Open questions from the phase-space approach to weighted Fourier extension inequalities: strengthening the parabolic Sobolev–Mizohata–Takeuchi estimate, higher-dimensional tomographic Wigner distributions, fractional marginals, and higher-dimensional Flandrin-type bounds

The phase-space approach of Bennett–Gutiérrez–Nakamura–Oliveira (Forum of Mathematics, Sigma, 2025) reformulates the classical Stein and Mizohata–Takeuchi problems for weighted $L^2$ bounds on Fourier extension operators in terms of geometric Wigner distributions, and proves unconditional Sobolev variants of them. The paper explicitly poses four open questions that define the remaining open space of this framework: (1) whether the parabolic Sobolev–Mizohata–Takeuchi estimate can be strengthened by replacing the supremum over the midpoint hull of the Fourier support with the supremum over the actual support (Question 11.1); (2) whether the planar tomographic construction of geometric Wigner distributions via the X-ray transform extends to all dimensions, perhaps through the Radon transform (Question 11.2); (3) whether oblique or fractional phase-space marginals yield fractional forms of the Stein and Mizohata–Takeuchi inequalities (Question 11.3); and (4) whether the planar Flandrin-type inequality with an $\varepsilon$-loss extends to spatial dimensions $d>1$ (Question 11.4). No resolution of these four questions is known: the 2025–2026 counterexamples of Cairo, Cairo–Zhang, Fenves, and Bennett–Kovač–Nakamura–Oliveira refute the global and local endpoint Mizohata–Takeuchi conjectures, but they target the $L^\infty$-marginal endpoint inequalities, not the Sobolev-norm variants, the tomographic construction, or the Flandrin-type bound.

