The 2002 paper of Muscalu, Tao and Thiele proved uniform LpL^p estimates for nn-linear multiplier forms with modulation symmetry whose symbol is singular on a one-dimensional subspace Γ′=span(v)\Gamma'=\mathrm{span}(v) of the frequency hyperplane, with constants uniform over all degenerations of Γ′\Gamma', but only in the strict local L2L^2 range 2<pi<∞2<p_i<\infty, and it explicitly left two directions open: no analogue of the anisotropic metric dvd_v and its symbol condition is known for singular subspaces of rank k>1k>1, and the uniform bounds have not been extended beyond 2<pi<∞2<p_i<\infty for the general symbol class. Later literature settles the rank-one, bilinear-Hilbert-transform-specific parts: Grafakos and Li (2004) and Li (2006) for uniform bilinear Hilbert transform bounds, Oberlin and Thiele (2011) for the full range in the Walsh model, and Uraltsev and Warchalski (2022, preprint) for the full open range of exponents uniformly over the bilinear Hilbert transform family. Fraccaroli, Saari and Thiele (2025) proved the multivariate trilinear uniform theorem for fixed quasiconformal distortion, uniformly in dilations and rotations, and showed that the missing uniformity in the distortion would imply LpL^p bounds for the triangular Hilbert transform, a central open problem beyond current techniques; that uniformity in KK is a separate open question of their paper, not part of the 2002 question, which concerns nn-linear forms on R\mathbb{R}. The surviving open core is precise and nonempty: identify the intrinsic higher-rank anisotropic symbol class and prove bounds uniform in the degeneration, and extend the uniform bounds for general symbols beyond the strict local L2L^2 range.

Background

Multi-linear multiplier forms. For n≥3n\ge 3 and Schwartz functions f1,…,fnf_1,\dots,f_n on R\mathbb{R}, let Γ={ξ∈Rn: ∑i=1nξi=0}\Gamma=\{\xi\in\mathbb{R}^n:\ \sum_{i=1}^n\xi_i=0\} and let mm be a multiplier on Γ\Gamma. The associated nn-linear form is Λm(f1,…,fn)=∫Γm(ξ)∏i=1nf^i(ξi) dξ,\Lambda_m(f_1,\dots,f_n)=\int_\Gamma m(\xi)\prod_{i=1}^n\hat f_i(\xi_i)\,d\xi, where f^\hat f denotes the Fourier transform. One seeks bounds ∣Λm(f1,…,fn)∣≤C∏i∥fi∥Lpi|\Lambda_m(f_1,\dots,f_n)|\le C\prod_i\|f_i\|_{L^{p_i}} whenever ∑i1/pi=1\sum_i 1/p_i=1; by duality this is the boundedness of an (n−1)(n-1)-linear operator Lp1×⋯×Lpn−1→Lpn′L^{p_1}\times\cdots\times L^{p_{n-1}}\to L^{p_n'}. For Coifman-Meyer symbols, i.e. symbols smooth away from the origin with ∣∂αm(ξ)∣≤C∣ξ∣−∣α∣|\partial^\alpha m(\xi)|\le C|\xi|^{-|\alpha|} (the class governing classical paraproducts, the dyadic-model building blocks used to decompose products of functions), such bounds hold for all 1<pi<∞1<p_i<\infty.

Modulation symmetry. The form Λm\Lambda_m is invariant under modulating each input, fi(x)↦eiλixfi(x)f_i(x)\mapsto e^{i\lambda_i x}f_i(x), because the induced frequency shifts cancel on Γ\Gamma. This symmetry places such operators outside the classical Calderon-Zygmund machinery for singular integrals and is the source of difficulty for the bilinear Hilbert transform (BHT) Hα,β(f,g)(x)=p.v.∫Rf(x−αt)g(x−βt) dt/tH_{\alpha,\beta}(f,g)(x)=\mathrm{p.v.}\int_{\mathbb{R}} f(x-\alpha t)g(x-\beta t)\,dt/t, i.e. the trilinear form with multiplier sgn(β1ξ1+β2ξ2+β3ξ3)\mathrm{sgn}(\beta_1\xi_1+\beta_2\xi_2+\beta_3\xi_3), whose boundedness was proved by Lacey and Thiele.

