betaFeedback

WorksMathematics

γγ-Valued Stein–Tomas Restriction for Schatten Classes

Abstract

Let Σ⊂R3Σ\subset\mathbb R^3 be a compact C2C^2 hypersurface with nonvanishing Gaussian curvature. We prove the sharp matrix-dimension-independent Schatten-valued Stein–Tomas range RΣ:Lp(R3;Sa)⟶γ(L2(Σ),Sa),1<a<∞,1≤p≤min⁡{a,4/3}. R_Σ:L^p(\mathbb R^3;S^a)\longrightarrowγ(L^2(Σ),S^a), \qquad 1<a<\infty,\quad 1\le p\le\min\{a,4/3\}. The core endpoint is EΣ:γ(L2(Σ),S4)→L4(R3;S4)E_Σ:γ(L^2(Σ),S^4)\to L^4(\mathbb R^3;S^4). After squaring the quartic norm, multiplication lands in the Hilbert–Schmidt class. A local coarea argument splits resonant quartets according to the shorter adjacent frequency chord, producing a dimension-free row/column estimate for elliptic and hyperbolic curvature. We also show that Gaussian type 4/34/3 alone does not characterize the corresponding endpoint for arbitrary Banach spaces. Applications include matrix Herglotz waves, a two-dimensional Schrödinger L4L^4 estimate, and free spectral-measure bounds.

Introduction

The restriction problem and the Stein–Tomas line

The Fourier restriction problem asks whether the Fourier transform of an LpL^{p} function, which a priori is defined only almost everywhere in the ambient frequency space, can be meaningfully restricted to a lower-dimensional curved set. For a smooth compact hypersurface Σ⊂Rn\Sigma\subset\mathbb{R}^{n} with surface measure dσd\sigma, the classical L2L^{2} restriction problem is

∥f^Σ∥L2(Σ)≤C∥f∥Lp(Rn).(1)\left\|\widehat{f}_{\Sigma}\right\|_{L^{2}(\Sigma)} \le C\left\|f\right\|_{L^{p}(\mathbb{R}^{n})}. \tag*{(1)}

The curvature-sensitive theory emerged through Stein’s early restriction observation, the Fefferman–Stein bounds recorded in Fefferman’s work on strongly singular convolution operators [7], and Tomas’s 1975 theorem [29]. Tomas obtained the full nonendpoint range and explicitly noted that Stein had supplied the endpoint by complex interpolation; see the historical account in [21] and Stein’s monograph [24]. For the sphere, the Stein–Tomas range is

1≤p≤pST:=2(n+1)n+3.1 \le p \le p_{\mathrm{ST}} := \frac{2(n+1)}{n+3}.

At n=3n=3 this becomes pST=4/3p_{\mathrm{ST}}=4/3. The Knapp concentration example shows that the exponent cannot be improved in the scalar Lp→L2L^p \to L^2 scale.

The dual extension operator is

EΣg(x)=∫Σeix⋅ωg(ω) dσ(ω),(2)E_{\Sigma}g(x)=\int_{\Sigma}e^{ix\cdot\omega}g(\omega)\,\mathrm{d}\sigma(\omega), \tag*{(2)}

and the scalar endpoint in three dimensions is

∥EΣg∥L4(R3)≲Σ∥g∥L2(Σ).(3)\lVert E_{\Sigma}g\rVert_{L^4(\mathbb{R}^3)}\lesssim_{\Sigma}\lVert g\rVert_{L^2(\Sigma)}. \tag*{(3)}

For the sphere S2S^2, the sharp endpoint and its extremizers have been studied in detail; in particular, Foschi proved that constants are global maximizers for the adjoint sphere inequality [9], and the even exponent 44 makes the convolution structure especially transparent [21].

The restriction viewpoint quickly entered PDE. Strichartz extended the extension method to quadratic surfaces and connected it explicitly with spacetime decay for wave and Klein–Gordon type equations [25]. Carleson–Sjölin oscillatory integrals [5] and Stein’s oscillatory-integral theory enlarged the framework from translation-invariant surfaces to variable phases. Sogge’s spectral-cluster estimates on compact manifolds [23] and the uniform Sobolev inequalities of Kenig–Ruiz–Sogge [16] show another major route from restriction geometry to spectral and elliptic PDE. Keel–Tao’s abstract endpoint Strichartz theorem [15] systematized the dispersive consequences.

A second revolution came from bilinear and multilinear restriction. Tao–Vargas–Vega [27], Tao’s sharp bilinear paraboloid estimate [26], the multilinear restriction theorem of Bennett–Carbery–Tao [2], the Bourgain–Guth multilinear approach [3], and Guth’s polynomial-partitioning work [12] established restriction as a central meeting point of harmonic analysis, incidence geometry, and dispersive PDE. These developments concern the scalar restriction conjecture beyond the Stein–Tomas line. Our problem is different: the ambient exponent remains exactly the classical Stein–Tomas exponent, while the coefficient space becomes noncommutative.

Why a coefficient-space theorem is needed: a multichannel dispersive model

A simple PDE model already shows why a componentwise scalar estimate does not capture the problem studied here. Consider a matrix-valued dispersive band

U(t,x)=∫ei(x⋅ξ−tλ(ξ))G(ξ) dξ,G(ξ)∈S4,(4)U(t,x)=\int e^{i(x\cdot\xi-t\lambda(\xi))}G(\xi)\,\mathrm{d}\xi,\qquad G(\xi)\in S^4, \tag*{(4)}

where the phase λ\lambda is one branch of a constant-coefficient system and has nondegenerate Hessian on the frequency patch under consideration. In a system with frequency-dependent polarization one naturally encounters

G(ξ)=P(ξ)A(ξ)Q(ξ),G(\xi)=P(\xi)A(\xi)Q(\xi),

with spectral projectors or channel maps P(ξ),Q(ξ)P(\xi),Q(\xi) that cannot be simultaneously diagonalized as ξ\xi varies. For example, even in two internal dimensions the rank-one projections

P(ξ)=v(ξ)v(ξ)∗,v(ξ)=(cos⁡ϑ(ξ),sin⁡ϑ(ξ))T,P(\xi)=v(\xi)v(\xi)^*,\qquad v(\xi)=\left(\cos\vartheta(\xi),\sin\vartheta(\xi)\right)^T,

generically satisfy [P(ξ),P(η)]≠0[P(\xi),P(\eta)] \ne0. Estimating every matrix entry separately and then summing either loses the coherent channel geometry or introduces constants depending on the number of channels. The natural invariant quantity is instead the Gaussian row/column norm of the whole amplitude. On a nondegenerate graph, the principal estimate proved below gives precisely

∥U∥Lt,x4(S4)≲∥G∥γ(Lξ2,S4).\|U\|_{L^4_{t,x}(S^4)} \lesssim\|G\|_{\gamma(L^2_\xi,S^4)}.

with no dependence on matrix dimension. Thus the theorem is designed for dispersive systems in which the propagating field itself is operator valued, rather than for a scalar field copied into several independent channels. A useful second geometry, very different from the sphere, is the saddle

ϕhyp(ξ)=ξ12−ξ22.(5)\phi_{\mathrm{hyp}}(\xi)=\xi_1^2-\xi_2^2. \tag*{(5)}

Its Hessian has determinant −4-4, so compact patches have strictly negative Gaussian curvature and fall directly under the local theorem. The scalar restriction theory for hyperbolic surfaces is classical and has been developed much further in the modern bilinear and polynomial-partitioning literature; see, for example, Buschenhenke–Müller–Vargas [4]. What is new in the present setting is not the scalar saddle geometry itself, but the dimension-free coherent estimate for noncommuting Schatten-valued amplitudes on that geometry. The same local theorem also covers genuinely nonquadratic perturbations such as

ϕε(ξ)=ξ12−ξ22+εξ13.\phi_{\varepsilon}(\xi)=\xi_1^2-\xi_2^2+\varepsilon\xi_1^3.

on a sufficiently small patch where the Hessian remains nonsingular. This example makes clear that the argument does not rely on positive curvature, exact quadratic structure, or spherical symmetry.

The new coefficient geometry

Let SaS^a be the Schatten class on a separable Hilbert space. We study

RΣf:=f^∣Σ:Lp(R3;Sa)⟶γ(L2(Σ),Sa).(6)R_\Sigma f:=\widehat{f}\big|_\Sigma:L^p(\mathbb{R}^3;S^a)\longrightarrow\gamma(L^2(\Sigma),S^a). \tag*{(6)}

The target is not the Bochner space L2(Σ;Sa)L^2(\Sigma;S^a). It is the space of γ\gamma-radonifying operators from the Hilbert space L2(Σ)L^2(\Sigma) into SaS^a, the natural Gaussian replacement for Hilbert-valued square functions; see [30, 14]. For a finite orthonormal system (hj)(h_j) and coefficients (xj)⊂Sa(x_j)\subset S^a,

∥∑jhj⊗xj∥γ=(E∥∑jγjxj∥Sa2)1/2.\left\|\sum_j h_j\otimes x_j\right\|_{\gamma} = \left(\mathbb{E}\left\|\sum_j\gamma_jx_j\right\|_{S^a}^{2}\right)^{1/2}.

For a≥2a\ge2, noncommutative Khintchine identifies this Gaussian norm with simultaneous row and column square functions [19, 20]:

∥∑jhj⊗xj∥γ≍amax⁡{∥(∑jxj∗xj)1/2∥a,∥(∑jxjxj∗)1/2∥a}.(7)\left\|\sum_j h_j\otimes x_j\right\|_{\gamma} \asymp_a \max\left\{ \left\|\left(\sum_j x_j^*x_j\right)^{1/2}\right\|_{a}, \left\|\left(\sum_j x_jx_j^*\right)^{1/2}\right\|_{a} \right\}. \tag*{(7)}

Thus the first critical non-Hilbert endpoint in dimension three is dual to

EΣ:γ(L2(Σ),S4)⟶L4(R3;S4).(8)E_\Sigma:\gamma(L^2(\Sigma),S^4)\longrightarrow L^4(\mathbb{R}^3;S^4). \tag*{(8)}

The key point is that the power 4 is perfectly matched to the Schatten product law

S4⋅S4⊂S2,S^{4}\cdot S^{4}\subset S^{2},

and S2S^{2} is Hilbert. The scalar identity “square the L4L^{4} norm and use Plancherel” therefore survives, but the order of matrix multiplication matters. A fixed ordering produces a genuine crossing commutator. The proof below avoids estimating that commutator separately. Instead, on each resonant quartet we compare two adjacent frequency chords and pair the Hilbert–Schmidt factors along the longer one. On the region where one chord is shorter, its length controls the resonance fiber while the longer chord controls the coarea denominator; the resulting fiber mass is uniformly bounded. This direct two-region decomposition is the geometric mechanism behind the dimension-free estimate.

