $N \geq m$ , nesting gives $P_{N} w = w$ . Therefore $∥ P_{N} v - v ∥_{2} \leq ∥ P_{N} (v - w) ∥_{2} + ∥ v - w ∥_{2} \leq 2 ∥ v - w ∥_{2} < ε$ , by step 1.1. Thus $P_{N} v \to v$ in $L^{2} (D; C)$ .

DefinitionDefinition: Literature-sourcedProof: Not applicableOpen item page →

Spherical harmonics and partial-wave conventions in three dimensions

Definition

Assume Countable Choice and Dependent Choice. Use the unit sphere $S^{2}$ and its polar surface measure $d σ$ from The polar surface set function on the unit sphere. For each integer $ℓ \geq 0$ , let $H_{ℓ}$ be the complex vector space of restrictions to $S^{2}$ of homogeneous degree- $ℓ$ polynomials $P$ on $R^{3}$ satisfying the Euclidean harmonic equation $Δ P = 0$ . For complex-valued polynomials or functions, $Δ$ and its polar-coordinate formulas act componentwise on real and imaginary parts. In polar coordinates $x = r ω$ , define $Δ_{S^{2}}$ by $Δ_{R^{3}} = \partial_{r}^{2} + \frac{2}{r} \partial_{r} + \frac{1}{r^{2}} Δ_{S^{2}},$ equivalently $Δ_{S^{2}} = \frac{1}{\sin θ} \partial_{θ} (\sin θ \partial_{θ}) + \frac{1}{\sin^{2} θ} \partial_{ϕ}^{2}$ away from the poles. Thus if $Y \in H_{ℓ}$ is the restriction of $P$ , then $Δ_{S^{2}} Y = - ℓ (ℓ + 1) Y$ .

For $ℓ \geq 0$ , define the normalized zonal polynomial $P_{ℓ} (t)$ to be the degree- $ℓ$ polynomial satisfying $(1 - t^{2}) P_{ℓ}^{''} (t) - 2 t P_{ℓ}^{'} (t) + ℓ (ℓ + 1) P_{ℓ} (t) = 0, P_{ℓ} (1) = 1.$ The later harmonic-basis lemma proves existence and uniqueness of this normalized polynomial and identifies it with the usual Legendre convention.

The term partial wave of degree $ℓ$ means the $H_{ℓ}$ component of an angular expansion in $L^{2} (S^{2}, d σ; C)$ . Completeness, orthogonality, finite dimension, and the evaluation bound for these components are established in Spherical harmonics form a complete basis with the addition formula ↗.

For a radial field $u (r, ω)$ , its degree- $ℓ$ component is written $u_{ℓ} (r, ω) \in H_{ℓ}$ ; any orthonormal basis within this finite-dimensional space gives the equivalent scalar mode equations. The definition does not select such a basis globally.

LemmaStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-6-luna)Open item page →

The three-dimensional free-space scattering kernel is square integrable

Statement

Assume Countable Choice. Let $D \subset R^{3}$ be bounded and Lebesgue measurable, let $M : = ∣ D ∣$ , $d_{D} : = diam (D)$ , $q \in L^{\infty} (D; C)$ , and $k \in R$ . If $M > 0$ , define $a (x, y) = G_{k} (x - y) q (y)$ off the diagonal and set it to zero on the diagonal. Then $a \in L^{2} (D \times D)$ and $∥ a ∥_{L^{2} (D \times D)}^{2} \leq \frac{∥ q ∥_{\infty}^{2} M d_{D}}{4 π} .$ If $M = 0$ , every measurable kernel on $D \times D$ represents the zero product $L^{2}$ class.

Facts & Assumptions

Given: Countable Choice, the bounded measurable set $D$ , its positive finite measure and diameter, $q \in L^{\infty} (D; C)$ , and real $k$ .

[F1]

For every nonnegative Borel function $g$ on $R^{3}$ , polar coordinates give $\int g (y) d y = \int_{0}^{\infty} \int_{S^{2}} g (r ω) r^{2} d σ (ω) d r$ (Polar coordinates decompose Lebesgue measure into r^{n-1} dr d sigma); a closed ball of radius $r$ has volume $4 π r^{3} / 3$ (A closed three-dimensional ball of radius $r \geq 0$ has volume $4 π r^{3} / 3$ ).

[F2]

Tonelli applies to nonnegative functions on the completed product of the finite Lebesgue measure spaces (Tonelli and Fubini for the completed product, with only almost-everywhere section measurability).

[F3]

A countable subset of $R^{3}$ , in particular a singleton, is Lebesgue null (Every at most countable subset of $R^{n}$ is Lebesgue null; in particular $λ_{1} (Q) = 0$ ).

[F4]

For real $t$ , $∣ e^{i t} ∣ = 1$ ( $\exp (x + i y) = e^{x} (\cos y + i \sin y)$ , $∣ \exp (x + i y) ∣ = e^{x}$ , and $e^{i π} + 1 = 0$ ).

[F5]

Applying The essential supremum is attained as the least essential bound to the measurable real function $∣ q ∣$ gives a measurable representative of $q$ with $∣ q (y) ∣ \leq ∥ q ∥_{\infty}$ everywhere after changing it on a null set; such a change does not affect the product $L^{2}$ class by completed-product Tonelli.

Proof

Given: The data of the Statement and a measurable representative of $q$ satisfying $∣ q (y) ∣ \leq ∥ q ∥_{\infty}$ almost everywhere.

1.1givenF2F3F4F5casesalgebra

If $M = 0$ , then $(D \times D, λ_{3} \times λ_{3})$ is null and the conclusion follows. Suppose $M > 0$ . Off the Borel diagonal, $G_{k} (x - y) q (y)$ is product-measurable, and setting it to zero on the diagonal preserves measurability. For $x \neq y$ , [F4] and real $k$ give $∣ e^{i k ∣ x - y ∣} ∣ = 1$ ; choosing the representative in [F5] yields $∣ a (x, y) ∣^{2} \leq ∥ q ∥_{\infty}^{2} (16 π^{2} ∣ x - y ∣^{2})^{- 1}$ . The diagonal is product-null by [F2] and F3, so its assigned value does not alter the class.

1.2F1givenalgebra

In the positive-measure case $d_{D} > 0$ . For each fixed $x \in D$ , $D \subseteq \overline{B} (x, d_{D})$ . By F1 with the Borel function $g (y) = ∣ x - y ∣^{- 2} 1_{0 < ∣ x - y ∣ \leq d_{D}}$ , polar coordinates give $\int_{D} ∣ x - y ∣^{- 2} d y \leq \int_{0}^{d_{D}} \int_{S^{2}} r^{- 2} r^{2} d σ (ω) d r = 4 π d_{D}$ , using $σ (S^{2}) = 4 π$ , which follows by applying polar coordinates to the unit ball and using its volume $4 π / 3$ (A closed three-dimensional ball of radius $r \geq 0$ has volume $4 π r^{3} / 3$ ).

2.1F2step 1.1step 1.2cases-exhaustivealgebra∎

In the positive-measure case, integrating the bound of step 1.1 in $x$ and applying F2 gives $\int_{D} \int_{D} ∣ a (x, y) ∣^{2} d y d x \leq \frac{∥ q ∥_{\infty}^{2}}{16 π^{2}} M (4 π d_{D}) = \frac{∥ q ∥_{\infty}^{2} M d_{D}}{4 π} < \infty$ . Thus $a \in L^{2} (D \times D)$ . Steps 1.1 and 1.2 cover respectively $M = 0$ and $M > 0$ .

LemmaStatement: AI-adaptedProof: AI-adaptedprecheck passOpen item page →

The volume far-field map is bounded and is the uniform far-field limit

Statement

Assume Countable Choice. Let $D, q, k$ be as in The three-dimensional volume integral scattering model, with $0 < ∣ D ∣ < \infty$ , and put $R_{0} > \sup_{y \in D} ∣ y ∣$ . Then $F_{k} : L^{2} (D; C) \to C (S^{2}; C)$ is linear and $∥ F_{k} v ∥_{\infty} \leq \frac{∥ q ∥_{\infty} \sqrt{∣ D ∣}}{4 π} ∥ v ∥_{2} .$ It is Lipschitz in direction with constant at most $∣ k ∣ R_{0} ∥ q ∥_{\infty} \sqrt{∣ D ∣} ∥ v ∥_{2} / (4 π)$ . For every $r \geq 2 R_{0}$ , the exterior volume potential of The three-dimensional volume integral scattering model satisfies $\sup_{θ \in S^{2}} ∣ r e^{- i k r} V_{k} v (r θ) - F_{k} v (θ) ∣ \leq \frac{2 R_{0} + ∣ k ∣ R_{0}^{2}}{4 π r} ∥ q ∥_{\infty} \sqrt{∣ D ∣} ∥ v ∥_{2} .$ Thus $F_{k} v$ is the uniform far-field amplitude under the outgoing $e^{i k r} / r$ convention.

Facts & Assumptions

Given: Countable Choice, $D, q, k$ as in the Statement, $v \in L^{2} (D; C)$ , and $R_{0} > \sup_{y \in D} ∣ y ∣$ .

[F1]

Cauchy–Schwarz gives $∥ q v ∥_{L^{1} (D)} \leq ∥ q ∥_{\infty} \sqrt{∣ D ∣} ∥ v ∥_{2}$ (The complex $L^{2}$ pairing is well-defined and satisfies Cauchy–Schwarz).