Singular subspaces and rank. Muscalu, Tao and Thiele (J. Amer. Math. Soc. 15 (2002)) treated multipliers that are singular along a kk-dimensional subspace Γ′⊆Γ\Gamma'\subseteq\Gamma (the rank of the multiplier), satisfying the isotropic condition ∣∂αm(ξ)∣≤C dist(ξ,Γ′)−∣α∣|\partial^\alpha m(\xi)|\le C\,\mathrm{dist}(\xi,\Gamma')^{-|\alpha|}. For 0≤k<n/20\le k<n/2 with Γ′\Gamma' non-degenerate (a graph over every choice of kk coordinates), they proved bounds for all 1<pi<∞1<p_i<\infty; k=0k=0 recovers paraproducts and n=3n=3, k=1k=1 the BHT.

The uniform problem. The constants in that non-degenerate theorem deteriorate as Γ′\Gamma' degenerates: already for the BHT the known constants blow up polynomially when two of the parameters βi\beta_i collide, the regime relevant to Calderon commutators (iterated commutators of the Hilbert transform with multiplication by Lipschitz functions) and the Cauchy integral on Lipschitz curves. The source paper of this problem (Muscalu, Tao and Thiele, J. Anal. Math. 88 (2002)) initiated the uniform theory in the rank-one case: for Γ′=span(v)\Gamma'=\mathrm{span}(v) with all vi≠0v_i\ne 0 and ∑vi=0\sum v_i=0, they introduced the anisotropic metric dv(x,y)=max⁡1≤i≤n∣xi−yi∣/∣vi∣d_v(x,y)=\max_{1\le i\le n}|x_i-y_i|/|v_i| and the symbol condition ∣∂ξαm(ξ)∣≤C∏i=1n(vi dv(ξ,Γ′))−αi,|\partial_\xi^\alpha m(\xi)|\le C\prod_{i=1}^n\bigl(v_i\,d_v(\xi,\Gamma')\bigr)^{-\alpha_i}, and proved (Theorem 1.2) bounds for 2<pi<∞2<p_i<\infty with constants uniform in vv, that is, uniform over the entire degeneration of Γ′\Gamma'. The paper states verbatim that the authors do not know how to modify this condition in the higher rank case k>1k>1, mainly because they have no natural analogue of the metric dvd_v, and that one expects to go beyond the case 2<pi<∞2<p_i<\infty but does not pursue this there.

Modern reformulation and signposts. For trilinear forms on Rd\mathbb{R}^d (functions of dd variables), Fraccaroli, Saari and Thiele reorganized the rank-one theory: the singular set is the image of the diagonal {(τ,τ,τ)}\{(\tau,\tau,\tau)\} under a block matrix L=L1⊕L2⊕L3L=L_1\oplus L_2\oplus L_3, and the natural symbol class allows LL to be KK-quasiconformal, meaning each block satisfies ∥Ln∥opd≤Kdet⁡Ln\|L_n\|_{\mathrm{op}}^{d}\le K\det L_n, where dd is the number of variables of the input functions. Their 2025 theorem gives bounds in the strict local L2L^2 range (all pi>2p_i>2) that are uniform in isotropic dilations and rotations of LL, for fixed KK; uniformity in the distortion parameter KK is not known and is posed by them as a separate open direction, the case d=K=1d=K=1 of their theorem recovering the uniform bilinear Hilbert transform family. Two signposts mark the frontier. Among the partially degenerate configurations in d=2d=2, the twisted paraproduct (a bilinear operator on R2\mathbb{R}^2 of paraproduct type with entangled frequency variables) and the dyadic triangular Hilbert transform with one essentially one-dimensional input are bounded, while the triangular Hilbert transform p.v.∫RF(x+t,y)G(x,y+t)H(x,y) dt/t\mathrm{p.v.}\int_{\mathbb{R}}F(x+t,y)G(x,y+t)H(x,y)\,dt/t has no known LpL^p bounds, and bounds uniform in KK would imply such bounds. On the exponent side, the full open range of exponents is known uniformly for the specific BHT family (Uraltsev and Warchalski 2022, via embedding bounds for iterated outer Lebesgue spaces, a scale of spaces quantifying the size and density of families of wave packets, i.e. functions simultaneously localized in position and frequency; and Oberlin and Thiele 2011 for the Walsh model, a dyadic prototype of the BHT built from quartile rectangles), but for general anisotropic symbols the uniform theorem remains confined to 2<pi<∞2<p_i<\infty.