Main results

Our first theorem is the dimension-free matrix endpoint. We refer to it as the Schatten-valued Stein–Tomas extension theorem; the notation γ(L2(Σ),S4)\gamma(L^{2}(\Sigma),S^{4}) is independent of the particular finite-dimensional realization of S4S^{4}.

Theorem 1.1. Let Σ⊂R3\Sigma\subset\mathbb{R}^{3} be a compact C2C^{2} hypersurface with nonvanishing Gaussian curvature and smooth positive density dσ\mathrm{d}\sigma. Then

EΣ:γ(L2(Σ),S4)⟶L4(R3;S4)(9)E_{\Sigma}:\gamma(L^{2}(\Sigma),S^{4})\longrightarrow L^{4}(\mathbb{R}^{3};S^{4}) \tag*{(9)}

is bounded. The constant depends on finitely many local C2C^{2} geometric bounds, a positive lower curvature bound, the density, and a finite atlas, but not on matrix dimension.

By γ\gamma-trace duality this gives the critical restriction corollary, which is the small-Schatten endpoint used in the classification.

Corollary 1.2. Under the assumptions of Theorem 1.1,

RΣ:L4/3(R3;S4/3)⟶γ(L2(Σ),S4/3)(10)R_{\Sigma}:L^{4/3}(\mathbb{R}^{3};S^{4/3})\longrightarrow\gamma(L^{2}(\Sigma),S^{4/3}) \tag*{(10)}

is bounded with a matrix-dimension-independent constant.

Combining the critical endpoint with the Hilbert, large-Schatten, and universal L1L^{1} branches gives the sharp Schatten Stein–Tomas range in R3\mathbb{R}^{3}.

Theorem 1.3. Let 1<a<∞1<a<\infty. Then

RΣ:Lp(R3;Sa)⟶γ(L2(Σ),Sa)(11)R_{\Sigma}:L^{p}(\mathbb{R}^{3};S^{a})\longrightarrow\gamma(L^{2}(\Sigma),S^{a}) \tag*{(11)}

is bounded for

1≤p≤min⁡{a,4/3}.(12)1\le p\le\min\{a,4/3\}. \tag*{(12)}

For every such curved surface the scalar rank-one Knapp example forces p≤4/3p\le4/3. For a<2a<2, a dimension-independent estimate on the infinite Schatten class (equivalently, estimates uniform over all matrix dimensions) also forces p≤ap\le a by Gaussian type. Thus (12) is the sharp matrix-dimension-independent range.

For the paraboloid the proof is especially simple. The following corollary is the two-dimensional Schrödinger L4L^4 estimate with γ\gamma-radonifying Schatten data.

Corollary 1.4. For finite-rank Schwartz S4S^4-valued data ff on R2\mathbb{R}^2,

∥eitΔf∥L4(Rt×Rx2;S4)≲∥f^∥γ(L2(R2),S4).(13)\left\|e^{it\Delta}f\right\|_{L^4(\mathbb{R}_t\times\mathbb{R}^2_x;S^4)} \lesssim\left\|\widehat{f}\right\|_{\gamma(L^2(\mathbb{R}^2),S^4)}. \tag*{(13)}

The estimate extends by completion to the corresponding γ\gamma-data space.

At a glance: what the generalization changes

The following table separates exact scalar recovery from genuinely new coefficient-space information.

Classical objectClassical statementPresent coefficient-space statement
Stein–Tomas in R3\mathbb{R}^3L4/3(R3)→L2(Σ)L^{4/3}(\mathbb{R}^3)\to L^2(\Sigma)L4/3(R3;S4/3)→γ(L2(Σ),S4/3)L^{4/3}(\mathbb{R}^3;S^{4/3})\to\gamma(L^2(\Sigma),S^{4/3})
Herglotz/HelmholtzL2(S2)→Lx4L^2(S^2)\to L^4_xcoherent noncommuting angular amplitudes γ(L2,S4)→Lx4(S4)\gamma(L^2,S^4)\to L^4_x(S^4)
22D Schrödinger∣eitΔf∣Lt,x4≲∣f∣2\left|e^{it\Delta}f\right|_{L^4_{t,x}}\lesssim\left|f\right|_2∣eitΔf∣Lt,x4(S4)≲∣f^∣γ\left|e^{it\Delta}f\right|_{L^4_{t,x}(S^4)}\lesssim\left|\widehat f\right|_\gamma
Free spectral measurescalar L4/3→L4L^{4/3}\to L^4 density estimateS4/3S^{4/3}-valued spectral density with the same scaling
Gaussian surface fieldsscalar/Hilbert random amplitudesradonification of coherent S4S^4-valued cylindrical data
Variable phase / resolventCarleson–Sjölin and KRS theoriesnot claimed here; these are the next coefficient-space problems

Table 1.

The first three rows are direct specializations or consequences of the theorem. The last row is deliberately separated: a compact-surface theorem does not by itself provide variable-phase summation or the principal-value cancellation of a resolvent.

Relation to trace-ideal and noncommutative restriction results

Frank and Sabin proved orthonormal-function analogues of Stein–Tomas and Strichartz, together with Schatten trace-ideal versions of uniform Sobolev estimates and applications to limiting absorption, Hartree dynamics, Lieb–Thirring bounds, and scattering [11]. Their restriction theorem controls scalar densities generated by orthonormal systems. The present theorem controls an operator-valued surface amplitude itself through its two-sided row/column Gaussian geometry. Neither formulation formally contains the other.

Likewise, Fourier restriction on quantum Euclidean spaces [13] and Fourier–Wigner restriction for Schatten operators [17] are genuinely noncommutative restriction theories, but the Fourier transforms there are noncommutative/phase-space transforms. Müller recently gave a pseudodifferential-operator proof of the Fourier–Wigner/classical phase-space equivalence and improved the dependence of the corresponding constants on the support diameter [18]. These results remain different from (6): here the physical-variable Fourier transform is the ordinary Euclidean transform of an SaS^a-valued function, and the surface target is the Banach-valued γ\gamma space.

How the classical theory is recovered

Theorem 1.1 contains the classical endpoint in two elementary ways. Setting the coefficient algebra equal to C\mathbb{C} gives (3) directly. More generally, taking G(ω)=g(ω)x0G(\omega)=g(\omega)x_0 with a fixed rank-one x0x_0 reduces the estimate to scalar Stein–Tomas. Interpolation with the trivial L1→L2(Σ)L^1 \to L^2(\Sigma) restriction estimate recovers the full scalar range 1≤p≤4/31 \le p \le4/3.

A Hilbert space can also be embedded as a pure column in a Schatten class, so the theorem contains the Hilbert-valued L2→L4L^2 \to L^4 extension estimate as a special coefficient configuration. In the noncommutative setting this pure-row/pure-column case is much easier than the general theorem because only one square function survives. The genuinely new content is simultaneous left/right interaction.

The PDE consequences should be read similarly. Corollary 1.4 reduces, for scalar or rank-one data, to the classical two-dimensional L4L^4 Schrödinger Strichartz inequality. The Herglotz-wave consequence in § 8 reduces to the classical L4L^4 bound for Fourier extension from S2S^2. By contrast, the cone has one flat principal direction, so the wave-equation restriction problem is not a formal special case of Theorem 1.1; Strichartz’s quadratic-surface theory and later cone restriction estimates remain distinct inputs.

Schatten classes and γ\gamma-radonifying operators

Schatten classes

Let H\mathcal{H} be a separable Hilbert space. For 1≤a<∞1 \le a < \infty, the Schatten class Sa(H)S^a(\mathcal{H}) consists of compact operators AA whose singular values sj(A)s_j(A) belong to ℓa\ell^a:

∥A∥Sa=(∑jsj(A)a)1/a=(Tr⁡∣A∣a)1/a.\|A\|_{S^a}=\left(\sum_j s_j(A)^a\right)^{1/a}=\left(\operatorname{Tr}|A|^a\right)^{1/a}.

We write S∞=B(H)S^\infty=B(\mathcal{H}) when only the operator norm is relevant. Standard operator-ideal facts used below are collected, for example, in [22]; the type/cotype statement is classical in the local theory of Banach spaces, see [28]:

(i) duality (Sa)∗=Sa′(S^a)^*=S^{a'} for 1<a<∞1<a<\infty under the trace pairing Tr⁡(A∗B)\operatorname{Tr}(A^*B);

(ii) Schatten Hölder,

∥AB∥Sr≤∥A∥Sp∥B∥Sq,1r=1p+1q.\|AB\|_{S^r}\le\|A\|_{S^p}\|B\|_{S^q},\qquad\frac{1}{r}=\frac{1}{p}+\frac{1}{q}.

(iii) complex interpolation [Sa0,Sa1]θ=Sa[S^{a_0},S^{a_1}]_\theta=S^a with 1/a=(1−θ)/a0+θ/a11/a=(1-\theta)/a_0+\theta/a_1;

(iv) S2S^2 is the Hilbert–Schmidt space with inner product ⟨A,B⟩HS=Tr⁡(A∗B)\langle A,B\rangle_{HS}=\operatorname{Tr}(A^*B);

(v) SaS^a has Gaussian type min⁡{a,2}\min\{a,2\} and Gaussian cotype max⁡{a,2}\max\{a,2\}, up to the usual endpoint interpretation;

(vi) the norm is unitarily invariant, ∥UAV∥Sa=∥A∥Sa\|UAV\|_{S^a}=\|A\|_{S^a} for unitaries U,VU,V, and for an operator of rank at most NN,

∥A∥Sb≤∥A∥Sa≤N1/a−1/b∥A∥Sb,1≤a≤b≤∞.\|A\|_{S^b}\le\|A\|_{S^a}\le N^{1/a-1/b}\|A\|_{S^b},\qquad1\le a\le b\le\infty.

In particular all Schatten norms agree on a rank-one partial isometry.

The scale has familiar operator-theoretic meanings. S1\mathcal{S}^{1} is the trace class, the natural home of density matrices and nuclear operators; S2\mathcal{S}^{2} is the Hilbert–Schmidt class; S∞\mathcal{S}^{\infty} is the bounded-operator endpoint. The class S4\mathcal{S}^{4} is the first genuinely non-Hilbert Schatten space for which a quartic norm becomes Hilbertian after one multiplication. This is exactly the algebraic feature exploited below.