[F2]

For real $s, t$ , let $γ (r) = \exp (i (t + r (s - t)))$ on $[0, 1]$ . If $s = t$ , it is constant. Otherwise, with $z = i (t + r (s - t))$ , the complex derivative theorem The complex exponential is entire and its complex derivative is itself gives $\frac{γ (r + h) - γ (r)}{h} = i (s - t) \frac{\exp (z + i h (s - t)) - \exp z}{i h (s - t)} ⟶ i (s - t) \exp z .$ Thus $∣ γ^{'} (r) ∣ = ∣ s - t ∣$ by $\exp (x + i y) = e^{x} (\cos y + i \sin y)$ , $∣ \exp (x + i y) ∣ = e^{x}$ , and $e^{i π} + 1 = 0$ , and identifying $C$ with $R^{2}$ , If $f : [a, b] \to R^{m}$ is differentiable with integrable $f^{'}$ then $\int_{a}^{b} f^{'} = f (b) - f (a)$ ; and a bounded derivative makes $f$ Lipschitz gives $∣ e^{i s} - e^{i t} ∣ \leq ∣ s - t ∣$ .

[F3]

The integral convention for the exterior potential and the far-field map is fixed in The three-dimensional volume integral scattering model.

Proof

Given: The data in the Statement.

1.1F1F3algebra

By [F1], the defining integral of $F_{k} v$ is absolutely convergent and $∣ F_{k} v (θ) ∣ \leq (4 π)^{- 1} ∥ q v ∥_{1}$ . Taking the supremum proves the norm bound; linearity follows from linearity of the integral.

1.2F1F2F3

For $θ, θ^{'} \in S^{2}$ , [F2] gives $∣ e^{- i k θ \cdot y} - e^{- i k θ^{'} \cdot y} ∣ \leq ∣ k ∣ ∣ y ∣ ∣ θ - θ^{'} ∣$ . Integrating against $∣ q v ∣ / (4 π)$ and using [F1] proves the stated Lipschitz bound and hence continuity in direction.

1.3givenF2algebra

Fix $r \geq 2 R_{0}$ , $θ \in S^{2}$ , and $y \in D$ , and set $s : = ∣ r θ - y ∣$ . Then $s \geq r - R_{0} \geq r / 2$ , $∣ r - s ∣ \leq R_{0}$ , and $∣ s - (r - θ \cdot y) ∣ = \frac{∣ y ∣^{2} - (θ \cdot y)^{2}}{s + r - θ \cdot y} \leq \frac{R_{0}^{2}}{r},$ because the denominator is at least $(r - R_{0}) + (r - R_{0}) \geq r$ . Therefore $∣ r / s - 1 ∣ \leq 2 R_{0} / r$ and, by [F2], $∣ e^{i k (s - r)} - e^{- i k θ \cdot y} ∣ \leq ∣ k ∣ R_{0}^{2} / r$ .

2.1F1F3step 1.3algebra∎

The integrand identity $r e^{- i k r} G_{k} (r θ - y) = (4 π)^{- 1} (r / s) e^{i k (s - r)}$ and step 1.3 give a uniform pointwise error at most $(2 R_{0} + ∣ k ∣ R_{0}^{2}) / (4 π r)$ . Integrating against $∣ q v ∣$ , applying [F1], and taking the supremum over $θ$ proves the far-field estimate.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

Spherical harmonics form a complete basis with the addition formula

Statement

Assume Countable Choice and Dependent Choice. Use $H_{ℓ}$ from Spherical harmonics and partial-wave conventions in three dimensions. For every $ℓ \geq 0$ , $\dim_{C} H_{ℓ} = 2 ℓ + 1$ ; the spaces $H_{ℓ}$ are pairwise orthogonal in $L^{2} (S^{2}, d σ; C)$ and their algebraic sum is dense there. Thus they give a complete orthogonal spherical-harmonic expansion. For any orthonormal basis $Y_{ℓ 1}, \dots, Y_{ℓ, 2 ℓ + 1}$ of $H_{ℓ}$ , the basis-independent reproducing kernel $Z_{ℓ} (ω, η) : = \sum_{m = 1}^{2 ℓ + 1} Y_{ℓ m} (ω) \overline{Y_{ℓ m} (η)}$ satisfies $Z_{ℓ} (ω, ω) = (2 ℓ + 1) / (4 π)$ . In particular, $∣ g (ω) ∣ \leq \sqrt{\frac{2 ℓ + 1}{4 π}} ∥ g ∥_{L^{2} (S^{2})} (g \in H_{ℓ}) .$ For fixed $η$ , rotational invariance makes $Z_{ℓ} (ω, η)$ a degree- $ℓ$ zonal polynomial of $ω \cdot η$ ; normalizing $P_{ℓ} (1) = 1$ gives the usual addition identity $Z_{ℓ} (ω, η) = \frac{2 ℓ + 1}{4 π} P_{ℓ} (ω \cdot η)$ .

Facts & Assumptions

Given: Countable Choice and Dependent Choice, the sphere and its surface measure from Spherical harmonics and partial-wave conventions in three dimensions, and the finite-dimensional polynomial spaces used below.

[F1]

Homogeneous harmonic restrictions satisfy the polar Laplacian identity and use the polar surface measure (Spherical harmonics and partial-wave conventions in three dimensions, Polar coordinates decompose Lebesgue measure into r^{n-1} dr d sigma).

[F2]

Green's second identity holds on the unit ball for real $C^{2}$ functions under Countable Choice; its complex form follows by applying it to real and imaginary parts (Second Green identity).

[F3]

The uniform closure of a unital real function algebra is closed under pointwise maxima and minima (The uniform closure of a unital real function algebra is closed under absolute value, maximum, and minimum).

[F4]

A finite Radon measure on a locally compact Hausdorff space has $C_{c}$ dense in $L^{2}$ under Dependent Choice (C_c(X) is dense in L^p(mu) for a Radon measure). Every finite Borel measure on a second-countable locally compact Hausdorff space is Radon under Countable Choice (Radon measure on an LCH space, Locally finite Borel measures on second-countable LCH spaces are regular).

[F5]

Rank-nullity gives $\dim \ker Δ = \dim P_{ℓ} - \dim P_{ℓ - 2}$ once surjectivity of the polynomial Laplacian is proved locally (Rank-nullity: $\dim_{F} V = nullity T + rank T$ , Finite-dimensional vector space, and its dimension $\dim_{F} V$ ; infinite-dimensional means having no finite basis, Linear subspace of a vector space).

[F6]

The complex $L^{2}$ inner product satisfies Cauchy–Schwarz and has the norm induced by $d σ$ (Orthonormal families, complete orthonormal systems and Hilbert bases, The complex $L^{2}$ pairing is well-defined and satisfies Cauchy–Schwarz).

[F7]

Under Countable Choice, invertible linear changes of variables preserve Lebesgue measurability and scale measure by the absolute determinant (A linear map $T$ of $R^{n}$ sends Lebesgue measurable sets to Lebesgue measurable sets, with $λ_{n} (T [E]) = ∣ \det T ∣ λ_{n} (E)$ when $T$ is invertible and $T [E]$ Lebesgue null when it is not).

[F8]

A unital real algebra separating points interpolates any prescribed values at two distinct points (A unital separating real function algebra interpolates arbitrary values at two distinct points).

[F9]

$S^{2}$ is compact by Heine--Borel, hence locally compact. It is second countable: intersections with $S^{2}$ of rational-centred, rational-radius Euclidean balls form a countable base. Consequently the finite Borel surface measure is Radon by [F4].

[F10]

Every nonempty open $U \subseteq S^{2}$ has positive $σ$ -measure: the polar map is a homeomorphism, with inverse $(r, ω) \mapsto r ω$ , so it sends $(1 / 2, 3 / 4) \times U$ to a nonempty open subset of $R^{3}$ contained in the unit radial cone over $U$ . That cone has positive volume by containing a ball (A closed three-dimensional ball of radius $r \geq 0$ has volume $4 π r^{3} / 3$ ). The cone definition of $σ$ gives $σ (U) > 0$ . By [F1] and [F9], $S^{2}$ is compact and $σ$ is finite, so every continuous restriction belongs to $L^{2} (S^{2})$ . If it is zero as an $L^{2}$ class, it is therefore identically zero, so pointwise evaluation on $H_{ℓ} \subset L^{2} (S^{2})$ is well defined.

[A1]

Countable Choice is used in [F4] to show the finite Borel surface measure is Radon and in the cited proof of [F3]; Dependent Choice is used in the $C_{c}$ -density theorem. No Axiom of Choice is used for the finite-dimensional polynomial bases.

Proof

Given: The spaces $H_{ℓ}$ , surface measure $d σ$ , Countable Choice and Dependent Choice.

1.1F1givenalgebra

If $P_{ℓ} (x) = r^{ℓ} Y (ω)$ is homogeneous and harmonic, the polar Laplacian convention in [F1] gives $0 = Δ P_{ℓ} = r^{ℓ - 2} (ℓ (ℓ + 1) Y + Δ_{S^{2}} Y)$ . Hence $- Δ_{S^{2}} Y = ℓ (ℓ + 1) Y$ .

1.2F5givenalgebra

Let $P_{n}$ be the homogeneous complex polynomials of degree $n$ in three variables, with monomial basis of size $(\binom{n + 2}{2})$ . For homogeneous $R_{m}$ of degree $m$ , direct differentiation gives $Δ (∣ x ∣^{2 a} R_{m}) = 2 a (2 m + 2 a + 1) ∣ x ∣^{2 a - 2} R_{m} + ∣ x ∣^{2 a} Δ R_{m}$ . Given $S \in P_{n - 2}$ , put $J = ⌊ (n - 2) / 2 ⌋$ , $c_{j} = 2 (j + 1) (2 n - 2 j - 1)$ , $b_{0} = 1 / c_{0}$ , and $b_{j} = - b_{j - 1} / c_{j}$ for $1 \leq j \leq J$ . Then $Q = \sum_{j = 0}^{J} b_{j} ∣ x ∣^{2 (j + 1)} Δ^{j} S \in P_{n}$ satisfies $Δ Q = S$ : the coefficient of $∣ x ∣^{2 j} Δ^{j} S$ is $b_{j} c_{j} + b_{j - 1} = 0$ for $j \geq 1$ , the $S$ coefficient is $b_{0} c_{0} = 1$ , and $Δ^{J + 1} S = 0$ . Thus $Δ : P_{n} \to P_{n - 2}$ is onto. By [F5], its harmonic kernel has dimension $(\binom{n + 2}{2}) - (\binom{n}{2}) = 2 n + 1$ for $n \geq 2$ ; for $n = 0, 1$ all homogeneous polynomials are harmonic and have dimensions $1, 3$ . Restriction to $S^{2}$ is injective because homogeneity recovers $P (r ω) = r^{n} P (ω)$ , so $\dim H_{n} = 2 n + 1$ .