Problem Statement

Source-faithful statement, following Muscalu, Tao and Thiele, J. Anal. Math. 88 (2002), Theorem 1.2 and the remarks after it.

Let n≥3n\ge 3, let Γ={ξ∈Rn: ∑i=1nξi=0}\Gamma=\{\xi\in\mathbb{R}^n:\ \sum_{i=1}^n\xi_i=0\}, and, for a function mm on Γ\Gamma (the symbol), let Λm(f1,…,fn)=∫Γm(ξ)∏i=1nf^i(ξi) dξ\Lambda_m(f_1,\dots,f_n)=\int_\Gamma m(\xi)\prod_{i=1}^n\hat f_i(\xi_i)\,d\xi be the nn-linear multiplier form on Schwartz functions on R\mathbb{R}, where f^\hat f denotes the Fourier transform. One seeks strong-type bounds ∣Λm(f1,…,fn)∣≤C∏i=1n∥fi∥Lpi|\Lambda_m(f_1,\dots,f_n)|\le C\prod_{i=1}^n\|f_i\|_{L^{p_i}} under the scaling condition ∑i=1n1/pi=1\sum_{i=1}^n 1/p_i=1. The objective is to complete the uniform theory of Theorem 1.2 of the source paper by removing its two stated limitations; the two components below remove the two limitations of the same theorem.

(1) Higher rank. Theorem 1.2 concerns symbols singular on a one-dimensional subspace Γ′=span(v)⊆Γ\Gamma'=\mathrm{span}(v)\subseteq\Gamma, where v=(v1,…,vn)v=(v_1,\dots,v_n) has all coordinates nonzero and summing to zero; writing dv(x,y)=max⁡1≤i≤n∣xi−yi∣/∣vi∣d_v(x,y)=\max_{1\le i\le n}|x_i-y_i|/|v_i| and dv(ξ,Γ′)=inf⁡η∈Γ′dv(ξ,η)d_v(\xi,\Gamma')=\inf_{\eta\in\Gamma'}d_v(\xi,\eta), its hypothesis is the anisotropic symbol condition ∣∂ξαm(ξ)∣≤C∏i=1n(vi dv(ξ,Γ′))−αi|\partial_\xi^\alpha m(\xi)|\le C\prod_{i=1}^n\bigl(v_i\,d_v(\xi,\Gamma')\bigr)^{-\alpha_i} for all partial derivatives up to some finite order, and its conclusion is a strong-type bound for 2<pi<∞2<p_i<\infty with a constant CC uniform over all such vv. For a singular subspace Γ′⊆Γ\Gamma'\subseteq\Gamma of dimension k>1k>1 (the mandatory rank range is k<n/2k<n/2, where fixed-subspace boundedness is known for non-degenerate Γ′\Gamma' and only the uniformity is missing; the degenerate boundary k≥n/2k\ge n/2, which contains the multilinear Hilbert transforms, the higher-input analogues of the bilinear Hilbert transform, lies outside the required bound — even the fixed-subspace LpL^p estimates are unknown there — and is mentioned only to delimit how far uniformity extends), identify the intrinsic anisotropic symbol condition on mm — the analogue for k>1k>1 of the metric dvd_v and the displayed condition — and prove that under this condition the strong-type bound holds for 2<pi<∞2<p_i<\infty with CC independent of the position of Γ′\Gamma' as Γ′\Gamma' approaches degenerate configurations, in the same sense in which Theorem 1.2 is uniform over all vectors vv with nonzero coordinates summing to zero. A solution must recover the known special cases as specialisations of the new anisotropic condition within this framework: specialising to the isotropic condition at fixed non-degenerate Γ′\Gamma' must give the non-degenerate theorem (1<pi<∞1<p_i<\infty), and specialising to the dvd_v condition at k=1k=1 must give the rank-one uniform Theorem 1.2 (2<pi<∞2<p_i<\infty); the exponent range beyond 2<pi<∞2<p_i<\infty belongs to component (2), not to this component's new claim.