The Hilbert–Schmidt product identity is the reason a=4a=4 is special:

∥AB∗∥S22=Tr⁡(A∗AB∗B),A,B∈S4.(14)\lVert AB^{*}\rVert_{\mathcal{S}^{2}}^{2}=\operatorname{Tr}(A^{*}AB^{*}B),\qquad A,B\in\mathcal{S}^{4}. \tag*{(14)}

This is exactly the noncommutative analogue of the scalar product appearing after squaring an L4L^{4} extension norm.

gamma-radonifying operators

Let HH be a Hilbert space and XX a Banach space. If

T=∑j=1Nhj⊗xj:H→XT=\sum_{j=1}^{N}h_{j}\otimes x_{j}:H\to X

with (hj)(h_{j}) orthonormal, set

∥T∥γ(H,X)=(E∥∑j=1Nγjxj∥X2)1/2.(15)\lVert T\rVert_{\gamma(H,X)}=\left(\mathbb{E}\left\lVert\sum_{j=1}^{N}\gamma_{j}x_{j}\right\rVert_{X}^{2}\right)^{1/2}. \tag*{(15)}

The completion of finite-rank operators is γ(H,X)\gamma(H,X); see [30, 14]. The ideal property states that if A:H0→HA:H_{0}\to H and B:X→YB:X\to Y are bounded, then

∥BTA∥γ(H0,Y)≤∥B∥∥T∥γ∥A∥.(16)\lVert BTA\rVert_{\gamma(H_{0},Y)}\leq\lVert B\rVert\lVert T\rVert_{\gamma}\lVert A\rVert. \tag*{(16)}

For Hilbert XX, γ(H,X)\gamma(H,X) is the Hilbert–Schmidt class. For non-Hilbert XX, the norm records Gaussian geometry that cannot in general be replaced by a Bochner L2L^{2} norm.

For 2≤a<∞2\leq a<\infty, noncommutative Khintchine gives

(E∥∑jγjxj∥a2)1/2≃amax⁡{∥(∑jxj∗xj)1/2∥a,∥(∑jxjxj∗)1/2∥a}.(17)\left(\mathbb{E}\left\lVert\sum_{j}\gamma_{j}x_{j}\right\rVert_{a}^{2}\right)^{1/2}\simeq_{a}\max\left\{\left\lVert\left(\sum_{j}x_{j}^{*}x_{j}\right)^{1/2}\right\rVert_{a},\left\lVert\left(\sum_{j}x_{j}x_{j}^{*}\right)^{1/2}\right\rVert_{a}\right\}. \tag*{(17)}

For 1<a≤21<a\leq2, the corresponding norm is the infimum over decompositions xj=yj+zjx_{j}=y_{j}+z_{j} of the sum of the column norm of (yj)(y_{j}) and the row norm of (zj)(z_{j}); see [19, 20].

For an S4\mathcal{S}^{4}-valued simple function GG on a measure space, write

CG=(∫G∗G)1/2,RG=(∫GG∗)1/2,(18)C_{G}=\left(\int G^{*}G\right)^{1/2},\qquad R_{G}=\left(\int GG^{*}\right)^{1/2}, \tag*{(18)}

and

∥G∥Γ4=max⁡{∥CG∥4,∥RG∥4}.(19)\lVert G\rVert_{\Gamma_{4}}=\max\{\lVert C_{G}\rVert_{4},\lVert R_{G}\rVert_{4}\}. \tag*{(19)}

Then

∥G∥γ(L2,S4)≃∥G∥Γ4.(20)\lVert G\rVert_{\gamma(L^{2},\mathcal{S}^{4})}\simeq\lVert G\rVert_{\Gamma_{4}}. \tag*{(20)}

Two elementary transfer principles

We record two facts needed later. The first is the universal L1L^1 restriction principle, which requires no geometry of the coefficient Banach space.

Proposition 2.1. For every Banach space XX and compact finite-measure hypersurface Σ\Sigma,

RΣ:L1(R3;X)→γ(L2(Σ),X)R_\Sigma: L^1(\mathbb{R}^3;X) \to\gamma(L^2(\Sigma),X)

is bounded.

Proof. For x∈R3x \in\mathbb{R}^3, let kx(ω)=e−ix⋅ωk_x(\omega)=e^{-ix\cdot\omega}. The restriction operator associated with a simple XX-valued ff is the Bochner integral of rank-one operators

Tf=∫R3kx⊗f(x) dx.T_f=\int_{\mathbb{R}^3} k_x\otimes f(x)\,dx.

Since ∥kx⊗y∥γ=∥kx∥L2(Σ)∥y∥X=σ(Σ)1/2∥y∥X\|k_x\otimes y\|_\gamma=\|k_x\|_{L^2(\Sigma)}\|y\|_X=\sigma(\Sigma)^{1/2}\|y\|_X, the triangle inequality in γ\gamma yields

∥Tf∥γ≤σ(Σ)1/2∥f∥L1(X).\|T_f\|_\gamma\leq\sigma(\Sigma)^{1/2}\|f\|_{L^1(X)}.

Density completes the proof.

The second elementary principle is the Hilbert-output factorization, which transfers a scalar Hilbert-input estimate to Schatten-valued γ\gamma data on the range 1<a≤21<a\leq2.

Proposition 2.2. Let T:H→Lq(M)T:H\to L^q(M) be bounded, where HH is Hilbert and 2≤q≤∞2\leq q\leq\infty. For 1<a≤21<a\leq2,

T⊗ISa:γ(H,Sa)→Lq(M;Sa)T\otimes I_{S^a}:\gamma(H,S^a)\to L^q(M;S^a)

is bounded with a constant depending only on aa and ∥T∥\|T\|.

Proof. It suffices to treat a finite orthonormal family (hj)(h_j) and coefficients xjx_j. Put

A=(∑jxj∗xj)1/2.A=\left(\sum_j x_j^*x_j\right)^{1/2}.

Choose rr by 1/a=1/2+1/r1/a=1/2+1/r. On the support of AA, write

B=A1−a/2,yj=xjAa/2−1,xj=yjB.B=A^{1-a/2},\qquad y_j=x_jA^{a/2-1},\qquad x_j=y_jB.

Spectral regularization removes the possible kernel. Then

∑j∥yj∥22=Tr⁡Aa,∥B∥r=(Tr⁡Aa)1/r.\sum_j\|y_j\|_2^2=\operatorname{Tr} A^a,\qquad\|B\|_r=(\operatorname{Tr} A^a)^{1/r}.

Hilbert-valued tensorization of TT and Schatten Hölder imply

∥∑j(Thj)xj∥Lq(Sa)≤∥T∥∥A∥a.\left\|\sum_j(Th_j)x_j\right\|_{L^q(S^a)}\leq\|T\|\|A\|_a.

A left factorization gives the row estimate. Taking the infimum over row/column decompositions and using the a≤2a\leq2 form of noncommutative Khintchine proves the claim.

By duality, Proposition 2.2 implies the full scalar Stein–Tomas range for restriction with values in SaS^a when a≥2a\geq2.

Local quartic resonance geometry

The proof of Theorem 1.1 is local on the surface. Let U⊂R2U \subset\mathbb{R}^2 be convex and

Γ(ξ)=(ξ,ϕ(ξ)),\Gamma(\xi)=(\xi,\phi(\xi)),

where ϕ∈C2(U)\phi\in C^2(U) and D2ϕD^2\phi is uniformly nonsingular. After shrinking UU around a fixed point, we may assume that D2ϕD^2\phi stays uniformly close to one invertible matrix. We initially work with Lebesgue measure dξd\xi; smooth positive surface densities are inserted in § 5.

For G:U→S4G:U \to S^4, extended by zero outside UU, define

EϕG(x,t)=∫R2ei(x⋅ξ+tϕ(ξ))G(ξ) dξ.(21)E_\phi G(x,t)=\int_{\mathbb{R}^2}e^{i(x\cdot\xi+t\phi(\xi))}G(\xi)\,d\xi. \tag*{(21)}

We first make rigorous the Fourier-side object obtained by squaring this extension. For u∈R2u \in\mathbb{R}^2 put

qu(ξ)=ϕ(ξ+u)−ϕ(ξ).(22)q_u(\xi)=\phi(\xi+u)-\phi(\xi). \tag*{(22)}

Whenever both ξ\xi and ξ+u\xi+u belong to UU,

∇ξqu(ξ)=(∫01D2ϕ(ξ+tu) dt)u.(23)\nabla_\xi q_u(\xi)=\left(\int_0^1D^2\phi(\xi+tu)\,dt\right)u. \tag*{(23)}

The next result is the difference-map transversality lemma. It turns nondegeneracy of the Hessian into a uniform coarea description of each difference fiber.

Lemma 3.1. After shrinking UU if necessary, there is c0>0c_0>0 such that

∣∇ξqu(ξ)∣≥c0∣u∣(24)\lvert\nabla_\xi q_u(\xi)\rvert\ge c_0\lvert u\rvert \tag*{(24)}

for every admissible (ξ,u)(\xi,u). More precisely, for u≠0u\ne0 one may choose

eu=H0u∣H0u∣,H0=D2ϕ(ξ0),e_u=\frac{H_0u}{\lvert H_0u\rvert},\qquad H_0=D^2\phi(\xi_0),

so that

∣eu⋅∇ξqu(ξ)∣≥c0∣u∣(25)\lvert e_u\cdot\nabla_\xi q_u(\xi)\rvert\ge c_0\lvert u\rvert \tag*{(25)}

with one fixed sign throughout the patch. Consequently every nonempty level set of quq_u is, in coordinates adapted to eue_u, a single uniformly Lipschitz graph; in particular its length inside the fixed compact support is bounded uniformly after ∣u∣\lvert u\rvert is bounded away from zero.

Proof. Fix ξ0∈U\xi_0\in U and write H0=D2ϕ(ξ0)H_0=D^2\phi(\xi_0). If s0s_0 is the smallest singular value of H0H_0, shrink the relevant convex neighborhood until

∥D2ϕ(ζ)−H0∥≤s0/4.\lVert D^2\phi(\zeta)-H_0\rVert\le s_0/4.

Then the averaged Hessian in (23) differs from H0H_0 by at most s0/4s_0/4, which proves (24). Pairing the same identity with eu=H0u/∣H0u∣e_u=H_0u/\lvert H_0u\rvert gives (25), after a harmless further shrinking. Since the directional derivative never changes sign, quq_u is strictly monotone along every line parallel to eue_u; the implicit-function theorem and the upper Hessian bound give the asserted uniform graph and slope bounds.

For ε>0\varepsilon>0 define the S2S^2-valued difference density

Kε(u,s)=1{∣u∣≥ε}∫{qu(ξ)=s}G(ξ)∗G(ξ+u)∣∇ξqu(ξ)∣ dH1(ξ).(26)K_\varepsilon(u,s)=\mathbf{1}_{\{|u|\geq\varepsilon\}}\int_{\{q_u(\xi)=s\}}\frac{G(\xi)^*G(\xi+u)}{|\nabla_\xi q_u(\xi)|}\,dH^1(\xi). \tag*{(26)}

For each fixed ε>0\varepsilon>0, Lemma 3.1 gives a uniformly Lipschitz description of the quq_u-fibers for ∣u∣≥ε|u|\geq\varepsilon. Since GG is bounded and compactly supported, (26) is then bounded on a compact (u,s)(u,s)-set, and hence Kε∈L2(R3;S2)K_\varepsilon\in L^2(\mathbb{R}^3;S^2). It is the coarea density of the pushforward of

1{∣η−ξ∣≥ε}G(ξ)∗G(η) dξ dη\mathbf{1}_{\{|\eta-\xi|\geq\varepsilon\}}G(\xi)^*G(\eta)\,d\xi\,d\eta

under the difference map

(ξ,η)⟼Γ(η)−Γ(ξ).(\xi,\eta)\longmapsto\Gamma(\eta)-\Gamma(\xi).