2.1F1F2step 1.1algebra

Let $P_{ℓ}, Q_{m}$ be homogeneous harmonic polynomials of degrees $ℓ, m$ . Apply [F2] on the unit ball to the real and imaginary components of $P_{ℓ}, \overline{Q_{m}}$ . Their volume Laplacians vanish, while homogeneity gives outward normal derivatives $\partial_{ν} P_{ℓ} = ℓ P_{ℓ}$ and $\partial_{ν} \overline{Q_{m}} = m \overline{Q_{m}}$ on $S^{2}$ . Therefore $(ℓ - m) \int_{S^{2}} P_{ℓ} (ω) \overline{Q_{m} (ω)} d σ (ω) = 0$ . If $ℓ \neq m$ , this proves orthogonality.

2.2step 1.2givenalgebra

For $P \in P_{n}$ , define $H = P + \sum_{j = 1}^{⌊ n / 2 ⌋} a_{j} ∣ x ∣^{2 j} Δ^{j} P$ , where $a_{0} = 1$ and $a_{j + 1} = - a_{j} / [2 (j + 1) (2 n - 2 j - 1)]$ . The differentiation identity in step 1.2 shows that the coefficients of each $∣ x ∣^{2 j} Δ^{j + 1} P$ cancel in $Δ H$ , so $H$ is harmonic. Since $P - H$ is divisible by $∣ x ∣^{2}$ , restriction to $S^{2}$ writes $P ∣_{S^{2}}$ as an element of $H_{n}$ plus the restriction of a homogeneous degree- $(n - 2)$ polynomial. Induction on degree decomposes every polynomial restriction into a finite sum of spherical harmonics.

2.3F1F6F7F9F10step 1.2algebra

Fix $ℓ$ . By [F10], the $L^{2}$ inner product is positive definite on the finite-dimensional space of continuous restrictions, so finite-dimensional Gram--Schmidt gives an orthonormal basis. The sum $Z_{ℓ}$ is independent of this basis because a unitary change of basis preserves the sum of rank-one coordinate kernels. For $R \in S O (3)$ and Borel $E \subseteq S^{2}$ , the polar-surface definition represents $σ (E)$ as three times the Lebesgue measure of its unit radial cone $C_{E} = {r ω : 0 < r \leq 1, ω \in E}$ . Since $R$ is a homeomorphism, $R E$ is Borel, $R C_{E} = C_{R E}$ , and $∣ \det R ∣ = 1$ ; [F7] therefore gives $σ (R E) = σ (E)$ . If $P$ is homogeneous and harmonic of degree $ℓ$ , then $P \circ R^{- 1}$ is homogeneous of the same degree and remains harmonic: termwise differentiation of monomials in the real and imaginary parts gives $\partial_{i} \partial_{i} (P \circ R^{- 1}) (x) = \sum_{a, b} (R^{- 1})_{a i} (R^{- 1})_{b i} (\partial_{a} \partial_{b} P) (R^{- 1} x)$ ; summing over $i$ and using $R^{- 1} (R^{- 1})^{T} = I$ proves $Δ (P \circ R^{- 1}) = (Δ P) \circ R^{- 1} = 0$ . Thus composition with $R^{- 1}$ is unitary on $L^{2} (S^{2}, d σ)$ and preserves $H_{ℓ}$ ; the rotated basis is again orthonormal, so basis independence gives $Z_{ℓ} (R ω, R η) = Z_{ℓ} (ω, η)$ . In particular $Z_{ℓ} (ω, ω)$ is constant because rotations act transitively on $S^{2}$ . Integrating the diagonal gives $\int_{S^{2}} Z_{ℓ} (ω, ω) d σ = 2 ℓ + 1$ . Polar coordinates applied to the unit ball and its volume $4 π / 3$ (A closed three-dimensional ball of radius $r \geq 0$ has volume $4 π r^{3} / 3$ ) give $σ (S^{2}) = 4 π$ , so the diagonal constant is $(2 ℓ + 1) / (4 π)$ . Cauchy–Schwarz for the finite coefficient vector yields the displayed evaluation bound.

3.1A1F3F4F8F9step 2.1step 2.2

Let $A$ be the restrictions of real polynomials to $S^{2}$ ; constants, sums, and products show that it is a unital real algebra, and coordinate functions separate distinct points. Fix $f \in C (S^{2}; R)$ and $ε > 0$ . For each fixed $η \in S^{2}$ , form the family of open sets $U_{p} : = {ω : p (ω) > f (ω) - ε / 2}$ for all $p \in A$ with $p (η) = f (η)$ . This family covers $S^{2}$ : for each $ω$ , [F8] supplies an interpolant $p$ with $p (η) = f (η)$ and $p (ω) = f (ω)$ , using the constant polynomial when $ω = η$ . By [F9], compactness gives a finite subcover corresponding to $p_{1}, \dots, p_{s}$ . Their maximum $g_{η} \in \overline{A}$ by [F3] satisfies $g_{η} (η) = f (η)$ and $g_{η} > f - ε / 2$ everywhere. This proves the existence of such a function for every $η$ without choosing a family indexed by $S^{2}$ . Define the set $G : = {g \in \overline{A} : g > f - ε / 2 on S^{2}}$ and the open sets $V_{g} : = {ω : g (ω) < f (ω) + ε / 2}$ for $g \in G$ . The functions whose existence was just proved show that these sets cover $S^{2}$ , so [F9] gives a finite subcover $V_{g_{1}}, \dots, V_{g_{t}}$ . The minimum $h : = \min_{i} g_{i}$ lies in $\overline{A}$ by [F3] and satisfies $f - ε / 2 < h < f + ε / 2$ everywhere. Since $ε$ was arbitrary, $f \in \overline{A}$ and the real polynomial restrictions are dense in $C (S^{2}; R)$ . Applying this to real and imaginary parts gives density of complex polynomial restrictions in $C (S^{2}; C)$ . By [F4] and [F9], $d σ$ is Radon and $C (S^{2}; C) = C_{c} (S^{2}; C)$ is dense in $L^{2} (S^{2}, d σ; C)$ . Step 2.2 therefore implies that the span of the mutually orthogonal spaces $H_{ℓ}$ is dense in $L^{2}$ .

4.1givenstep 1.1step 2.3algebra∎

For fixed $η$ , rotations fixing $η$ leave $Z_{ℓ} (ω, η)$ invariant. Writing an invariant polynomial in coordinates parallel and perpendicular to $η$ shows it is a polynomial in $t = ω \cdot η$ on $S^{2}$ . Since it belongs to $H_{ℓ}$ , the spherical-Laplacian formula in the Definition gives $(1 - t^{2}) p^{''} - 2 t p^{'} + ℓ (ℓ + 1) p = 0$ for this polynomial $p$ . If $p (t) = \sum_{j = 0}^{ℓ} c_{j} t^{j}$ , coefficient comparison gives $(j + 2) (j + 1) c_{j + 2} + [ℓ (ℓ + 1) - j (j + 1)] c_{j} = 0$ for $0 \leq j < ℓ$ ; downward recursion determines every coefficient from $c_{ℓ}$ , and a nonzero solution cannot have degree $d < ℓ$ because its leading coefficient would be multiplied by $ℓ (ℓ + 1) - d (d + 1) \neq 0$ . Thus polynomial solutions of degree at most $ℓ$ form a one-dimensional space of degree exactly $ℓ$ . By step 2.3, $p (1) = Z_{ℓ} (η, η) = (2 ℓ + 1) / (4 π) > 0$ . Consequently $P_{ℓ} (t) : = 4 π p (t) / (2 ℓ + 1)$ is the unique degree- $ℓ$ solution with $P_{ℓ} (1) = 1$ from the Definition, and $Z_{ℓ} (ω, η) = \frac{2 ℓ + 1}{4 π} P_{ℓ} (ω \cdot η)$ .

LemmaStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-6-luna)Open item page →

The outgoing three-dimensional Helmholtz kernel is a fundamental solution

Statement

Assume Countable Choice. For real $k$ , let $Φ_{k} (x) = e^{i k ∣ x ∣} / (4 π ∣ x ∣)$ for $x \neq 0$ . In the sense of distributions on $R^{3}$ , $(- Δ - k^{2}) Φ_{k} = δ_{0} .$ Here $δ_{0}$ is the point-mass distribution defined by $⟨ δ_{0}, φ ⟩ = φ (0)$ for every $φ \in C_{c}^{\infty} (R^{3})$ . Consequently, if $g \in L^{1} (R^{3})$ has bounded support, the integral $\int_{R^{3}} Φ_{k} (x - y) g (y) d y$ is absolutely finite for almost every $x$ . Define $V$ by this integral almost everywhere (and arbitrarily on the exceptional null set). Then $V \in L_{l o c}^{1}$ and $(- Δ - k^{2}) V = g$ distributionally.

Facts & Assumptions

Given: Countable Choice, real $k$ , the outgoing kernel convention in The three-dimensional volume integral scattering model, and a compactly supported $g \in L^{1} (R^{3})$ .

[F1]

The three-dimensional volume integral scattering model: $Φ_{k} (r) = e^{i k r} / (4 π r)$ for $r > 0$ .

[F2]

Polar coordinates decompose Lebesgue measure into r^{n-1} dr d sigma: radial integration in $R^{3}$ uses $r^{2} d r d σ$ with $σ (S^{2}) = 4 π$ .