Background

To a smooth codimension-one submanifold $S$ of $\mathbb{R}^n$ carrying surface measure $d\sigma$, one associates the Fourier extension operator $\widehat{gd\sigma}(x)=\int_S g(u)e^{-2\pi i x\cdot u}\,d\sigma(u)$, where $g\in L^2(d\sigma)$; estimating such operators is the subject of (Fourier) restriction theory, with applications from dispersive PDE to analytic number theory. A question posed by Elias M. Stein in 1978 asks whether the weighted $L^2$ bound $\int_{\mathbb{R}^n}|\widehat{gd\sigma}(x)|^2w(x)\,dx\le C\int_S|g(u)|^2\sup_{v\in T_uS}Xw(N(u),v)\,d\sigma(u)$ holds for all nonnegative weights $w$, where $N:S\to S^{n-1}$ is the Gauss map (the unit normal field), $T_uS=N(u)^\perp$ is the tangent space, and $Xw(\omega,v)=\int_{-\infty}^{\infty}w(v+t\omega)\,dt$ is the classical X-ray transform (the integral of $w$ along the line through $v$ in direction $\omega$). The simpler Mizohata–Takeuchi inequality replaces the $u$-integral on the right by $C\|g\|_{L^2(d\sigma)}^2\sup_{(u,v)\in TS}Xw(N(u),v)$ with the supremum restricted to $u\in\operatorname{supp}(g)$; it arose from work of Mizohata and Takeuchi on well-posedness of Schrödinger equations. Both conjectures were recently refuted in this generality: Cairo constructed a counterexample to the Mizohata–Takeuchi inequality (and hence to Stein's) for every $C^2$ hypersurface not contained in a hyperplane, with a $\log R$ loss when the weight is restricted to a ball $B(0,R)$, and subsequent work of Cairo–Zhang, Fenves, and Bennett–Kovač–Nakamura–Oliveira produced power-loss failures of the local forms and explicit combinatorial counterexamples on the paraboloid. The paper (Bennett–Gutiérrez–Nakamura–Oliveira) develops a phase-space perspective that predates and survives these refutations. For a solution $u$ of the linear Schrödinger equation $\partial_t u=2\pi i\Delta_x u$ on $\mathbb{R}^d\times\mathbb{R}$ with initial data $u_0\in L^2(\mathbb{R}^d)$ (the paraboloid case of the extension problem), one uses the Wigner distribution $W(g_1,g_2)$, the phase-space density $W(g_1,g_2)(x,v)=\int g_1(x+y/2)\overline{g_2(x-y/2)}e^{-2\pi i v\cdot y}\,dy$ standard in quantum mechanics and time-frequency analysis, which encodes the pair $(g_1,g_2)$ jointly in position $x$ and velocity $v$. The key identity is $\int|u(x,t)|^2w(x,t)\,dx\,dt=\iint W(u_0,u_0)(x,v)\rho^*w(x,v)\,dx\,dv$, where $\rho^*w(x,v)=\int_{\mathbb{R}}w(x-2tv,t)\,dt$ is the X-ray transform of $w$ along the line through $(x,0)$ with direction $(-2v,1)$. Since $W(u_0,u_0)$ integrates in $v$ to $|u_0(x)|^2$ and in $x$ to $|\widehat{u_0}(v)|^2$, this identity suggests the ideal (and now known false) parabolic Mizohata–Takeuchi bound with $\sup_{v\in\operatorname{supp}(\widehat{u_0})}\|\rho^*w(\cdot,v)\|_{L^\infty_x}$ on the right. What the paper proves unconditionally are Sobolev variants: for $s<d/2$ and $\operatorname{supp}(\widehat{u_0})\subseteq B(0;1)$, $\int|u|^2w\lesssim\sup_{v\in\frac12(\operatorname{supp}\widehat{u_0}+\operatorname{supp}\widehat{u_0})}\|\rho^*w(\cdot,v)\|_{\dot H_x^s}\|u_0\|_2^2$, where $\dot H_x^s$ is the homogeneous $L^2$-Sobolev space in the $x$-variable, with norm $\|f\|_{\dot H_x^s}=\|\,|\xi|^s\widehat{f}(\xi)\|_{L^2_\xi}$; the supremum runs over the midpoint hull of the Fourier support, a larger set than the support itself. The proofs rest on bounding $\|W(u_0,u_0)(\cdot,v)\|_{\dot H_x^{-s}}$ by bilinear fractional integrals of $|\widehat{u_0}|^2$, and the arguments work for general strictly convex $S$, for signed weights, and with constants controlled by powers of a curvature quotient $Q(S)$ (not claimed optimal in $n>2$, per Remarks 1.10 and 4.14 of the paper). Two further strands complete the framework. First, for strictly convex planar curves the paper constructs a geometric Wigner distribution $W_S(g,g)$ tomographically from the local intensity $|\widehat{gd\sigma}|^2$ via the $S$-parametrised X-ray transform $X_Sw(u,v)=Xw(N(u),v)$: Proposition 9.1 shows $\lim_{\lambda\to\infty}K(u)(-\Delta_v)^{1/2}X_S(\Phi^\lambda|\widehat{gd\sigma}|^2)(u,v)=W_S(g,g)(u,v)$, where $K$ is the Gaussian curvature, $\Phi^\lambda$ is a smooth cut-off at scale $\lambda$, and $(-\Delta_v)^{1/2}$ is the half-Laplacian (the fractional power one-half of the Laplacian in $v$) implementing the classical X-ray inversion. Second, the phase-space formulation connects to a conjecture of Flandrin from time-frequency analysis, in its weakened modern form $\int_K W(g,g)\lesssim\|g\|_2^2$ uniformly over convex subsets $K$ of phase space (the original constant-1 form was refuted by Delourme–Duyckaerts–Lerner); the paper proves the planar case with an $\varepsilon$-loss, $\int_KW(u_0,u_0)\le C_\varepsilon|K|^\varepsilon\|u_0\|_2^2$ for convex $K\subset\mathbb{R}^2$, and shows this Flandrin-type bound implies the parabolic Mizohata–Takeuchi inequality for a natural class of weights. After the 2025–2026 refutations of the endpoint conjectures, these Sobolev variants, the tomographic construction, and the planar Flandrin-type bound are among the strongest unconditional results in the weighted extension program, and Section 11 of the paper poses exactly the questions that remain open within this framework.