Consequently, up to the harmless Fourier-normalization constant,

Kε⟶Fx,t(EϕG)∗EϕG(27)K_\varepsilon\longrightarrow\mathcal{F}_{x,t}(E_\phi G)^*E_\phi G \tag*{(27)}

in the sense of tempered distributions as ε↓0\varepsilon\downarrow0. In fact, for smooth compactly supported finite-rank GG the omitted region ∣η−ξ∣<ε|\eta-\xi|<\varepsilon has S2S^2-valued total variation tending to zero, so the underlying pushforward measures converge in total variation.

The local theorem will follow once we prove the uniform estimate

sup⁡ε>0∥Kε∥L2(R3;S2)2≲ϕ∥(∫UG∗G)1/2∥44+∥(∫UGG∗)1/2∥44.(28)\sup_{\varepsilon>0}\|K_\varepsilon\|_{L^2(\mathbb{R}^3;S^2)}^2\lesssim_\phi\left\|\left(\int_U G^*G\right)^{1/2}\right\|_4^4+\left\|\left(\int_U GG^*\right)^{1/2}\right\|_4^4. \tag*{(28)}

To identify the geometry in this L2L^2 norm, expand (26) and write the second point on the same quq_u-fiber as ξ+v\xi+v. Equality of the two quq_u values is precisely

Φξ,v(u)=ϕ(ξ+u)−ϕ(ξ)−ϕ(ξ+u+v)+ϕ(ξ+v)=0.(29)\Phi_{\xi,v}(u)=\phi(\xi+u)-\phi(\xi)-\phi(\xi+u+v)+\phi(\xi+v)=0. \tag*{(29)}

The four corresponding frequencies are

ξ1=ξ,ξ2=ξ+u,ξ4=ξ+v,ξ3=ξ+u+v.(30)\xi_1=\xi,\qquad\xi_2=\xi+u,\qquad\xi_4=\xi+v,\qquad\xi_3=\xi+u+v. \tag*{(30)}

To make the re-disintegration precise, fix δ>0\delta>0 and work first on ∣u∣,∣v∣≥δ|u|,|v|\geq\delta. On this regular region the scalar coarea theorem and Fubini define one resonance measure dνξ(u,v)d\nu_\xi(u,v) on {Φξ,v(u)=0}\{\Phi_{\xi,v}(u)=0\}, and it has the two equivalent iterated representations

∫F dνξ=∫∣u∣≥δ∫{Φξ,v(u)=0}F(u,v)∣∇vΦξ,v(u)∣ dH1(v) du(31)\int F\,d\nu_\xi=\int_{|u|\geq\delta}\int_{\{\Phi_{\xi,v}(u)=0\}}\frac{F(u,v)}{|\nabla_v\Phi_{\xi,v}(u)|}\,dH^1(v)\,du \tag*{(31)}
=∫∣v∣≥δ∫{Φξ,v(u)=0}F(u,v)∣∇uΦξ,v(u)∣ dH1(u) dv.(32)=\int_{|v|\geq\delta}\int_{\{\Phi_{\xi,v}(u)=0\}}\frac{F(u,v)}{|\nabla_u\Phi_{\xi,v}(u)|}\,dH^1(u)\,dv. \tag*{(32)}

Indeed both sides are the coarea disintegration of the same hypersurface measure for the scalar level set Φ=0\Phi=0; equivalently, this follows by applying coarea in one variable and then Fubini, with all gradients bounded away from zero on the truncated region. This observation is what permits us below to choose the uu- or vv-fiber according to which chord is longer.

If

Aξ(u)=G(ξ)∗G(ξ+u),A_\xi(u)=G(\xi)^*G(\xi+u),

then the pointwise trace in the expansion of ∥Kε∥22\lVert K_{\varepsilon}\rVert_{2}^{2} is

Tr⁡(Aξ(u)∗Aξ+v(u))=Tr⁡(G2∗G1G4∗G3).\operatorname{Tr}\left(A_{\xi}(u)^{*}A_{\xi+v}(u)\right)=\operatorname{Tr}\left(G_{2}^{*}G_{1}G_{4}^{*}G_{3}\right).

This is the complex conjugate, after cyclic permutation, of

Tr⁡(G1∗G2G3∗G4).\operatorname{Tr}\left(G_{1}^{*}G_{2}G_{3}^{*}G_{4}\right).

Hence the two expressions have exactly the same absolute value. Thus it suffices to control the absolute quartic form on the resonance set (29). The estimates are uniform in δ\delta, so Remark 3.5 removes the auxiliary truncation.

The next lemma is the resonance-fiber transversality statement. It is the four-wave analogue of the preceding difference-map estimate.

Lemma 3.2. After shrinking UU if necessary, there is c1>0c_{1}>0 such that for all admissible ξ\xi, uu, vv,

∣∇uΦξ,v(u)∣≥c1∣v∣.(33)\lvert\nabla_{u}\Phi_{\xi,v}(u)\rvert\ge c_{1}\lvert v\rvert. \tag*{(33)}

More precisely, for every v≠0v\ne0 one can choose a unit vector eve_{v}, depending only on vv and the center Hessian, such that

∣ev⋅∇uΦξ,v(u)∣≥c1∣v∣.(34)\lvert e_{v}\cdot\nabla_{u}\Phi_{\xi,v}(u)\rvert\ge c_{1}\lvert v\rvert. \tag*{(34)}

The sign of this directional derivative is fixed on the patch.

Proof. With H0H_{0} and s0s_{0} as above, the fundamental theorem of calculus gives

∇uΦξ,v(u)=−(∫01D2ϕ(ξ+u+tv) dt)v.(35)\nabla_{u}\Phi_{\xi,v}(u)=-\left(\int_{0}^{1}D^{2}\phi(\xi+u+tv)\,\mathrm{d}t\right)v. \tag*{(35)}

The averaged Hessian stays within s0/4s_{0}/4 of H0H_{0}, proving (33). Taking

ev=H0v∣H0v∣e_{v}=\frac{H_{0}v}{\lvert H_{0}v\rvert}

and shrinking once more makes ev⋅∇uΦe_{v}\cdot\nabla_{u}\Phi bounded away from zero with one fixed sign, proving (34).

The scale-critical geometric input is the short-chord fiber bound: the short chord controls the length of the resonance fiber, while the long chord controls the coarea denominator.

Lemma 3.3. There is CC such that

sup⁡ξ, v≠0∫∣u∣≤∣v∣Φξ,v(u)=0dH1(u)∣∇uΦξ,v(u)∣≤C.(36)\sup_{\xi,\ v\ne0}\int_{\substack{\lvert u\rvert\le\lvert v\rvert\\ \Phi_{\xi,v}(u)=0}}\frac{\mathrm{d}\mathcal{H}^{1}(u)}{\lvert\nabla_{u}\Phi_{\xi,v}(u)\rvert}\le C. \tag*{(36)}

The symmetric estimate with uu and vv interchanged holds on ∣v∣<∣u∣\lvert v\rvert<\lvert u\rvert.

Proof. By Lemma 3.2, in coordinates whose first axis is eve_{v}, the zero set of Φξ,v\Phi_{\xi,v} is a uniformly Lipschitz graph over the orthogonal coordinate. Its projection from the region ∣u∣≤∣v∣\lvert u\rvert\le\lvert v\rvert has length at most 2∣v∣2\lvert v\rvert, while the directional derivative, and hence the full gradient, is bounded below by c1∣v∣c_{1}\lvert v\rvert. Coarea in this graph coordinate therefore gives a uniform bound. The symmetric statement follows from

Φξ,v(u)=Φξ,u(v).\Phi_{\xi,v}(u)=\Phi_{\xi,u}(v).

Remark 3.4 (The quadratic model). If ϕ(ξ)=12ξTHξ\phi(\xi)=\frac{1}{2}\xi^{T}H\xi with HH symmetric and invertible, then the resonance identity is exact:

Φξ,v(u)=−uTHv.\Phi_{\xi,v}(u)=-u^{T}Hv.

Thus, for fixed vv, the resonant uu-fiber is a line perpendicular to HvHv; its intersection with ∣u∣≤∣v∣|u|\le|v| has length O(∣v∣)O(|v|) and the coarea denominator is ∣Hv∣≃H∣v∣|Hv|\simeq_H |v|. Lemma 3.3 is therefore immediate in the quadratic case. The preceding proof shows that on a sufficiently small curved patch the general C2C^2 geometry is a uniform perturbation of exactly this model.

Remark 3.5 (Degenerate axes). The coarea representation of the expanded L2L^2 norm degenerates at v=0v=0, and symmetrically at u=0u=0. These axes have zero measure in the product of the regular fibers. Equivalently, one may first truncate ∣u∣,∣v∣≥δ|u|,|v|\ge\delta, use the two-region estimates below uniformly in δ\delta, and then pass to the limit. The direct short-chord fiber bound is uniform down to zero; in particular the apparent 1/∣v∣1/|v| singularity is locally integrable in the two-dimensional vv variable.

The two-chord proof of the local S4S^4 extension theorem

The next result is the local graph theorem. It is the core analytic statement of the paper: once it is proved on a sufficiently small graph patch, the compact-surface theorem follows by a finite atlas and the ideal property of γ\gamma.

Theorem 4.1. Under the assumptions of § 3,

∥EϕG∥L4(R3;S4)≲ϕmax⁡{∥(∫UG∗G)1/2∥4,∥(∫UGG∗)1/2∥4}.(37)\|E_\phi G\|_{L^4(\mathbb{R}^3;S^4)}\lesssim_\phi\max\left\{\left\|\left(\int_U G^*G\right)^{1/2}\right\|_4,\left\|\left(\int_U GG^*\right)^{1/2}\right\|_4\right\}. \tag*{(37)}

The estimate is uniform over finite matrix dimensions.

Proof. It suffices to treat smooth finite-rank GG compactly supported in UU. By the discussion preceding (28), it is enough to prove (28). Expanding ∥Kε∥L2(S2)2\|K_\varepsilon\|_{L^2(S^2)}^2, applying the coarea formula off the degenerate axes, and then inserting absolute values gives the resonant quartic integral associated with (29). Split this integral into the two regions

∣u∣≤∣v∣and∣v∣<∣u∣.|u|\le|v| \qquad\text{and}\qquad|v|<|u|.

The restriction ∣u∣≥ε|u|\ge\varepsilon only removes part of these regions, so all estimates below are uniform in ε\varepsilon.