[F3]

Second Green identity: assuming Countable Choice, Green's second identity applies on the smooth annulus to real $C^{2}$ functions; its complex-bilinear form follows by splitting both functions into real and imaginary parts.

[F4]

Fubini's theorem for L^1 functions on a sigma-finite product and Tonelli and Fubini for the completed product, with only almost-everywhere section measurability: absolutely integrable product functions may be integrated in either order, with the stated almost-everywhere section interpretation.

[F5]

Test function space d of an open set, Test function topology, Fixed support test function frechet space, Distribution, and Distributional derivative: $D (R^{3}) = C_{c}^{\infty}$ ; a distribution is a continuous complex-linear functional on this test space, and derivatives are defined by integration by parts against tests. The seminorm $p_{0} (φ) = \sup_{x} ∣ φ (x) ∣$ is continuous on each fixed-support test space, so point evaluation $φ \mapsto φ (0)$ is continuous.

[F6]

A locally integrable function defines a regular distribution, and under Countable Choice the embedding is continuous (Regular distribution from a locally integrable function, Locally integrable functions embed in distributions, The Axiom of Countable Choice ( ${A C}_{ω}$ )).

[F8]

A countable union of measurable null sets is null (Finite and countable subadditivity of measures).

[F7]

The complex exponential has derivative $\exp^{'} (z) = \exp z$ (The complex exponential is entire and its complex derivative is itself), and for real $t$ it satisfies $∣ e^{i t} ∣ = 1$ ( $\exp (x + i y) = e^{x} (\cos y + i \sin y)$ , $∣ \exp (x + i y) ∣ = e^{x}$ , and $e^{i π} + 1 = 0$ ). For real $r$ and $k \neq 0$ , its real-parameter derivative follows directly from the complex difference quotient: $\frac{e^{i k (r + h)} - e^{i k r}}{h} = i k \frac{e^{i k r + i k h} - e^{i k r}}{i k h} ⟶ i k e^{i k r} .$ For $k = 0$ the function is constant. The defining exponential series also gives $e^{i k ϵ} = 1 + i k ϵ + O (ϵ^{2})$ as $ϵ ↓ 0$ (The complex exponential by its power series).

Proof

Given: The data in the Statement and an arbitrary $φ \in C_{c}^{\infty} (R^{3})$ .

1.1F1F2F5F6F7algebra

Writing $r = ∣ x ∣$ , [F7] gives $Φ_{k}^{'} (r) = e^{i k r} (i k r - 1) / (4 π r^{2})$ ; direct substitution yields $Φ_{k}^{''} + 2 Φ_{k}^{'} / r + k^{2} Φ_{k} = 0$ for $r > 0$ . Also [F7] gives $∣ Φ_{k} (x) ∣ = 1 / (4 π ∣ x ∣)$ , whose integral over each centered ball is finite by [F2]. Hence $Φ_{k} \in L_{l o c}^{1}$ and [F6] supplies its regular distribution. Point evaluation is continuous by [F5], so $δ_{0}$ is also a distribution. Thus $(- Δ - k^{2}) Φ_{k} = 0$ away from the origin as a classical equation and the distributional pairing in the next step is defined.

2.1F2F3F5F6F7step 1.1algebra

Choose an outer sphere beyond $supp φ$ and apply [F3] on the annulus between it and radius $ϵ$ . Apply the real identity to the four real/imaginary pairings to obtain the complex bilinear identity. The outer boundary term is zero, and the inner boundary orientation yields $\int_{∣ x ∣ > ϵ} Φ_{k} (x) (- Δ - k^{2}) φ (x) d x = \int_{∣ x ∣ = ϵ} (Φ_{k} \partial_{r} φ - φ \partial_{r} Φ_{k}) d S$ . The first term is $O (ϵ)$ since $Φ_{k} = O (ϵ^{- 1})$ and the sphere area is $4 π ϵ^{2}$ . From the derivative formula in step 1.1 and the exponential series, $e^{i k ϵ} (i k ϵ - 1) = - 1 + O (ϵ^{2})$ , so $\partial_{r} Φ_{k} = - 1 / (4 π ϵ^{2}) + O (1)$ . The second term therefore tends to $φ (0)$ : after writing $x = ϵ ω$ , it equals $\int_{S^{2}} φ (ϵ ω) / (4 π) d σ + O (ϵ^{2})$ . The omitted ball contributes $o (1)$ because $∣ Φ_{k} ∣ = O (1 / r)$ is locally integrable by [F2]. By [F6], this limit is precisely $⟨ (- Δ - k^{2}) Φ_{k}, φ ⟩ = ⟨ δ_{0}, φ ⟩$ .

3.1F2F3F4F8step 2.1algebra∎

Let $A$ be any bounded measurable set and let $C$ be a bounded set containing the support of $g$ . Polar coordinates show $\sup_{y \in C} \int_{A} ∣ Φ_{k} (x - y) ∣ d x < \infty$ , since $A - y$ lies in a fixed-radius ball and $\int_{0}^{R} r^{- 1} r^{2} d r < \infty$ . Tonelli then gives $\int_{A} \int_{C} ∣ Φ_{k} (x - y) ∣ ∣ g (y) ∣ d y d x < \infty$ , so the defining convolution is absolutely finite for almost every $x \in A$ and belongs to $L^{1} (A)$ . Taking $A = B (0, n)$ for each integer $n \geq 1$ and using [F8] on the countable union of exceptional null sets proves absolute finiteness for almost every $x \in R^{3}$ . For every bounded measurable $A$ , the same estimate gives $V \in L^{1} (A)$ , hence $V \in L_{l o c}^{1}$ . With $A = supp φ$ , the same bound makes $g (y) Φ_{k} (x - y) (- Δ - k^{2}) φ (x)$ absolutely integrable, so [F4] permits Fubini. Step 2.1 translated by $y$ gives $⟨ (- Δ - k^{2}) V, φ ⟩ = \int g (y) φ (y) d y$ . The preceding argument proves $V \in L_{l o c}^{1}$ and the a.e. existence of its integral representative.

LemmaStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-6-luna)Open item page →

The volume integral operator is well defined and bounded on L2

Statement

Assume Countable Choice. Let $D$ have finite Lebesgue measure. For every product-measurable kernel $a \in L^{2} (D \times D; C)$ , the formula $T_{a} v (x) = \int_{D} a (x, y) v (y) d y$ defines a representative-independent complex-linear map $L^{2} (D; C) \to L^{2} (D; C)$ and $∥ T_{a} ∥ \leq ∥ a ∥_{L^{2} (D \times D)}$ . Consequently, under The three-dimensional volume integral scattering model with $0 < ∣ D ∣ < \infty$ , $∥ K_{k} ∥_{L^{2} \to L^{2}} \leq ∥ G_{k} (\cdot - \cdot) q (\cdot) ∥_{L^{2} (D \times D)} \leq ∥ q ∥_{\infty} \sqrt{\frac{∣ D ∣ diam (D)}{4 π}} .$ For $∣ D ∣ = 0$ the operator is zero.

Facts & Assumptions

Given: Countable Choice, a finite-measure $D$ , a square-integrable product-measurable kernel $a$ , and, for the scattering bound, $q, k$ from The three-dimensional volume integral scattering model.

[F1]

The kernel $a (x, y) = G_{k} (x - y) q (y)$ is a product $L^{2}$ class with the displayed bound (The three-dimensional free-space scattering kernel is square integrable).

[F2]

Complex $L^{2}$ Cauchy–Schwarz is valid and the pairing is representative-independent (The complex $L^{2}$ pairing is well-defined and satisfies Cauchy–Schwarz).

[F3]

A product $L^{1}$ function has integrable, measurable sections almost everywhere, and its section integral is measurable (Tonelli and Fubini for the completed product, with only almost-everywhere section measurability, Fubini's theorem for L^1 functions on a sigma-finite product).

Proof

Given: The data above and $u \in L^{2} (D; C)$ .

1.1F1F2F3given

On the product space, Cauchy–Schwarz gives $\int_{D \times D} ∣ a (x, y) u (y) ∣ d x d y \leq ∥ a ∥_{2} (∣ D ∣ \int_{D} ∣ u (y) ∣^{2} d y)^{1 / 2} < \infty$ . Hence the section integral defining $T_{a} u$ exists absolutely for almost every $x$ and is measurable by F3.

2.1F3step 1.1algebra

If $u = u^{'}$ almost everywhere, then $a (x, y) (u - u^{'}) (y) = 0$ for product-almost every $(x, y)$ by Tonelli on $D \times {u \neq u^{'}}$ ; F3 makes their section integrals equal for almost every $x$ . Changing $a$ on a product-null set likewise changes only a.e.-zero sections. Thus the output class depends only on the kernel and input classes. Integral linearity on $L^{1}$ proves complex linearity.

3.1F1F2F3step 2.1algebra

For almost every $x$ , Cauchy–Schwarz in $y$ yields $∣ T_{a} u (x) ∣^{2} \leq (\int_{D} ∣ a (x, y) ∣^{2} d y) ∥ u ∥_{2}^{2}$ . Integrating this inequality in $x$ with Tonelli gives $∥ T_{a} u ∥_{2}^{2} \leq ∥ a ∥_{2}^{2} ∥ u ∥_{2}^{2}$ . Step 2.1 and this estimate prove the general assertion.

4.1F1step 3.1algebra∎

Apply step 3.1 to $a (x, y) = G_{k} (x - y) q (y)$ and then use The three-dimensional free-space scattering kernel is square integrable for its explicit square norm. If $∣ D ∣ = 0$ , both domain and codomain are the zero $L^{2}$ space.

TheoremStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-6-luna)Open item page →

Exponential coefficient decay gives a uniform partial-wave truncation bound

Statement

Assume Countable Choice and Dependent Choice. Let $b_{ℓ} \in H_{ℓ}$ for every integer $ℓ \geq 0$ , where $H_{ℓ}$ is defined in Spherical harmonics and partial-wave conventions in three dimensions. Suppose constants $C \geq 0$ and $ρ > 1$ satisfy $∥ b_{ℓ} ∥_{L^{2} (S^{2})} \leq C ρ^{- ℓ}$ for all $ℓ$ . Then $b (ω) : = \sum_{ℓ = 0}^{\infty} b_{ℓ} (ω)$ converges uniformly on $S^{2}$ , and for every integer $L \geq 0$ , $\sup_{ω \in S^{2}} ∣ b (ω) - \sum_{ℓ = 0}^{L} b_{ℓ} (ω) ∣ \leq C \sqrt{\frac{2 L + 3}{4 π}} \frac{ρ^{- (L + 1)}}{(1 - ρ^{- 1})^{2}} .$ This is a conditional angular-coefficient bound; it does not assert exponential decay for general scattering data.

Facts & Assumptions

Given: The harmonic spaces, constants $C \geq 0$ , $ρ > 1$ , and coefficient bounds above.

[F1]

Spherical harmonics form a complete basis with the addition formula: for $g \in H_{ℓ}$ , $∥ g ∥_{\infty} \leq \sqrt{(2 ℓ + 1) / (4 π)} ∥ g ∥_{2}$ .

Proof

Given: The data in the Statement.

1.1F1givenalgebra

By [F1], $∥ b_{ℓ} ∥_{\infty} \leq C \sqrt{(2 ℓ + 1) / (4 π)} ρ^{- ℓ}$ . The scalar series of these bounds converges because $ρ > 1$ , so the partial sums of $\sum b_{ℓ}$ are uniformly Cauchy and converge uniformly to a function $b$ .

2.1givenstep 1.1algebra∎

For $ℓ = L + 1 + n$ with $n \geq 0$ , $\sqrt{2 ℓ + 1} \leq (n + 1) \sqrt{2 L + 3}$ : squaring reduces this to $2 L + 3 + 2 n \leq (2 L + 3) (n + 1)^{2}$ , whose difference is nonnegative. Therefore the tail in step 1.1 is at most $C \sqrt{(2 L + 3) / (4 π)} ρ^{- (L + 1)} \sum_{n = 0}^{\infty} (n + 1) ρ^{- n}$ . For $t = ρ^{- 1} \in (0, 1)$ , multiplying the geometric series by itself or differentiating its partial sums gives $\sum_{n \geq 0} (n + 1) t^{n} = (1 - t)^{- 2}$ . This yields the claimed bound.

TheoremStatement: AI-adaptedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

The volume integral scattering operator is compact

Statement

Assume AC and the data of The three-dimensional volume integral scattering model with $0 < ∣ D ∣ < \infty$ . Then $K_{k} : L^{2} (D; C) \to L^{2} (D; C)$ is compact and $∥ K_{k} ∥ \leq ∥ G_{k} (x - y) q (y) ∥_{L^{2} (D \times D)} \leq ∥ q ∥_{\infty} \sqrt{\frac{∣ D ∣ diam (D)}{4 π}} .$

The proof supplies both the finite-measure Hilbert–Schmidt rescaling route required by the kernel interface and the singular-truncation/Schur route specified for this volume kernel. No uniqueness or outgoing-scattering solvability is asserted here.

Facts & Assumptions

Given: AC, $D, q, k$ as in The three-dimensional volume integral scattering model, the integral operator $K_{k}$ , and $0 < ∣ D ∣ < \infty$ .

[F1]

The kernel $a (x, y) = G_{k} (x - y) q (y)$ is square integrable with $∥ a ∥_{2}^{2} \leq ∥ q ∥_{\infty}^{2} ∣ D ∣ diam (D) / (4 π)$ (The three-dimensional free-space scattering kernel is square integrable).

[F2]

On a finite source and target measure space, $T_{a}$ is unitarily equivalent to the probability-space kernel operator with kernel $\sqrt{M_{X} M_{Y}} a$ ; for $D$ this is $M a$ (Rescaling finite-measure kernel operators to probability spaces).

[F3]

Under AC, on a probability space a square-integrable kernel defines a bounded compact operator, with operator norm at most its kernel $L^{2}$ norm (Square integrable kernels define bounded compact integral operators).

[F4]

If $h (x, y)$ is a measurable kernel with $\sup_{x} \int ∣ h (x, y) ∣ d y \leq A$ and $\sup_{y} \int ∣ h (x, y) ∣ d x \leq B$ , then its integral operator has norm at most $\sqrt{A B}$ ; the weighted Cauchy–Schwarz/Tonelli proof is given in step 1.2.

[F5]

Under Countable Choice, $C_{c}^{\infty} (R^{3}; C)$ is dense in Euclidean $L^{2}$ (Complex finite-simple and smooth compact-support density for finite p).

[F6]

A uniformly bounded, equicontinuous family of real functions on a compact metric space has compact closure in the sup norm, under Countable Choice and Dependent Choice (Arzelà--Ascoli for real $C (K)$ under Countable Choice and Dependent Choice: compact closure iff equicontinuous and pointwise bounded).

[F7]

A norm limit of compact operators into a Banach space is compact under Countable Choice (Norm limit of compact operators is compact); AC supplies Countable Choice and Dependent Choice (AC supplies the countable and dependent choices used in Banach integration).

[F8]

Any product $L^{2}$ kernel $h$ defines a bounded operator with norm at most $∥ h ∥_{2}$ (The volume integral operator is well defined and bounded on L2).

[F9]

Polar coordinates give $\int_{B (x, ρ)} ∣ x - y ∣^{- 2} d y = 4 π ρ$ and the unit sphere area is $4 π$ (Polar coordinates decompose Lebesgue measure into r^{n-1} dr d sigma, A closed three-dimensional ball of radius $r \geq 0$ has volume $4 π r^{3} / 3$ ).

[F10]

The closure $\overline{D}$ is compact by Heine–Borel and $\overline{D} \times \overline{D}$ is compact as a finite product (Heine-Borel in $R^{n}$ : with the Euclidean metric a subset of $R^{n}$ is compact if and only if it is closed and bounded, and the proof by bisection uses no choice principle; the same holds on the real line, A product of finitely many compact spaces is compact in the product topology).

[F11]

Complex $L^{2} (D; C)$ is complete under Countable Choice (Complex Lp completeness and almost-everywhere subsequences), and AC supplies Countable Choice by AC supplies the countable and dependent choices used in Banach integration.

[F12]

Applying The essential supremum is attained as the least essential bound to $∣ q ∣$ gives $∣ q ∣ \leq ∥ q ∥_{\infty}$ almost everywhere; changing a representative on that measurable null set gives the everywhere-bounded $q_{0}$ used in the Schur estimate.

[F13]

Cauchy--Schwarz on the finite-measure set $D$ gives $\int_{D} ∣ v ∣ \leq \sqrt{∣ D ∣} ∥ v ∥_{2}$ (The complex $L^{2}$ pairing is well-defined and satisfies Cauchy–Schwarz).

Proof

Given: AC and the scattering data in the Statement.

1.1givenF1F2F3F11

By [F11] and AC, $X = L^{2} (D; C)$ is a Banach space. Normalize $D$ to the probability measure $d ν = d x / M$ . By [F2], the operator $K_{k}$ is unitarily equivalent to the integral operator on $L^{2} (D, ν)$ with kernel $\tilde{a} = M a$ . Its kernel norm equals $∥ a ∥_{L^{2} (D \times D, d x d y)}$ by [F2]. Applying [F3] to this probability-space kernel gives compactness of the normalized operator and its norm bound; unitary equivalence transfers both to $K_{k}$ .

1.2givenF1F9F12algebra

We also verify the singular-truncation route. By [F12], choose a measurable representative $q_{0}$ and set it to zero on the measurable null set where $∣ q_{0} ∣ > ∥ q ∥_{\infty}$ . This does not change the operator, since the kernel changes only on a product-null set, and now $∣ q_{0} (y) ∣ \leq ∥ q ∥_{\infty}$ for every $y$ . For $ε > 0$ , put $χ_{ε} (r) = 0$ for $r \leq ε$ , $χ_{ε} (r) = (r - ε) / ε$ for $ε < r < 2 ε$ , and $χ_{ε} (r) = 1$ for $r \geq 2 ε$ . Let $K_{ε}$ have kernel $χ_{ε} (∣ x - y ∣) G_{k} (x - y) q_{0} (y)$ . The difference kernel is supported where $∣ x - y ∣ < 2 ε$ and bounded in modulus by $∥ q ∥_{\infty} / (4 π ∣ x - y ∣)$ . By [F9] its row and column integrals, for every fixed row and column, are at most $A = B \leq (∥ q ∥_{\infty} / (4 π)) 4 π \int_{0}^{2 ε} r d r = 2 ∥ q ∥_{\infty} ε^{2}$ . For each $x$ , weighted Cauchy–Schwarz gives $∣ T_{h} v (x) ∣^{2} \leq (\int ∣ h (x, y) ∣ d y) (\int ∣ h (x, y) ∣ ∣ v (y) ∣^{2} d y)$ ; integrating and using Tonelli yields $∥ T_{h} v ∥_{2}^{2} \leq A B ∥ v ∥_{2}^{2}$ . Hence $∥ K - K_{ε} ∥ \leq 2 ∥ q ∥_{\infty} ε^{2}$ .