Problem Statement

Resolve the four open questions posed in Section 11 of Bennett–Gutiérrez–Nakamura–Oliveira (2024/2025), which together delimit the remaining open space of the phase-space framework for weighted Fourier extension inequalities. The four questions share one objective — determining how far the framework's unconditional Sobolev-type and tomographic results can be pushed toward the (now refuted) ideal Stein and Mizohata–Takeuchi bounds — but they are largely independent: Section 11 of the paper presents them as a collection of questions, "some concrete and some more speculative", and resolving one neither presupposes nor substitutes for resolving another. Question 11.1 (strengthening the parabolic Sobolev–Mizohata–Takeuchi inequality). Let $u$ solve $\partial_tu=2\pi i\Delta_xu$ on $\mathbb{R}^d\times\mathbb{R}$ with data $u_0\in L^2(\mathbb{R}^d)$, $\operatorname{supp}(\widehat{u_0})\subseteq B(0;1)$, and let $\rho^*w(x,v)=\int_{\mathbb{R}}w(x-2tv,t)\,dt$; write $\|f\|_{\dot H_x^s}=\|\,|\xi|^s\widehat{f}(\xi)\|_{L^2_\xi}$ for the homogeneous Sobolev norm in the $x$-variable. The proven estimate (Theorem 2.8) states that for $s<d/2$ and weights $w$ (signed weights are permitted, per Remark 2.9 of the paper),

$$\int_{\mathbb{R}^d\times\mathbb{R}}|u(x,t)|^2w(x,t)\,dx\,dt\lesssim\sup_{v\in\frac12(\operatorname{supp}(\widehat{u_0})+\operatorname{supp}(\widehat{u_0}))}\|\rho^*w(\cdot,v)\|_{\dot H_x^s}\,\|u_0\|_2^2 .$$

Can this be strengthened, for all nonnegative weights $w$, to the same bound with the supremum taken over the actual Fourier support,

$$\int_{\mathbb{R}^d\times\mathbb{R}}|u(x,t)|^2w(x,t)\,dx\,dt\lesssim\sup_{v\in\operatorname{supp}(\widehat{u_0})}\|\rho^*w(\cdot,v)\|_{\dot H_x^s}\,\|u_0\|_2^2 ,$$

as suggested by the (false in the $L^\infty$-marginal form) ideal inequality (2.9)? Question 11.2 (tomographic Wigner distributions in higher dimensions). For strictly convex smooth planar curves $S$, Proposition 9.1 reconstructs the geometric Wigner distribution $W_S(g,g)$ (the curved analogue of the classical Wigner distribution, adapted to $S$) from the local intensity $|\widehat{gd\sigma}|^2$ via X-ray inversion: $\lim_{\lambda\to\infty}K(u)(-\Delta_v)^{1/2}X_S(\Phi^\lambda|\widehat{gd\sigma}|^2)(u,v)=W_S(g,g)(u,v)$, where $K(u)$ is the Gaussian curvature of $S$ at $u$, $\Phi^\lambda$ is a smooth cut-off localising at scale $\lambda$, and $(-\Delta_v)^{1/2}$ is the half-Laplacian in $v$ implementing the classical X-ray inversion. Is there an analogous tomographic construction of a Wigner distribution that functions in all dimensions, perhaps involving the Radon transform (which integrates a function over hyperplanes and is the natural higher-dimensional analogue of the X-ray transform's line integrals)? Question 11.3 (fractional Stein and Mizohata–Takeuchi inequalities). Do interesting fractional forms of the ideal parabolic inequalities (2.8)/(2.9) arise by replacing the standard phase-space marginal property $\int_{\mathbb{R}^d}W(u_0,u_0)(x,v)\,dv=|u_0(x)|^2$ (equation (2.4) of the paper, the classical marginal property underlying the velocity-averaging representation (2.5) and hence the adjoint X-ray transform $\rho^*$) with an oblique or fractional phase-space marginal of the Wigner distribution (marginals along oblique directions, as in the Radon–Wigner transform, or taken under the fractional Fourier transforms used in phase-space tomography; see Remark 9.2 of the paper)? Question 11.4 (Flandrin-type inequality with $\varepsilon$-loss in higher dimensions). Theorem 10.1 proves that for each $\varepsilon>0$ there exists $C_\varepsilon<\infty$ such that $\int_KW(u_0,u_0)\,dx\,dv\le C_\varepsilon|K|^\varepsilon\|u_0\|_2^2$ for all convex $K\subset\mathbb{R}^2$, where $W$ is the classical Wigner distribution. Does the analogous statement hold in higher dimensions $d>1$, i.e. for convex $K\subset\mathbb{R}^d\times\mathbb{R}^d$?