Case 1: ∣u∣≤∣v∣|u|\le|v|. Using cyclicity of the trace,

Tr⁡(G1∗G2G3∗G4)=⟨G1G4∗,G2G3∗⟩S2.(38)\operatorname{Tr}(G_1^*G_2G_3^*G_4)=\langle G_1G_4^*,G_2G_3^*\rangle_{S^2}. \tag*{(38)}

Hence

∣Tr⁡(G1∗G2G3∗G4)∣≤∥G1G4∗∥2∥G2G3∗∥2.|\operatorname{Tr}(G_1^*G_2G_3^*G_4)|\le\|G_1G_4^*\|_2\|G_2G_3^*\|_2.

In the parametrization (30), define

Fv(ξ)=∥G(ξ)G(ξ+v)∗∥2.F_v(\xi)=\|G(\xi)G(\xi+v)^*\|_2.

The two factors are Fv(ξ)F_v(\xi) and Fv(ξ+u)F_v(\xi+u). The resonant measure is invariant under (ξ,u)↦(ξ+u,−u)(\xi,u)\mapsto(\xi+u,-u) because

Φξ+u,v(−u)=−Φξ,v(u).\Phi_{\xi+u,v}(-u)=-\Phi_{\xi,v}(u).

Cauchy–Schwarz, the symmetry above, and Lemma 3.3 therefore give

T≤≲∬R2×U∥G(ξ)G(ξ+v)∗∥22[∫∣u∣≤∣v∣Φξ,v(u)=0dH1(u)∣∇uΦξ,v(u)∣]dξ dv≲∬U×U∣G(ξ)G(η)∗∣2 dξ dη=Tr⁡(∫UG∗G dξ)2=∥(∫UG∗G)1/2∥44.(39)T_{\leq} \lesssim\iint_{\mathbb{R}^{2}\times U} \lVert G(\xi)G(\xi+v)^{*}\rVert_{2}^{2}\left[\int_{\substack{|u|\leq|v|\\ \Phi_{\xi,v}(u)=0}} \frac{d\mathcal{H}^{1}(u)}{|\nabla_{u}\Phi_{\xi,v}(u)|}\right]d\xi\,dv \lesssim\iint_{U\times U}|G(\xi)G(\eta)^{*}|^{2}\,d\xi\,d\eta = \operatorname{Tr}\left(\int_{U}G^{*}G\,d\xi\right)^{2} = \left\lVert\left(\int_{U}G^{*}G\right)^{1/2}\right\rVert_{4}^{4}. \tag*{(39)}

Case 2: ∣v∣<∣u∣|v|<|u|. Use instead

∣Tr⁡(G1∗G2G3∗G4)∣≤∥G1∗G2∥2∥G3∗G4∥2.(40)\left|\operatorname{Tr}(G_{1}^{*}G_{2}G_{3}^{*}G_{4})\right|\leq\lVert G_{1}^{*}G_{2}\rVert_{2}\lVert G_{3}^{*}G_{4}\rVert_{2}. \tag*{(40)}

For fixed uu put

Hu(ξ)=∥G(ξ)∗G(ξ+u)∥2.H_{u}(\xi)=\lVert G(\xi)^{*}G(\xi+u)\rVert_{2}.

The second factor is Hu(ξ+v)H_{u}(\xi+v). The symmetry (ξ,v)↦(ξ+v,−v)(\xi,v)\mapsto(\xi+v,-v), the symmetric part of Lemma 3.3, and Cauchy–Schwarz give

T>≲∬R2×U∥G(ξ)∗G(ξ+u)∥22[∫∣v∣<∣u∣Φξ,v(u)=0dH1(v)∣∇vΦξ,v(u)∣]dξ du≲∬U×U∣G(ξ)∗G(η)∣2 dξ dη=Tr⁡(∫UGG∗ dξ)2=∥(∫UGG∗)1/2∥44.(41)T_{>}\lesssim\iint_{\mathbb{R}^{2}\times U}\lVert G(\xi)^{*}G(\xi+u)\rVert_{2}^{2}\left[\int_{\substack{|v|<|u|\\ \Phi_{\xi,v}(u)=0}}\frac{d\mathcal{H}^{1}(v)}{|\nabla_{v}\Phi_{\xi,v}(u)|}\right]d\xi\,du \lesssim\iint_{U\times U}|G(\xi)^{*}G(\eta)|^{2}\,d\xi\,d\eta = \operatorname{Tr}\left(\int_{U}GG^{*}\,d\xi\right)^{2} = \left\lVert\left(\int_{U}GG^{*}\right)^{1/2}\right\rVert_{4}^{4}. \tag*{(41)}

Combining (39) and (41) proves (28) uniformly in ε\varepsilon. Choose εk↓0\varepsilon_{k}\downarrow0. The uniform L2(R3;S2)L^{2}(\mathbb{R}^{3};S^{2}) bound and Hilbert-space weak compactness give a weakly convergent subsequence Kεk⇀KK_{\varepsilon_{k}}\rightharpoonup K. By (27), the same sequence converges distributionally to the Fourier transform of (EϕG)∗EϕG(E_{\phi}G)^{*}E_{\phi}G, so these distributions coincide. By weak lower semicontinuity, the limit satisfies the same L2L^{2} bound. Plancherel now yields

∥(EϕG)∗EϕG∥L2(S2)2≲∥(∫UG∗G)1/2∥44+∥(∫UGG∗)1/2∥44.\lVert(E_{\phi}G)^{*}E_{\phi}G\rVert_{L^{2}(S^{2})}^{2}\lesssim\left\lVert\left(\int_{U}G^{*}G\right)^{1/2}\right\rVert_{4}^{4}+\left\lVert\left(\int_{U}GG^{*}\right)^{1/2}\right\rVert_{4}^{4}.

Finally,

∥EϕG∥L4(S4)4=∥(EϕG)∗EϕG∥L2(S2)2,\lVert E_{\phi}G\rVert_{L^{4}(S^{4})}^{4}=\lVert(E_{\phi}G)^{*}E_{\phi}G\rVert_{L^{2}(S^{2})}^{2},

which proves (37). Density gives the stated completion.

Remark 4.2 (Why the crossing obstruction disappears). If one globally fixes the first pairing in (40), one is naturally led to a reordered quartic term plus a noncommutative commutator. Theorem 4.1 does not show that this commutator vanishes. It shows that estimating it separately is unnecessary. The operator ordering is chosen locally according to whether ∣u∣≤∣v∣|u|\leq|v| or ∣v∣<∣u∣|v|<|u|. This is the noncommutative analogue of allowing the geometry to choose the favorable bilinear factorization; no dyadic decomposition is needed once the two regions are separated.

Global curved surfaces

Proof of Theorem 1.1. Cover Σ\Sigma by finitely many coordinate patches Σj\Sigma_j on which, after a rigid rotation, the surface is the graph of a C2C^2 function ϕj\phi_j whose Hessian is uniformly nonsingular. The nonvanishing Gaussian-curvature assumption guarantees this after shrinking the patches. Choose a smooth partition of unity ∑jχj=1\sum_j \chi_j = 1 subordinate to this cover, with supp⁡χj⋐Σj\operatorname{supp}\chi_j \Subset\Sigma_j.

In graph coordinates the surface measure has the form

dσ=wj(ξ) dξd\sigma= w_j(\xi)\,d\xi

with wjw_j positive and bounded above and below. Let

Uj:L2(Σj,dσ)→L2(Uj,dξ),(UjG)(ξ)=wj(ξ)1/2G(Γj(ξ)),U_j : L^2(\Sigma_j,d\sigma) \to L^2(U_j,d\xi), \qquad(U_jG)(\xi) = w_j(\xi)^{1/2}G(\Gamma_j(\xi)),

which is unitary. The localized surface extension is then exactly the graph extension applied to

wj1/2Uj(χjG).w_j^{1/2}U_j(\chi_jG).

Multiplication by the bounded scalar function wj1/2w_j^{1/2} is a bounded operator on the Hilbert variable L2(Uj)L^2(U_j). Hence the ideal property of γ\gamma gives

∥wj1/2Uj(χjG)∥γ(L2(Uj),S4)≤∥wj1/2∥∞∥χjG∥γ(L2(Σ),S4)≲j∥G∥γ.\left\|w_j^{1/2}U_j(\chi_jG)\right\|_{\gamma(L^2(U_j),S^4)} \leq\left\|w_j^{1/2}\right\|_\infty\left\|\chi_jG\right\|_{\gamma(L^2(\Sigma),S^4)} \lesssim_j \left\|G\right\|_\gamma.

The local theorem and the triangle inequality in L4(S4)L^4(S^4) therefore give

∥EΣG∥L4(S4)≤∑j∥EΣ(χjG)∥L4(S4)≲Σ∥G∥γ.\left\|E_\Sigma G\right\|_{L^4(S^4)} \leq\sum_j\left\|E_\Sigma(\chi_jG)\right\|_{L^4(S^4)} \lesssim_\Sigma\left\|G\right\|_\gamma.

Finite-rank operators are dense in γ\gamma by definition; approximating the finitely many scalar L2L^2 coefficient functions by smooth functions supported inside the coordinate patches shows that the smooth finite-rank surface amplitudes used above form a dense core. This completes the proof. □\square

Remark 5.1 (Semifinite noncommutative L4L^4 spaces). The local proof only uses the tracial Hilbert space L2(M,τ)L^2(\mathcal{M},\tau), Hölder multiplication L4⋅L4⊂L2L^4\cdot L^4\subset L^2, and the row/column form of noncommutative Khintchine. The same argument therefore extends, by standard approximation, from Schatten classes to semifinite noncommutative L4(M,τ)L^4(\mathcal{M},\tau) spaces. We state the Schatten version as the main theorem because it is the sharp classification needed below and avoids additional measure-topology notation.

Restriction duality and the sharp Schatten range

The critical dual theorem

For 1<a<∞1<a<\infty, Schatten classes are KK-convex. The trace pairing identifies the dual of γ(H,Sa)\gamma(H,S^a) with γ(H,Sa′)\gamma(H,S^{a'}) up to constants depending only on aa; see [30]. Applying this to Theorem 1.1 proves Corollary 1.2.

The large-Schatten branch

Scalar Stein–Tomas gives

EΣ:L2(Σ)→L4(R3).E_{\Sigma}: L^2(\Sigma) \to L^4(\mathbb{R}^3).

Proposition 2.2 therefore gives

EΣ:γ(L2(Σ),Sb)→L4(R3;Sb),1<b≤2.E_{\Sigma}: \gamma(L^2(\Sigma), S^b) \to L^4(\mathbb{R}^3; S^b), \qquad1 < b \le2.

Dualizing at b=a′b = a' yields

RΣ:L4/3(R3;Sa)→γ(L2(Σ),Sa),2≤a<∞.(42)R_{\Sigma}: L^{4/3}(\mathbb{R}^3; S^a) \to\gamma(L^2(\Sigma), S^a), \qquad2 \le a < \infty. \tag*{(42)}

Interpolation with Proposition 2.1 yields the full 1≤p≤4/31 \le p \le4/3 range for a≥2a \ge2.