1.3givenF5F6F10F13construct

Fix $ε > 0$ . Extend the everywhere-bounded representative $q_{0}$ by zero to $R^{3}$ ; it lies in $L^{2} (R^{3})$ because $q \in L^{\infty} (D)$ and $∣ D ∣ < \infty$ . By [F5], choose $q_{j} \in C_{c}^{\infty} (R^{3}; C)$ with $∥ q_{j} - q_{0} ∥_{L^{2} (D)} \to 0$ . The kernel $g_{ε, j} (x, y) = χ_{ε} (∣ x - y ∣) G_{k} (x - y) q_{j} (y)$ extends continuously to the compact set $\overline{D} \times \overline{D}$ by assigning zero on $x = y$ ; the cutoff vanishes near the diagonal. For $∥ v ∥_{2} \leq 1$ , the integral is independent of the representative of $v$ , and by [F13] its outputs obey $∣ T_{ε, j} v (x) ∣ \leq \sqrt{∣ D ∣} ∥ g_{ε, j} ∥_{\infty}$ and $∣ T_{ε, j} v (x) - T_{ε, j} v (x^{'}) ∣ \leq \sqrt{∣ D ∣} \sup_{y} ∣ g_{ε, j} (x, y) - g_{ε, j} (x^{'}, y) ∣$ . Uniform continuity of $g_{ε, j}$ makes these outputs uniformly bounded and equicontinuous. Applying [F6] separately to real and imaginary parts gives compact closures $C_{R}, C_{I}$ in $C (\overline{D}; R)$ . The product $C_{R} \times C_{I}$ is compact by finite-product compactness. Under the homeomorphism $w \mapsto (Re w, Im w)$ , the closure of the complex output family is a closed subset of this compact product, hence compact in $C (\overline{D}; C)$ . The inclusion is continuous because $∥ w ∣_{D} ∥_{2} \leq \sqrt{∣ D ∣} ∥ w ∥_{C (\overline{D})}$ ; hence $T_{ε, j} : L^{2} (D) \to L^{2} (D)$ is compact.

2.1givenF5F7F8step 1.3algebra

Since $∣ χ_{ε} G_{k} ∣ \leq (4 π ε)^{- 1}$ , the square norm of the kernel difference between $K_{ε}$ and $T_{ε, j}$ is at most $∣ D ∣ (16 π^{2} ε^{2})^{- 1} ∥ q - q_{j} ∥_{L^{2} (D)}^{2}$ . [F8] therefore gives $∥ K_{ε} - T_{ε, j} ∥ \to 0$ . By [F7], $K_{ε}$ is compact.

3.1F1F2F3F7step 1.1step 1.2step 2.1algebra∎

As $ε ↓ 0$ , step 1.2 gives $∥ K - K_{ε} ∥ \leq 2 ∥ q ∥_{\infty} ε^{2} \to 0$ . Applying [F7] again shows $K$ is compact. Step 1.1 independently gives the displayed norm estimate and compactness with the finite-measure Hilbert–Schmidt normalization.

TheoremStatement: AI-adaptedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

Born truncation error for a small-norm volume integral operator

Statement

Assume Countable Choice and the data of The three-dimensional volume integral scattering model with $0 < ∣ D ∣ < \infty$ . If $ρ : = ∥ K_{k} ∥_{L^{2} \to L^{2}} < 1$ , then for every $f \in L^{2} (D; C)$ the equation $u = f + K_{k} u$ has a unique solution. For each integer $m \geq 0$ , define the Born partial sum $u_{m} : = \sum_{j = 0}^{m} K_{k}^{j} f$ . Then $∥ u - u_{m} ∥_{2} \leq \frac{ρ^{m + 1}}{1 - ρ} ∥ f ∥_{2} .$ No compactness, analytic continuation, or physical uniqueness hypothesis is used in this small-norm branch.

Facts & Assumptions

Given: Countable Choice, $f \in L^{2} (D; C)$ , and $K_{k}$ with operator norm $ρ < 1$ .

[F1]

$K_{k}$ is a bounded linear operator on complex $L^{2} (D)$ (The volume integral operator is well defined and bounded on L2).

[F2]

Complex $L^{2} (D)$ is complete under Countable Choice (Complex Lp completeness and almost-everywhere subsequences, The Axiom of Countable Choice ( ${A C}_{ω}$ )).

[F3]

For every integer $j \geq 0$ , $∥ K_{k}^{j} f ∥_{2} \leq ρ^{j} ∥ f ∥_{2}$ , by repeated application of the operator norm inequality in [F1].

Proof

Given: The data of the Statement, with $ρ = ∥ K_{k} ∥ < 1$ .

1.1givenF2F3algebra

Define $u_{N} : = \sum_{j = 0}^{N} K_{k}^{j} f$ . For $M > N$ , the triangle inequality and [F3] give $∥ u_{M} - u_{N} ∥_{2} \leq ∥ f ∥_{2} \sum_{j = N + 1}^{M} ρ^{j} \leq ∥ f ∥_{2} ρ^{N + 1} / (1 - ρ)$ . This tends to zero as $N \to \infty$ , so $(u_{N})$ is Cauchy and converges to some $u \in L^{2} (D; C)$ by [F2].

2.1F1F3step 1.1algebra

Finite telescoping gives $(I - K_{k}) u_{N} = f - K_{k}^{N + 1} f$ . By [F3], $∥ K_{k}^{N + 1} f ∥_{2} \leq ρ^{N + 1} ∥ f ∥_{2} \to 0$ ; boundedness in [F1] lets $K_{k} u_{N} \to K_{k} u$ . Taking limits yields $(I - K_{k}) u = f$ . If $(I - K_{k}) v = 0$ , then $∥ v ∥_{2} = ∥ K_{k} v ∥_{2} \leq ρ ∥ v ∥_{2}$ , and $ρ < 1$ forces $v = 0$ , so this solution is unique.

3.1F2F3step 1.1algebra∎

With $m$ fixed, [F3] and the geometric-series identity give $∥ u - u_{m} ∥_{2} \leq \sum_{j = m + 1}^{\infty} ∥ K_{k}^{j} f ∥_{2} \leq ∥ f ∥_{2} \sum_{j = m + 1}^{\infty} ρ^{j} = ρ^{m + 1} ∥ f ∥_{2} / (1 - ρ)$ . The argument includes $ρ = 0$ and $f = 0$ .

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-6-luna)Open item page →

Cell-volume product integration for the scattering kernel

Definition

Assume Countable Choice. Let $D, q, k$ be as in The three-dimensional volume integral scattering model, let $f \in L^{2} (D; C)$ , and assume $0 < ∣ D ∣ < \infty$ . Let $P_{N}$ be the dyadic cell-average projection of Dyadic cell-average projections converge strongly in L2 of a bounded set, and put $K_{N} : = P_{N} K_{k}$ and $f_{N} : = P_{N} f$ . This is a finite-rank volume product-integration rule on $L^{2} (D)$ ; the projection and kernel convention are fixed by those definitions.

For each positive-measure cell $E_{N, j}$ set $ϕ_{N, j} : = ∣ E_{N, j} ∣^{- 1 / 2} 1_{E_{N, j}}$ . The exact product-cell weights and right-hand side are $A_{i j}^{(N)} : = ⟨ K_{k} ϕ_{N, j}, ϕ_{N, i} ⟩ = \frac{1}{\sqrt{∣ E_{N, i} ∣ ∣ E_{N, j} ∣}} \int_{E_{N, i}} \int_{E_{N, j}} G_{k} (x - y) q (y) d y d x, b_{i}^{(N)} : = \frac{1}{\sqrt{∣ E_{N, i} ∣}} \int_{E_{N, i}} f (x) d x .$ For each cell pair, the double integral is absolutely convergent: the kernel $G_{k} (x - y) q (y)$ is in $L^{2} (D \times D)$ and $ϕ_{N, i} (x) ϕ_{N, j} (y)$ has product $L^{2}$ norm one, so Cauchy--Schwarz makes their product integrable. Completed-product Fubini under Countable Choice then gives the displayed equality with the inner product. The finite system is $(I - A^{(N)}) c = b^{(N)}$ , and its reconstruction is $u_{N} = \sum_{j} c_{j} ϕ_{N, j}$ . Equivalently, it solves $(I - K_{N}) u_{N} = f_{N}$ in $V_{N}$ . Its far-field output is the exact cell integral $F_{N} (c) (θ) : = F_{k} [u_{N}] (θ) = \sum_{j} \frac{c_{j}}{4 π \sqrt{∣ E_{N, j} ∣}} \int_{E_{N, j}} e^{- i k θ \cdot y} q (y) d y .$ A singular finite matrix has status singular; the definition makes no unconditional invertibility claim. The double integrals are taken over whole cell pairs, including diagonal pairs; no point value of $G_{k}$ at its singular diagonal is sampled. Finiteness is established by The volume integral operator is well defined and bounded on L2 and The volume far-field map is bounded and is the uniform far-field limit.

The rule uses exact cell weights as a mathematical discretization. This definition makes no claim about floating-point evaluation or approximate cell weights.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passOpen item page →

Fredholm solvability and residual stability for volume scattering

Statement

Assume AC and the hypotheses of The three-dimensional volume integral scattering model with $0 < ∣ D ∣ < \infty$ . If $I - K_{k}$ is injective, then for every $f \in L^{2} (D; C)$ there is a unique $u \in L^{2} (D; C)$ solving $(I - K_{k}) u = f$ , and $A_{k} : = I - K_{k}$ has a bounded inverse. Writing $R_{k} : = ∥ A_{k}^{- 1} ∥$ , every $v \in L^{2} (D; C)$ satisfies $∥ u - v ∥_{2} \leq R_{k} ∥ f - (I - K_{k}) v ∥_{2} .$ This is an operator-theoretic Fredholm branch; no physical uniqueness statement for the exterior PDE is asserted.

Facts & Assumptions

Given: AC, $D, q, k$ , compact $K_{k}$ , and injectivity of $I - K_{k}$ .

[A1]

AC supplies Countable Choice; complex $L^{2} (D)$ is complete under Countable Choice (The Axiom of Choice, The Axiom of Countable Choice ( ${A C}_{ω}$ ), Complex Lp completeness and almost-everywhere subsequences). Thus it is a Banach space (Banach space).

[F1]

The operator $K_{k}$ is compact (The volume integral scattering operator is compact).

[F2]

On a Banach space, if $K$ is compact and $I - K$ is injective, then $I - K$ is surjective with bounded inverse (Fredholm alternative for identity minus compact); this result explicitly assumes AC.