References

1. Jonathan Bennett, Susana Gutiérrez, Shohei Nakamura, Itamar Oliveira, A phase-space approach to weighted Fourier extension inequalities, Forum of Mathematics, Sigma 13 (2025), article e181, DOI: 10.1017/fms.2025.10127, arXiv:2406.14886, https://doi.org/10.1017/fms.2025.10127
2. Hannah Cairo, A counterexample to the Mizohata-Takeuchi conjecture, preprint (2025), arXiv:2502.06137, https://arxiv.org/abs/2502.06137
3. Hannah Cairo, Ruixiang Zhang, Power loss for the Mizohata-Takeuchi conjecture on C^k convex hypersurfaces, preprint (2025), arXiv:2512.08064, https://arxiv.org/abs/2512.08064
4. Inbo Gottlieb Fenves, Cusp Excursions, Lattice Points on Manifolds, and the Mizohata-Takeuchi Conjecture, preprint (2026), arXiv:2606.27020, https://arxiv.org/abs/2606.27020
5. Jonathan Bennett, Vjekoslav Kovač, Shohei Nakamura, Itamar Oliveira, Rectangles, triangles and Schrödinger waves, preprint (2026), arXiv:2606.30178, https://arxiv.org/abs/2606.30178
6. Anthony Carbery, Marina Iliopoulou, Hong Wang, Some sharp inequalities of Mizohata–Takeuchi-type, Revista Matemática Iberoamericana 40 (2024), 1387–1418, DOI: 10.4171/rmi/1463, arXiv:2302.11877, https://doi.org/10.4171/rmi/1463
7. Jonathan Bennett, Neal Bez, Susana Gutiérrez, Shohei Nakamura, Itamar Oliveira, Mizohata-Takeuchi inequalities for orthonormal systems, Journal of the London Mathematical Society 114 (2026), article e70666, DOI: 10.1112/jlms.70666, arXiv:2506.03783, https://doi.org/10.1112/jlms.70666
8. Anthony Carbery, Zane Kun Li, Yixuan Pang, Po-Lam Yung, A weighted formulation of refined decoupling and inequalities of Mizohata-Takeuchi-type for the moment curve, The Journal of Geometric Analysis 36 (2026), article 48, DOI: 10.1007/s12220-025-02262-3, arXiv:2510.04345, https://doi.org/10.1007/s12220-025-02262-3
9. Jonathan Bennett, Shohei Nakamura, Shobu Shiraki, Tomographic Fourier Extension Identities for Submanifolds of R^n, Selecta Mathematica 30 (2024), article 80, DOI: 10.1007/s00029-024-00970-2, arXiv:2212.12348, https://doi.org/10.1007/s00029-024-00970-2
10. Spyridon Dendrinos, Andrei Mustata, Marco Vitturi, A restricted 2-plane transform related to Fourier Restriction for surfaces of codimension 2, Analysis & PDE 18(2) (2025) 475-526, DOI: 10.2140/apde.2025.18.475, arXiv:2209.15530, https://doi.org/10.2140/apde.2025.18.475
11. Bérangère Delourme, Thomas Duyckaerts, Nicolas Lerner, On integrals over a convex set of the Wigner distribution, Journal of Fourier Analysis and Applications 26 (2020), article 6, DOI: 10.1007/s00041-019-09722-9, arXiv:1901.07262, https://doi.org/10.1007/s00041-019-09722-9
12. Nicolas Lerner, Integrating the Wigner Distribution on subsets of the phase space, a Survey, Memoirs of the European Mathematical Society 12 (2024), DOI: 10.4171/mems/12, arXiv:2102.08090, https://doi.org/10.4171/mems/12