Interpolation below S2

We briefly spell out the interpolation because the domain and coefficient exponents move together.

For 4/3≤a≤24/3 \le a \le2, interpolate at the fixed spatial exponent 4/34/3 between Corollary 1.2 and the Hilbert-valued endpoint

RΣ:L4/3(R3;S2)→γ(L2(Σ),S2).R_{\Sigma}: L^{4/3}(\mathbb{R}^3; S^2) \to\gamma(L^2(\Sigma), S^2).

For a finite orthonormal family (hj)j=1N⊂L2(Σ)(h_j)_{j=1}^N \subset L^2(\Sigma), compose RΣR_{\Sigma} with the Gaussian coefficient map

TN,hf=∑j=1Nγj∫Σf^(ω)hj(ω)‾ dσ(ω).T_{N,h}f = \sum_{j=1}^{N} \gamma_j \int_{\Sigma} \widehat{f}(\omega)\overline{h_j(\omega)}\,\mathrm{d}\sigma(\omega).

The two endpoint estimates are bounds into L2(Ω;S4/3)L^2(\Omega; S^{4/3}) and L2(Ω;S2)L^2(\Omega; S^2), uniformly in NN and in the orthonormal family. Ordinary complex interpolation applies because

[L4/3(R3;S4/3),L4/3(R3;S2)]θ=L4/3(R3;Sa)\left[L^{4/3}(\mathbb{R}^3; S^{4/3}), L^{4/3}(\mathbb{R}^3; S^2)\right]_{\theta} = L^{4/3}(\mathbb{R}^3; S^a)

and

[L2(Ω;S4/3),L2(Ω;S2)]θ=L2(Ω;Sa),\left[L^2(\Omega; S^{4/3}), L^2(\Omega; S^2)\right]_{\theta} = L^2(\Omega; S^a),

where 1/a=(1−θ)/(4/3)+θ/21/a = (1-\theta)/(4/3) + \theta/2. Thus the finite Gaussian operators have the SaS^a bound with the same uniformity. Taking the supremum over finite orthonormal systems yields the γ∞\gamma_{\infty} norm. Since SaS^a is reflexive for 1<a<∞1 < a < \infty and hence contains no copy of c0c_0, the Hoffmann–Jørgensen–Kwapień theorem identifies γ∞\gamma_{\infty} with γ\gamma; see [30]. Thus

RΣ:L4/3(Sa)→γ(Sa),4/3≤a≤2.(43)R_{\Sigma}: L^{4/3}(S^a) \to\gamma(S^a), \qquad4/3 \le a \le2. \tag*{(43)}

Interpolation with p=1p = 1 gives all smaller pp.

For 1<a<4/31 < a < 4/3, apply the same finite-Gaussian argument between the universal endpoint

L1(R3;S1)→γ(S1)L^1(\mathbb{R}^3; S^1) \to\gamma(S^1)

and the critical (p,a)=(4/3,4/3)(p,a) = (4/3,4/3) theorem. Interpolation of the domain Bochner spaces and the finite Gaussian targets gives the line p=ap = a: indeed the same parameter satisfies 1/p=1−θ/4=1/a1/p = 1-\theta/4 = 1/a. After taking the supremum, γ∞=γ\gamma_{\infty} = \gamma because SaS^a is reflexive. Hence

RΣ:La(R3;Sa)→γ(Sa),1<a<4/3.(44)R_{\Sigma}: L^a(\mathbb{R}^3; S^a) \to\gamma(S^a), \qquad1 < a < 4/3. \tag*{(44)}

Interpolating once more with the universal p=1p = 1 estimate gives 1≤p≤a1 \le p \le a.

Necessity

The scalar rank-one test f(x)=g(x)x0f(x)=g(x)x_0 shows that (11) cannot hold beyond the scalar Stein–Tomas exponent p=4/3p=4/3. To see the coefficient-space obstruction, one may use the standard physical-translate argument. Choose a compactly supported scalar gg whose Fourier transform does not vanish on a surface patch, and translations yjy_j so far apart that the functions

e−iyj⋅ωg^(ω)e^{-iy_j\cdot\omega}\widehat{g}(\omega)

form a uniformly Riesz sequence in L2(Σ)L^2(\Sigma). For

f(x)=∑jg(x−yj)xj,f(x)=\sum_j g(x-y_j)x_j,

disjoint physical supports give

∥f∥Lp(Sa)=∥g∥p(∑j∥xj∥p)1/p,\|f\|_{L^p(S^a)}=\|g\|_p\left(\sum_j\|x_j\|^p\right)^{1/p},

while a bounded left inverse in the Hilbert variable of the γ\gamma norm yields Gaussian type pp for SaS^a. Since SaS^a has maximal type aa for a<2a<2, one must have p≤ap\leq a. This proves the sharpness in Theorem 1.3.

Type alone does not characterize the general Banach endpoint

The Schatten theorem reaches the coefficient-space type threshold, but this is a special feature of the Schatten geometry rather than a consequence of type alone. The following proposition gives a concrete contrast.

Proposition 6.1. Let 2<q<∞2<q<\infty and let

X=L4/3,q((0,1)3)X=L^{4/3,q}((0,1)^3)

be equipped with an equivalent Banach lattice norm. Then XX has Gaussian type 4/34/3, but the spherical endpoint

RS2:L4/3(R3;X)⟶γ(L2(S2),X)R_{S^2}:L^{4/3}(\mathbb{R}^3;X)\longrightarrow\gamma(L^2(S^2),X)

is not bounded.

Proof. The type assertion follows from the classical Lorentz-space theorem of Creekmore [6]: since 4/3<q4/3<q and 4/3<24/3<2, the space L4/3,qL^{4/3,q} has Rademacher type 4/34/3, equivalently Gaussian type 4/34/3. Suppose, to the contrary, that the displayed restriction estimate holds. Let f∈C∞(R3)f\in C^\infty(\mathbb{R}^3) and put Ω=(0,1)3\Omega=(0,1)^3. For L≥1L\geq1 define

FL(x)(y)=f(x−Ly),y∈Ω.F_L(x)(y)=f(x-Ly),\qquad y\in\Omega.

For every fixed xx, the distribution function of y↦f(x−Ly)y\mapsto f(x-Ly) is bounded by L−3L^{-3} times that of ff. Hence the Lorentz dilation law and monotonicity give

∥FL(x)∥X≲qL−9/4∥f∥L4/3,q(R3).\|F_L(x)\|_X\lesssim_q L^{-9/4}\|f\|_{L^{4/3,q}(\mathbb{R}^3)}.

If supp⁡f⊂[−R,R]3\operatorname{supp} f \subset[-R,R]^3, then FLF_L is supported in LΩ+[−R,R]3L\Omega+ [-R,R]^3 as a function of xx, so

lim sup⁡L→∞∥FL∥L4/3(R3;X)≲q∥f∥L4/3,q(R3).(45)\limsup_{L\to\infty}\|F_L\|_{L^{4/3}(\mathbb{R}^3;X)}\lesssim_q \|f\|_{L^{4/3,q}(\mathbb{R}^3)}. \tag*{(45)}

On the sphere,

(RS2FL)(ω)(y)=e−iLy⋅ωf^(ω).(R_{S^2}F_L)(\omega)(y)=e^{-iLy\cdot\omega}\widehat{f}(\omega).

For a Banach function lattice the Gaussian norm dominates the pointwise square-function norm. Indeed, for a finite orthonormal family (hj)(h_j) and coefficient functions xj∈Xx_j\in X,

(E∥∑jγjxj∥X2)1/2≥E∥∑jγjxj∥X≥∥E∣∑jγjxj∣∥X=c∥(∑j∣xj∣2)1/2∥X.\left(\mathbb{E}\left\|\sum_j\gamma_jx_j\right\|_X^2\right)^{1/2}\geq\mathbb{E}\left\|\sum_j\gamma_jx_j\right\|_X\geq\left\|\mathbb{E}\left|\sum_j\gamma_jx_j\right|\right\|_X=c\left\|\left(\sum_j|x_j|^2\right)^{1/2}\right\|_X.

Applying this to finite orthogonal projections in L2(S2)L^2(S^2), using the γ\gamma-ideal property and then monotone convergence, gives

∥RS2FL∥γ(L2(S2),X)≳q∥(∫S2∣(RS2FL)(ω)(⋅)∣2 dσ(ω))1/2∥X.\|R_{S^2}F_L\|_{\gamma(L^2(S^2),X)}\gtrsim_q \left\|\left(\int_{S^2}|(R_{S^2}F_L)(\omega)(\cdot)|^2\,d\sigma(\omega)\right)^{1/2}\right\|_X.

The square function on the right is the constant function ∥RS2f∥2\|R_{S^2}f\|_2. Combining this observation with (45) and letting L→∞L\to\infty would therefore imply the scalar Lorentz estimate

∥RS2f∥2≲q∥f∥L4/3,q(R3).\|R_{S^2}f\|_2\lesssim_q\|f\|_{L^{4/3,q}(\mathbb{R}^3)}.

Bak and Seeger proved the stronger positive endpoint L4/3,2→L2(S2)L^{4/3,2}\to L^2(S^2) and showed that the second Lorentz index 22 is optimal: the estimate fails for every L4/3,qL^{4/3,q} with q>2q>2 [1]. Their finite multiscale Knapp examples may be truncated smoothly in physical space without changing the contradiction, because they are rapidly decreasing and L1→L2(S2)L^1\to L^2(S^2) restriction is bounded. This contradiction proves the proposition.

Remark 6.2. Proposition 6.1 shows that the general Banach-space endpoint contains information beyond an attained Gaussian type exponent. In particular, the positive Schatten result cannot be recovered from a type-only transfer principle. The conclusion is already visible on finite-dimensional coefficient spaces: for every fixed q>2q>2, the translation construction may be approximated in the internal variable by step functions and in the physical variable in both L4/3L^{4/3} and L1L^1. Proposition 2.1 controls the resulting output error. Hence one obtains finite-dimensional subspaces XN⊂L4/3,qX_N\subset L^{4/3,q} whose type-4/34/3 constants stay uniformly bounded while their best spherical endpoint restriction constants tend to infinity. Thus no dimension-free endpoint theorem can depend only on the type constant.

The same argument singles out L4/3,2L^{4/3,2} as a natural Lorentz borderline. Bak–Seeger give the scalar estimate L4/3,2(R3)→L2(S2)L^{4/3,2}(\mathbb{R}^3)\to L^2(S^2), so the translation obstruction above disappears at the second Lorentz index 22; that scalar fact, however, does not by itself imply the L4/3,2L^{4/3,2}-valued γ\gamma estimate.

Classical restriction theory as a special and neighboring theory

The purpose of this section is twofold: to make explicit which classical statements are direct consequences of our theorem, and to distinguish them from neighboring results that motivate but do not formally follow from it.