[F3]

A compact linear operator is bounded, so $A_{k} = I - K_{k}$ is bounded (Compact linear operator, A bounded linear operator between normed spaces).

Proof

Given: AC and the data in the Statement.

1.1givenA1F1F2

By [A1], $L^{2} (D; C)$ is Banach. By [F1] $K_{k}$ is compact, and the assumed injectivity permits [F2] to be applied to $A_{k} = I - K_{k}$ . Therefore $A_{k}$ is surjective and has a bounded inverse; for each $f$ the unique solution is $u = A_{k}^{- 1} f$ .

2.1givenF3step 1.1algebra∎

For arbitrary $v \in L^{2} (D; C)$ , linearity gives $A_{k} (u - v) = A_{k} u - A_{k} v = f - A_{k} v$ . Applying the bounded inverse from step 1.1 yields $u - v = A_{k}^{- 1} (f - A_{k} v)$ , so $∥ u - v ∥_{2} \leq ∥ A_{k}^{- 1} ∥ ∥ f - A_{k} v ∥_{2} = R_{k} ∥ f - (I - K_{k}) v ∥_{2}$ .

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

Radial volume scattering separates into matched partial-wave boundary problems

Statement

Assume Countable Choice and Dependent Choice. Let $R > 0$ , $k > 0$ , and $Q \in C ([0, R]; C)$ . Set $q (x) = Q (∣ x ∣)$ on $B_{R}$ and $q (x) = 0$ outside $B_{R}$ . Choose for every $ℓ$ an orthonormal basis $Y_{ℓ m}$ of $H_{ℓ}$ . Suppose $u^{i n c}$ is a classical free Helmholtz field on all of $R^{3}$ , and $u = u^{i n c} + V$ satisfies the volume-potential equation $V (x) = \int_{B_{R}} G_{k} (x - y) q (y) u (y) d y$ . Assume $u$ is continuous on ${\overline{B}}_{R}$ , piecewise $C^{2}$ on either side of $∣ x ∣ = R$ , and $u$ and its radial derivative are continuous across that interface. For every spherical harmonic $Y_{ℓ m}$ assume its coefficient $w_{ℓ m} (r) : = r \int_{S^{2}} u (r ω) \overline{Y_{ℓ m} (ω)} d σ (ω)$ satisfies the regular-origin condition $w_{ℓ m} (r) = O (r^{ℓ + 1})$ as $r ↓ 0$ . Explicitly assume the incident regular-mode form: for every $ℓ, m$ there is $α_{ℓ m} \in C$ such that $r \int_{S^{2}} u^{i n c} (r ω) \overline{Y_{ℓ m} (ω)} d σ (ω) = α_{ℓ m} r j_{ℓ} (k r)$ for $r > 0$ . The outgoing radiation condition is $\lim_{r \to \infty} \sup_{ω \in S^{2}} ∣ r (\partial_{r} V (r ω) - i k V (r ω)) ∣ = 0$ . Then each reduced mode satisfies $w_{ℓ m}^{''} (r) + (k^{2} + Q (r) - \frac{ℓ (ℓ + 1)}{r^{2}}) w_{ℓ m} (r) = 0 (0 < r < R),$ is regular at $0$ , and has for $r > R$ the form $w_{ℓ m} (r) = α_{ℓ m} r j_{ℓ} (k r) + β_{ℓ m} r h_{ℓ}^{(1)} (k r) .$ The interior and exterior values and first derivatives agree at $R$ . This is a conditional mode reduction and matching statement; it asserts neither existence nor uniqueness of the total field.

Facts & Assumptions

Given: The radial contrast, classical field on $R^{3}$ , regular-origin hypothesis for the total modes, explicit incident regular-mode forms, and outgoing hypothesis in the Statement.

[F1]

The three-dimensional volume integral scattering model and The outgoing three-dimensional Helmholtz kernel is a fundamental solution: the volume-potential equation implies $(- Δ - k^{2}) (u - u^{i n c}) = q u$ distributionally.

[F2]

Spherical harmonics and partial-wave conventions in three dimensions: $Δ_{S^{2}} Y_{ℓ m} = - ℓ (ℓ + 1) Y_{ℓ m}$ .

[F3]

Spherical harmonics form a complete basis with the addition formula: under Countable Choice and Dependent Choice, the degree spaces are orthogonal and complete in $L^{2} (S^{2})$ .

[F4]

Spherical Bessel and outgoing Hankel recurrences solve the radial Helmholtz equation: $j_{ℓ}$ and $h_{ℓ}^{(1)}$ solve the free radial equation, have nonzero Wronskian, and respectively have regular and outgoing asymptotics.

Proof

Given: The data in the Statement.

1.1F1F2algebra

The boundedness of $Q$ and continuity of $u$ on the compact ball imply $q u \in L^{1} (B_{R})$ with bounded support, so [F1] applies. Since $u^{i n c}$ solves $(- Δ - k^{2}) u^{i n c} = 0$ , [F1] gives $Δ u + (k^{2} + Q (r)) u = 0$ in $0 < r < R$ . For fixed $r$ , multiply the polar Laplacian identity by $\overline{Y_{ℓ m}}$ , integrate over $S^{2}$ , and integrate the angular Laplacian by parts; the periodic azimuthal boundary terms cancel and the polar boundary terms vanish with the factor $\sin θ$ . By [F2], the angular term is $- ℓ (ℓ + 1)$ times the mode coefficient $a_{ℓ m} (r) : = w_{ℓ m} (r) / r$ . Hence $a_{ℓ m}^{''} + \frac{2}{r} a_{ℓ m}^{'} + (k^{2} + Q (r) - \frac{ℓ (ℓ + 1)}{r^{2}}) a_{ℓ m} = 0.$ Substitution of $w = r a$ yields the asserted reduced ODE.

1.2givenalgebra

The stated origin condition is precisely the regular branch condition for this reduced mode and is retained as a hypothesis, so no singular solution is admitted. Because $u$ and its first derivatives are continuous across $r = R$ , integration against the fixed smooth harmonic gives continuity of $w_{ℓ m}$ and $w_{ℓ m}^{'}$ there.

1.3F1F4given

Outside $B_{R}$ , the scattered field $V$ solves the free Helmholtz equation by [F1]. Its angular coefficient $v_{ℓ m} (r) = \int_{S^{2}} V (r ω) \overline{Y_{ℓ m} (ω)} d σ$ solves the free radial equation, whose independent solutions are $j_{ℓ} (k r)$ and $h_{ℓ}^{(1)} (k r)$ by the nonzero Wronskian in [F4]; write $v_{ℓ m} = a j_{ℓ} (k r) + b h_{ℓ}^{(1)} (k r)$ . The uniform radiation condition, integrated against the bounded $Y_{ℓ m}$ , gives $r (v_{ℓ m}^{'} - i k v_{ℓ m}) \to 0$ . Put $ϑ = k r - ℓ π / 2$ and $c = (- i)^{ℓ + 1}$ . The derivative recurrence in [F4] for $ℓ \geq 1$ , and direct differentiation of the base functions for $ℓ = 0$ , give $j_{ℓ}^{'} (z) = \cos (z - ℓ π / 2) / z + O (z^{- 2})$ and $(h_{ℓ}^{(1)})^{'} (z) = i c e^{i z} / z + O (z^{- 2})$ , so $r (\partial_{r} - i k) j_{ℓ} (k r) = e^{- i ϑ} + O (r^{- 1})$ while $r (\partial_{r} - i k) h_{ℓ}^{(1)} (k r) = O (r^{- 1})$ . Thus the radiation limit forces $a = 0$ , and the scattered mode is $β_{ℓ m} h_{ℓ}^{(1)} (k r)$ . Adding the stipulated incident mode gives the exterior expression in the Statement.

2.1F3step 1.2step 1.3∎

The interior/exterior continuity of the total mode and its radial derivative from step 1.2 gives the two matching equalities at $R$ . Completeness in [F3] makes these modal statements the corresponding partial-wave decomposition, without asserting that the mode equations alone solve the full inverse or uniqueness problem.

Numerical interface

For a fixed finite set of orders and a positive matching radius, Soundness and eventual absolute precision for spherical Bessel enclosures supplies certified scaled values and argument derivatives of $j_{ℓ}$ and outgoing $h_{ℓ}^{(1)}$ on positive real argument enclosures. Scaled spherical Bessel evaluation on positive real arguments also permits requesting their unscaled counterparts. At $z = k R > 0$ , the exterior matching columns are $R j_{ℓ} (z)$ and $R h_{ℓ}^{(1)} (z)$ , with radial derivatives $j_{ℓ} (z) + z j_{ℓ}^{'} (z)$ and $h_{ℓ}^{(1)} (z) + z (h_{ℓ}^{(1)})^{'} (z)$ . These identities follow by the product and chain rules. Thus the supplied first argument derivatives suffice, including at $ℓ = 0$ .

The real-argument interface does not restrict the theorem's complex contrast $Q$ : $Q$ enters the interior ODE while the exterior free wave number $k$ remains positive real. No evaluation at a complex wave number is used here. Certified exterior matching columns do not certify integration of the interior ODE, solvability of the complete matching system, total-field uniqueness, or the angular cutoff. Those claims retain their separate hypotheses and proofs. The mathematical mode-reduction statement above remains valid independently of whether numerical parameter enclosures have been provided.

For a numerical evaluation of these columns, supplied terminating positive-real parameter interfaces for $k$ and $R$ induce an argument interface for $k R$ . On a width request $2^{- n}$ , successively request both parameter boxes at indices $q = n, n + 1, \dots$ , skip until both lower endpoints are positive, multiply their dyadic intervals exactly, and stop when the product width is at most $2^{- n}$ . Each call terminates by hypothesis; positivity holds eventually because $k, R > 0$ . The parameter boxes are eventually bounded, and the product-width bound $width (X Y) \leq \sup ∣ X ∣ width (Y) + \sup ∣ Y ∣ width (X)$ tends to zero. Thus the scan terminates for every requested $n$ and retains the exact, generally non-dyadic product. The same interval operations propagate the factors $R$ and $z = k R$ in the displayed matching columns. This numerical representation condition does not restrict the theorem's exact real parameters.