Direct scalar recovery

Take Sa=CS^a = \mathbb{C}, or equivalently restrict Theorem 1.1 to a fixed rank-one operator direction. Then CG=RG=∥G∥L2C_G = R_G = \lVert G\rVert_{L^2} and Theorem 1.1 becomes

∥EΣg∥L4(R3)≲∥g∥2.\lVert E_{\Sigma}g\rVert_{L^4(\mathbb{R}^3)} \lesssim\lVert g\rVert_2.

Duality gives the three-dimensional Stein–Tomas endpoint

∥f^∣Σ∥2≲∥f∥4/3,\lVert\widehat{f}|_{\Sigma}\rVert_2 \lesssim\lVert f\rVert_{4/3},

and interpolation with L1L^1 gives 1≤p≤4/31 \le p \le4/3. Thus the classical Tomas–Stein theorem on compact curved surfaces in R3\mathbb{R}^3 is literally the scalar subcase of the present result.

For S2S^2 with the standard normalization, sharp constants and extremizers constitute a separate sharp-constant problem; see Foschi and Oliveira e Silva [9, 21]. Our theorem is dimension-free in the internal matrix size, not a claim about the optimal scalar geometric constant.

Fefferman, Tomas, Stein, and Carleson–Sjölin

Stein first observed that curvature can permit Lp→L2L^p \to L^2 restriction to the sphere; early Fefferman–Stein bounds are recorded in Fefferman’s 1970 paper [7]. Tomas [29] obtained the full nonendpoint range in all dimensions and ended his note by crediting Stein with the endpoint extension by complex interpolation; see [21] for a concise historical account. Fefferman’s ball-multiplier theorem [8] then displayed the complementary force of Kakeya geometry in Fourier multiplier theory. Stein subsequently developed the oscillatory-integral framework in which restriction is now usually presented [24]. Carleson–Sjölin [5] replaced translation-invariant extension by variable-phase oscillatory integrals. Bak and Seeger later proved endpoint Lorentz refinements of Stein–Tomas and parallel endpoint improvements for Carleson–Sjölin–Hörmander operators and spectral projections [1]. Theorem 1.1 should therefore be viewed as a coefficient-space generalization of the three-dimensional Stein–Tomas endpoint, not as a generalization of the full variable-phase theory.

Strichartz and dispersive equations

Strichartz’s 1977 paper [25] treated broad classes of quadratic surfaces and explicitly connected restriction/extension with spacetime decay for solutions of wave and Klein–Gordon equations. The paraboloid yields Schrödinger evolution. Keel and Tao [15] later established an abstract dispersive theorem giving the endpoint admissible family for a broad class of propagators.

Our Corollary 1.4 is a coefficient-space analogue of one particularly symmetric classical Strichartz estimate: the Lt,x4L^4_{t,x} estimate in two spatial dimensions. It is not the forbidden two-dimensional endpoint Lt2Lx∞L^2_t L^\infty_x, nor does it by itself give the full Keel–Tao mixed-norm family. What is new is that the initial frequency amplitude may be genuinely noncommuting and is measured in the γ(L2,S4)\gamma(L^2,S^4) norm rather than the Bochner L2(S4)L^2(S^4) norm.

Bilinear, multilinear, and polynomial-partitioning restriction

The decisive geometric idea in our proof is bilinear in spirit: a quartic expression is factored into two Hilbert–Schmidt products, and the favorable factorization is chosen according to the relative lengths of two frequency chords. This is philosophically close to the bilinear restriction program of Tao–Vargas–Vega [27] and Tao [26], and to the multilinear transversality theory of Bennett–Carbery–Tao [2]. Bourgain–Guth [3] and Guth [12] use multilinear information and polynomial partitioning to go beyond the Stein–Tomas line in the scalar restriction conjecture. Our theorem does not improve the scalar restriction exponents; instead it uses the special even exponent 4 to obtain a new noncommutative amplification exactly at the Stein–Tomas point.

Sogge, spectral clusters, and Kenig–Ruiz–Sogge

Restriction estimates are local Euclidean models for spectral projection and Fourier integral estimates on manifolds. Sogge’s spectral-cluster bounds [23] are a fundamental manifestation of this principle. Kenig–Ruiz–Sogge [16] developed uniform Sobolev inequalities and unique continuation estimates in which the same curved-frequency geometry reappears through the resolvent. § 8 derives a free spectral-measure consequence of our theorem. A full γ\gamma/Schatten KRS resolvent theorem requires an additional singular-integral/cancellation argument and is intentionally not claimed here.

Orthonormal systems and trace ideals

Frank–Lewin–Lieb–Seiringer [10] and Frank–Sabin [11] developed restriction and Strichartz estimates for orthonormal systems, naturally expressed through Schatten trace ideals. Those estimates are powerful in many-body and density-matrix PDE. The present theorem concerns a different noncommutative object. Rather than a scalar density

∑jνj∣Ehj∣2\sum_j \nu_j |Eh_j|^2

we estimate a coherent matrix field

∑j(Ehj)xj\sum_j (Eh_j)x_j

with arbitrary noncommuting coefficients xjx_j. The right norm simultaneously sees

∑jxj∗xjand∑jxjxj∗.\sum_j x_j^*x_j \quad\text{and}\quad\sum_j x_jx_j^*.

This coherent row/column geometry is invisible in scalar occupation-number estimates.

Applications to PDE and spectral theory

Matrix Herglotz waves and Helmholtz

For G:S2→S4G:S^2\to S^4, define the matrix Herglotz wave

U(x)=∫S2eix⋅ωG(ω) dσ(ω).(46)U(x)=\int_{S^2}e^{ix\cdot\omega}G(\omega)\,d\sigma(\omega). \tag*{(46)}

Entrywise,

(Δ+1)U=0.(\Delta+1)U=0.

Theorem 1.1 gives

∥U∥L4(R3;S4)≲∥G∥γ(L2(S2),S4).(47)\|U\|_{L^4(\mathbb{R}^3;S^4)}\lesssim\|G\|_{\gamma(L^2(S^2),S^4)}. \tag*{(47)}

This controls coherent multichannel Helmholtz fields. In scattering language, the angular data may represent coupled polarization or channel amplitudes; noncommutativity allows interactions that cannot be diagonalized once and for all in the angular variable.

By duality, a compactly supported S4/3S^{4/3}-valued source has a γ\gamma-radonifiable far-field restriction:

∥f^∣S2∥γ(L2(S2),S4/3)≲∥f∥L4/3(R3;S4/3).\|\widehat{f}|_{S^2}\|_{\gamma(L^2(S^2),S^{4/3})}\lesssim\|f\|_{L^{4/3}(\mathbb{R}^3;S^{4/3})}.

The scalar far-field estimate is the rank-one special case.

Schrödinger evolution in two space dimensions

For the paraboloid

ϕ(ξ)=∣ξ∣2,\phi(\xi)=|\xi|^2,

the resonance function satisfies exactly

Φξ,v(u)=−2u⋅v.\Phi_{\xi,v}(u)=-2u\cdot v.

Thus the fiber is a line perpendicular to vv and

δ(u⋅v) du=∣v∣−1dHv⊥1(u).\delta(u\cdot v)\,du=|v|^{-1}d\mathcal{H}^{1}_{v^\perp}(u).

Theorem 4.1 therefore holds globally for compactly supported data without a curvature-perturbation error. After the harmless Fourier normalization this is Corollary 1.4.

The significance of (13) is not merely “matrix-valued Strichartz.” A generic S4S^4-valued Lx2L^2_x function and a γ(Lx2,S4)\gamma(L^2_x,S^4) element have different geometry. The γ\gamma norm records randomized coherent superpositions and is the natural input norm for cylindrical Gaussian forcing and for many Banach-space stochastic integration problems.

Saddle dispersion and negative curvature

The hyperbolic model (5) gives a second exact quadratic test. If

V(t,x)=∫R2ei(x⋅ξ+t(ξ12−ξ22))G(ξ) dξ,V(t,x)=\int_{\mathbb{R}^2}e^{i(x\cdot\xi+t(\xi_1^2-\xi_2^2))}G(\xi)\,d\xi,

then the resonance function is

Φξ,v(u)=−2(u1v1−u2v2),\Phi_{\xi,v}(u)=-2(u_1v_1-u_2v_2),

so the coarea fiber is again a line and the two-chord proof applies without perturbative geometry. For compactly supported S4S^4-valued data,

∥V∥Lt,x4(S4)≲∥G∥γ(L2(R2),S4).(48)\|V\|_{L^4_{t,x}(S^4)}\lesssim\|G\|_{\gamma(L^2(\mathbb{R}^2),S^4)}. \tag*{(48)}

Equivalently, this is an L4L^4 estimate for the matrix-valued saddle-dispersive equation i∂tV+(∂x12−∂x22)V=0i\partial_tV+(\partial_{x_1}^2-\partial_{x_2}^2)V=0. Scalar hyperbolic restriction is part of the classical theory, including modern results for general smooth negative-curvature surfaces [4]; (48) displays the additional coherent row/column information retained by the Schatten-valued theorem.

Frequency-localized Klein–Gordon and other nondegenerate graphs

Let

ϕm(ξ)=m2+∣ξ∣2.\phi_m(\xi)=\sqrt{m^2+|\xi|^2}.

On every compact frequency annulus, D2ϕmD^2\phi_m is nonsingular, so Theorem 4.1 gives a frequency-localized S4S^4-valued Lt,x4L^4_{t,x} extension estimate for the two-dimensional Klein–Gordon flow. More generally, any compactly supported dispersive graph with uniformly nonsingular Hessian admits the same local theorem.

The cone is different: one principal curvature vanishes. Therefore the wave equation requires cone-specific restriction geometry and is not covered by simply setting ϕ(ξ)=∣ξ∣\phi(\xi)=|\xi| in Theorem 4.1. This distinction mirrors the classical literature beginning with Strichartz [25].

A free spectral-measure consequence

Let P=−ΔP=\sqrt{-\Delta} on R3\mathbb{R}^3. With the unit-sphere normalization, write

Rλf(ω)=f^(λω),Eλg(x)=∫S2eiλx⋅ωg(ω) dσ(ω).R_\lambda f(\omega)=\widehat{f}(\lambda\omega),\qquad E_\lambda g(x)=\int_{S^2}e^{i\lambda x\cdot\omega}g(\omega)\,d\sigma(\omega).

Scaling gives

∥Rλ∥L4/3(S4/3)→γ(S4/3)≲λ−3/4,\|R_\lambda\|_{L^{4/3}(S^{4/3})\to\gamma(S^{4/3})}\lesssim\lambda^{-3/4},

and Proposition 2.2 gives the matching extension bound with S4/3S^{4/3} values. Since

dEPdλ(λ)=cλ2EλRλ,\frac{dE_P}{d\lambda}(\lambda)=c\lambda^2E_\lambda R_\lambda,

we obtain

∥dEPdλ(λ)∥L4/3(S4/3)→L4(S4/3)≲λ1/2.(49)\left\|\frac{dE_P}{d\lambda}(\lambda)\right\|_{L^{4/3}(S^{4/3})\to L^4(S^{4/3})}\lesssim\lambda^{1/2}. \tag*{(49)}

For A=−ΔA=-\Delta and energy ρ=λ2\rho=\lambda^2,

∥dEAdρ(ρ)∥L4/3(S4/3)→L4(S4/3)≲ρ−1/4.(50)\left\|\frac{dE_A}{d\rho}(\rho)\right\|_{L^{4/3}(S^{4/3})\to L^4(S^{4/3})}\lesssim\rho^{-1/4}. \tag*{(50)}

The exponent is the same scaling exponent that appears on the conjugate line in free uniform Sobolev theory. However, integrating the spectral measure against (ρ−z)−1(\rho-z)^{-1} requires principal-value cancellation near the spectrum. Equations (49)–(50) are therefore a precursor to, not a substitute for, a KRS-type resolvent theorem.

Radonification of Gaussian surface fields

Let KK be a Hilbert noise space and

Ψ∈γ(K,γ(L2(Σ),S4)).\Psi\in\gamma\bigl(K,\gamma(L^2(\Sigma),S^4)\bigr).

By the ideal property (16) and Theorem 1.1,

EΣ∘Ψ∈γ(K,L4(R3;S4)),∥EΣΨ∥γ≲∥Ψ∥γ.(51)E_\Sigma\circ\Psi\in\gamma\bigl(K,L^4(\mathbb{R}^3;S^4)\bigr),\qquad\|E_\Sigma\Psi\|_\gamma\lesssim\|\Psi\|_\gamma. \tag*{(51)}

Thus a cylindrical Gaussian angular field whose covariance operator is radonifying at the surface level becomes an actual Lx4(S4)L^4_x(S^4)-valued Gaussian Herglotz field. This is the safe stochastic formulation: homogeneous spectral white noise with constant pointwise variance is not globally Lx4L^4_x without an additional radonifying/spatial localization assumption.

Why the noncommutative extension can matter

The classical scalar estimate treats one amplitude at a time. Hilbert tensorization treats many channels when their internal geometry is orthogonal. Orthonormal-density inequalities treat incoherent occupation numbers. Theorem 1.1 instead controls coherent sums with noncommuting internal amplitudes. This is relevant whenever the field itself is matrix/operator valued, for example:

  • finite or countable channel scattering with noncommuting coupling matrices;

  • density-operator or covariance amplitudes before taking a scalar trace;

  • matrix-valued random waves with correlated channels;

  • linearized systems in which spectral polarization changes with frequency and cannot be simultaneously diagonalized.

The theorem does not solve these PDEs by itself. It provides a dimension-independent linear estimate in exactly the coefficient geometry needed to begin perturbative, stochastic, or nonlinear arguments without first discarding coherent operator structure.

Further directions

Banach-space geometry

For a general Banach space XX, γ\gamma-valued restriction at exponent pp forces Gaussian type pp, and on bounded flat frequency sets type pp is sufficient. Proposition 6.1 shows that curvature does not make type alone sufficient at the three-dimensional Stein–Tomas endpoint: L4/3,qL^{4/3,q} has type 4/34/3 for q>2q > 2 but fails the endpoint restriction estimate. Thus a general characterization must contain finer information reflecting how curvature-generated contributions from different scales are aggregated. The Schatten theorem shows that noncommutative row/column geometry supplies enough additional structure to reach the type threshold in that scale. The borderline space L4/3,2L^{4/3,2} is an especially concrete next test. We do not propose a general characterization here; identifying an intrinsic Banach-space property behind endpoint γ\gamma-restriction is a separate problem.

Variable phases

The next natural extension is a γ(S4)→L4(S4)\gamma(S^4) \to L^4(S^4) theorem for Carleson–Sjölin/Hörmander oscillatory integrals with quantitatively controlled phase geometry. The local proof here suggests the correct mechanism: resonance fibers, two adjacent increments, and a scale-dependent choice of row or column factorization. Carrying this out uniformly under variable phase localization is a separate theorem.

Uniform resolvents

The spectral-measure estimate (50) has the KRS scaling, but a resolvent requires the cancellation of a singular Cauchy integral. A future Schatten/γ\gamma KRS theorem must therefore add a limiting-absorption or analytic-family argument. Frank–Sabin’s trace-ideal uniform Sobolev theory [11] provides an important neighboring model, while the present theorem supplies a different coherent coefficient geometry.

Higher dimensions and even exponents

In dimension two the Stein–Tomas dual exponent for the circle is 6. A three-fold alternating product of S6S^6 coefficients again lands in S2S^2, suggesting a trilinear analogue of the present proof. More generally, even exponents may permit a hierarchy of noncommutative multilinearizations. The combinatorics of resonance fibers becomes substantially more complicated, and we do not pursue it here.

Acknowledgments and disclosure

The authors received no external funding for this work.

The authors used Gemini for language, organization and bibliographic assistance. The authors take full responsibility for the mathematical content and the cited sources.

References

References

  1. [1]J.-G. Bak and A. Seeger, Extensions of the Stein–Tomas theorem, Math. Res. Lett. 18 (2011), 767–781.arxiv.org/abs/1004.4948
  2. [2]J. Bennett, A. Carbery, and T. Tao, On the multilinear restriction and Kakeya conjectures, Acta Math. 196 (2006), 261–302.DOI
  3. [3]J. Bourgain and L. Guth, Bounds on oscillatory integral operators based on multilinear estimates, Geom. Funct. Anal. 21 (2011), 1239–1295.arxiv.org/abs/1012.3760
  4. [4]S. Buschenhenke, D. Müller, and A. Vargas, Fourier restriction for smooth hyperbolic 2-surfaces, Math. Ann. 387 (2023), 17–56.arxiv.org/abs/2010.10449
  5. [5]L. Carleson and P. Sjölin, Oscillatory integrals and a multiplier problem for the disc, Studia Math. 44 (1972), 287–299.
  6. [6]J. Creekmore, Type and cotype in Lorentz Lp,q spaces, Indag. Math. (Proc.) 84 (1981), no. 2, 145–152.
  7. [7]C. Fefferman, Inequalities for strongly singular convolution operators, Acta Math. 124 (1970), 9–36.DOI
  8. [8]C. Fefferman, The multiplier problem for the ball, Ann. of Math. (2) 94 (1971), 330–336.DOI
  9. [9]D. Foschi, Global maximizers for the sphere adjoint Fourier restriction inequality, J. Funct. Anal. 268 (2015), 690–702.arxiv.org/abs/1310.2510
  10. [10]R. L. Frank, M. Lewin, E. H. Lieb, and R. Seiringer, Strichartz inequality for orthonormal functions, J. Eur. Math. Soc. 16 (2014), 1507–1526.arxiv.org/abs/1306.1309
  11. [11]R. L. Frank and J. Sabin, Restriction theorems for orthonormal functions, Strichartz inequalities, and uniform Sobolev estimates, Amer. J. Math. 139 (2017), 1649–1691.arxiv.org/abs/1404.2817
  12. [12]L. Guth, Restriction estimates using polynomial partitioning II, Acta Math. 221 (2018), 81–142.arxiv.org/abs/1603.04250
  13. [13]G. Hong, X. Lai, and L. Wang, Fourier restriction estimates on quantum Euclidean spaces, Adv. Math. 430 (2023), Article 109232.arxiv.org/abs/2209.01570
  14. [14]T. Hytönen, J. van Neerven, M. Veraar, and L. Weis, Analysis in Banach Spaces. Volume II: Probabilistic Methods and Operator Theory, Springer, 2017.DOI
  15. [15]M. Keel and T. Tao, Endpoint Strichartz estimates, Amer. J. Math. 120 (1998), 955–980.DOI
  16. [16]C. E. Kenig, A. Ruiz, and C. D. Sogge, Uniform Sobolev inequalities and unique continuation for second order constant coefficient differential operators, Duke Math. J. 55 (1987), 329–347.DOI
  17. [17]F. Luef and H. J. Samuelsen, Fourier restriction for Schatten class operators and functions on phase space, Int. Math. Res. Not. IMRN 2025 (2025), no. 2, rnae291.arxiv.org/abs/2407.16259
  18. [18]D. Müller, Bounds on pseudodifferential operators and Fourier restriction for Schatten classes, J. Funct. Anal. 290 (2026), no. 6, Article 111311.arxiv.org/abs/2412.05879
  19. [19]F. Lust-Piquard, Inégalités de Khintchine dans Cₚ (1 < p < ∞), C. R. Acad. Sci. Paris Sér. I Math. 303 (1986), 289–292.
  20. [20]F. Lust-Piquard and G. Pisier, Non-commutative Khintchine and Paley inequalities, Ark. Mat. 29 (1991), 241–260.
  21. [21]D. Oliveira e Silva, The endpoint Stein–Tomas inequality: old and new, São Paulo J. Math. Sci. 18 (2024), 1015–1024.DOI
  22. [22]B. Simon, Trace Ideals and Their Applications, 2nd ed., Math. Surveys Monogr. 120, Amer. Math. Soc., 2005.
  23. [23]C. D. Sogge, Concerning the Lᵖ norm of spectral clusters for second-order elliptic operators on compact manifolds, J. Funct. Anal. 77 (1988), 123–138.DOI
  24. [24]E. M. Stein, Harmonic Analysis: Real-Variable Methods, Orthogonality, and Oscillatory Integrals, Princeton Math. Ser. 43, Princeton Univ. Press, 1993.
  25. [25]R. S. Strichartz, Restrictions of Fourier transforms to quadratic surfaces and decay of solutions of wave equations, Duke Math. J. 44 (1977), 705–714.DOI
  26. [26]T. Tao, A sharp bilinear restriction estimate for paraboloids, Geom. Funct. Anal. 13 (2003), 1359–1384.DOI
  27. [27]T. Tao, A. Vargas, and L. Vega, A bilinear approach to the restriction and Kakeya conjectures, J. Amer. Math. Soc. 11 (1998), 967–1000.DOI
  28. [28]N. Tomczak-Jaegermann, Banach–Mazur Distances and Finite-Dimensional Operator Ideals, Pitman Monographs Surveys Pure Appl. Math. 38, Longman Scientific & Technical, 1989.
  29. [29]P. A. Tomas, A restriction theorem for the Fourier transform, Bull. Amer. Math. Soc. 81 (1975), 477–478.DOI
  30. [30]J. van Neerven, γ-radonifying operators—a survey, Proc. Centre Math. Appl. Austral. Nat. Univ. 44 (2010), 1–61; arXiv:0911.3788.arxiv.org/abs/0911.3788

Paper details

vlnc:W_g6d3fbqd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 2

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Gemini (multiple versions)

View statement
Looking for related papers…
Contents