TheoremStatement: AI-adaptedProof: AI-generatedprecheck passOpen item page →

Fredholm convergence and far-field error for cell-volume product integration

Statement

Assume AC and the data of The three-dimensional volume integral scattering model with $0 < ∣ D ∣ < \infty$ . Let $f \in L^{2} (D; C)$ , let $P_{N}, K_{N} = P_{N} K_{k}, f_{N} = P_{N} f$ and the exact cell-pair matrices be as in Cell-volume product integration for the scattering kernel, and suppose $I - K_{k}$ is injective. Then there is $N_{0}$ such that for all $N \geq N_{0}$ the matrix $I - A^{(N)}$ is nonsingular and its reconstruction $u_{N} \in V_{N}$ is the unique solution of $(I - K_{N}) u_{N} = f_{N}$ . If $u \in L^{2} (D; C)$ solves $(I - K_{k}) u = f$ , then for a constant $C < \infty$ independent of $N$ and $f$ , $∥ u_{N} - u ∥_{2} \leq C (∥ (P_{N} - I) K_{k} u ∥_{2} + ∥ (P_{N} - I) f ∥_{2}) .$ Writing $C_{F} : = ∥ q ∥_{\infty} \sqrt{∣ D ∣} / (4 π)$ , the exact far-field outputs obey $∥ F_{k} [u_{N}] - F_{k} [u] ∥_{C (S^{2})} \leq C_{F} C (∥ (P_{N} - I) K_{k} u ∥_{2} + ∥ (P_{N} - I) f ∥_{2}) ⟶ 0.$ This gives eventual solvability and convergence for the stated cell-volume product rule; it does not assert operator-norm convergence $K_{N} \to K_{k}$ or finite-precision error bounds.

Facts & Assumptions

Given: AC, $D, q, k$ as in the Statement, $f \in L^{2} (D; C)$ , the dyadic projections and exact product-cell rule, injectivity of $I - K_{k}$ , and any $u \in L^{2} (D; C)$ solving $(I - K_{k}) u = f$ .

[F1]

Dyadic cell-average projections converge strongly in L2 of a bounded set: $P_{N}$ is an orthogonal projection, $∥ P_{N} ∥ \leq 1$ , and $P_{N} v \to v$ for each $v \in L^{2} (D)$ .

[F2]

The volume integral scattering operator is compact: $K_{k}$ is compact and bounded on $L^{2} (D)$ under AC.

[F3]

Collective compactness gives eventual Fredholm invertibility and solution error bounds: on a nonzero Banach space, for compact $K$ , bounded $K_{N}$ with compact closure of $⋃_{N} K_{N} (B_{X})$ , pointwise norm convergence $K_{N} x \to K x$ , and injective $I - K$ , all sufficiently late $I - K_{N}$ are invertible with a uniform inverse bound, and the solution error is $C (∥ (K_{N} - K) u ∥ + ∥ f_{N} - f ∥)$ .

[F4]

Cell-volume product integration for the scattering kernel: $K_{N} = P_{N} K_{k}$ , $f_{N} = P_{N} f$ , and $I - K_{N}$ on $V_{N}$ has the displayed matrix $I - A^{(N)}$ in the normalized cell-indicator basis.

[F5]

The volume far-field map is bounded and is the uniform far-field limit: $∥ F_{k} v ∥_{\infty} \leq C_{F} ∥ v ∥_{2}$ .

[F6]

AC supplies Countable Choice and Dependent Choice (The Axiom of Choice, AC supplies the countable and dependent choices used in Banach integration); complex $L^{2} (D)$ is complete under Countable Choice (Complex Lp completeness and almost-everywhere subsequences), and under Countable Choice and Dependent Choice a metric space is compact exactly when it is complete and totally bounded (For a metric space, compact, countably compact, limit point compact, sequentially compact, and complete together with totally bounded are all equivalent, given countable choice and dependent choice).

Proof

Given: The data in the Statement.

1.1F2F3F6given

Let $X = L^{2} (D; C)$ . Since $∣ D ∣ > 0$ , $X \neq {0}$ (a normalized indicator of any positive-measure cell or subset is nonzero). By [F6] and AC, $X$ is a Banach space. [F2] gives compactness of $K_{k}$ , and the assumed injectivity makes the Fredholm premise of [F3] applicable once the approximation hypotheses are checked.

1.2F1F2F4

By [F1] and [F4], $K_{N} = P_{N} K_{k}$ is bounded linear and $∥ K_{N} ∥ \leq ∥ K_{k} ∥$ . For every $v \in X$ , [F1] applied to $K_{k} v$ gives $K_{N} v = P_{N} K_{k} v \to K_{k} v$ in norm.

2.1F1F2F3F6step 1.1algebra

Let $B_{X}$ be the closed unit ball and $C = \overline{K_{k} (B_{X})}$ . Compactness in [F2] makes $C$ compact. Strong convergence $P_{N} z \to z$ from [F1] is uniform for $z \in C$ : given $ϵ > 0$ , cover $C$ by finitely many $ϵ / 3$ balls with centers $z_{1}, \dots, z_{s}$ , choose $N_{0}$ so that $∥ P_{N} z_{j} - z_{j} ∥ < ϵ / 3$ for every center and $N \geq N_{0}$ , then use $∥ P_{N} ∥ \leq 1$ to get $\sup_{z \in C} ∥ P_{N} z - z ∥ < ϵ$ . To show the tail union is totally bounded, start with any radius $δ > 0$ , apply this uniform estimate with $ϵ = δ / 3$ , and cover $C$ by finitely many $δ / 3$ balls; each $P_{N} z$ with $N \geq N_{0}$ then lies within $2 δ / 3$ of one of their centers. Thus $⋃_{N \geq N_{0}} P_{N} (C)$ is totally bounded. Each of the finitely many earlier sets $P_{N} (C)$ is compact as a continuous image of $C$ , hence totally bounded. Their union $U : = ⋃_{N \geq 1} P_{N} (C)$ is totally bounded. Its closure is totally bounded too: for any $δ > 0$ , take a finite $δ / 2$ -net of $U$ ; every point of $\overline{U}$ is within $δ / 2$ of some point of $U$ , which is within $δ / 2$ of a net center. The closure is complete: every Cauchy sequence in $\overline{U}$ converges in the Banach space $X$ from step 1.1, and closedness of $\overline{U}$ keeps that limit in $\overline{U}$ . By [F6], $\overline{U}$ is compact. Since $K_{N} (B_{X}) = P_{N} K_{k} (B_{X}) \subset P_{N} (C)$ , the family $(K_{N})$ is collectively compact in the precise sense required by [F3].

3.1F1F3step 1.1step 1.2step 2.1

Now all hypotheses of [F3] hold by steps 1.1, 1.2, and 2.1. Therefore there are $N_{0}$ and a uniform $C$ such that $I - K_{N}$ is invertible for $N \geq N_{0}$ and its solutions satisfy $∥ u_{N} - u ∥_{2} \leq C (∥ (K_{N} - K_{k}) u ∥_{2} + ∥ f_{N} - f ∥_{2}) = C (∥ (P_{N} - I) K_{k} u ∥_{2} + ∥ (P_{N} - I) f ∥_{2})$ . The right side tends to zero by [F1].

4.1F4F5step 3.1algebra∎

For $N \geq N_{0}$ , the full-space solution satisfies $u_{N} = f_{N} + P_{N} K_{k} u_{N} \in V_{N}$ . In the normalized cell-indicator basis of [F4], its equation is exactly $(I - A^{(N)}) c = b^{(N)}$ . If that finite matrix had a nonzero null vector, its corresponding nonzero element of $V_{N}$ would be in the kernel of $I - K_{N}$ , contradicting invertibility; thus the matrix is nonsingular and its reconstruction is the unique operator solution. Finally [F5] gives the stated far-field bound, and step 3.1 makes its right side tend to zero.

5 · Examples, counterexamples and false statements

None yet.

Volume integral scattering and partial wave error

1 · Prerequisites

2 · Summary

3 · Logical flowchart

4 · Definitions, theorems and proofs

The three-dimensional volume integral scattering model

Definition

Rescaling finite-measure kernel operators to probability spaces

Statement

Facts & Assumptions

Proof

Dyadic cell-average projections converge strongly in L2 of a bounded set

Statement

Facts & Assumptions

Proof

Spherical harmonics and partial-wave conventions in three dimensions

Definition

The three-dimensional free-space scattering kernel is square integrable

Statement

Facts & Assumptions

Proof

The volume far-field map is bounded and is the uniform far-field limit

Statement

Facts & Assumptions

Proof

Spherical harmonics form a complete basis with the addition formula

Statement

Facts & Assumptions

Proof

The outgoing three-dimensional Helmholtz kernel is a fundamental solution

Statement

Facts & Assumptions

Proof

The volume integral operator is well defined and bounded on L2

Statement

Facts & Assumptions

Proof

Exponential coefficient decay gives a uniform partial-wave truncation bound

Statement

Facts & Assumptions

Proof

The volume integral scattering operator is compact

Statement

Facts & Assumptions

Proof

Born truncation error for a small-norm volume integral operator

Statement

Facts & Assumptions

Proof

Cell-volume product integration for the scattering kernel

Definition

Fredholm solvability and residual stability for volume scattering

Statement

Facts & Assumptions

Proof

Radial volume scattering separates into matched partial-wave boundary problems

Statement

Facts & Assumptions

Proof

Numerical interface

Fredholm convergence and far-field error for cell-volume product integration

Statement

Facts & Assumptions

Proof

5 · Examples, counterexamples and false statements

Sources