(2) Exponent range. For the rank-one anisotropic symbol class of Theorem 1.2, prove the uniform bounds for the full range 1<pi<∞1<p_i<\infty known in the non-degenerate theory, with constants still independent of the degeneration parameter. (For the specific BHT family this is settled in the full range by Uraltsev and Warchalski, 2022, preprint; the open part is the general symbol class. In the related multivariate quasiconformal setting of Fraccaroli, Saari and Thiele the same strict-local-L2L^2 restriction persists and is noted by them to be likely not sharp.)

The source paper additionally noted that the endpoint ∑i=1n−1θi=2\sum_{i=1}^{n-1}\theta_i=2 of its localized tree estimate (Proposition 7.1) is plausibly attainable; that technical sub-question belongs to the same program but is tied to the 2002 proof architecture and is not an independent objective here. The problem concerns the entire degenerating family and the general symbol class of Theorem 1.2; the Walsh model is a dyadic prototype of the bilinear Hilbert transform and does not by itself settle the general symbol case.

References

  1. Camil Muscalu, Terence Tao, Christoph Thiele. Uniform estimates on multi-linear operators with modulation symmetry. Journal d'Analyse Mathématique 88 (2002), 255-309. DOI: 10.1007/BF02786579. https://doi.org/10.1007/BF02786579 (arXiv:math/0107059, https://arxiv.org/abs/math/0107059). Source paper of the open problem.
  2. Camil Muscalu, Terence Tao, Christoph Thiele. Multi-linear operators given by singular multipliers. Journal of the American Mathematical Society 15 (2002), no. 2, 469-496. DOI: 10.1090/S0894-0347-01-00379-4. https://doi.org/10.1090/S0894-0347-01-00379-4 (arXiv:math/9910039, https://arxiv.org/abs/math/9910039).
  3. Michael Lacey, Christoph Thiele. L^p estimates on the bilinear Hilbert transform for 2 < p < infinity. Annals of Mathematics (2) 146 (1997), no. 3, 693-724. DOI: 10.2307/2952458. https://doi.org/10.2307/2952458.
  4. Loukas Grafakos, Xiaochun Li. Uniform bounds for the bilinear Hilbert transforms, I. Annals of Mathematics (2) 159 (2004), 889-933. DOI: 10.4007/annals.2004.159.889. https://doi.org/10.4007/annals.2004.159.889.
  5. Xiaochun Li. Uniform bounds for the bilinear Hilbert transforms, II. Revista Matemática Iberoamericana 22 (2006), no. 3, 1069-1126. DOI: 10.4171/RMI/483. https://doi.org/10.4171/rmi/483.
  6. Richard Oberlin, Christoph Thiele. New uniform bounds for a Walsh model of the bilinear Hilbert transform. Indiana University Mathematics Journal 60 (2011), no. 5, 1693-1712. DOI: 10.1512/iumj.2011.60.4445. https://doi.org/10.1512/iumj.2011.60.4445 (arXiv:1004.4019, https://arxiv.org/abs/1004.4019).
  7. Gennady Uraltsev, Michał Warchalski. The full range of uniform bounds for the bilinear Hilbert transform. arXiv:2205.09851 (2022). https://arxiv.org/abs/2205.09851. Preprint.
  8. Marco Fraccaroli, Olli Saari, Christoph Thiele. Phase space localizing operators. arXiv:2210.16164 (2022). https://arxiv.org/abs/2210.16164.
  9. Marco Fraccaroli, Olli Saari, Christoph Thiele. Uniform bounds for bilinear symbols with linear K-quasiconformally embedded singularity. Analysis & PDE 18 (2025), no. 9, 2293-2323. DOI: 10.2140/apde.2025.18.2293. https://doi.org/10.2140/apde.2025.18.2293 (arXiv:2402.11661, https://arxiv.org/abs/2402.11661).
  10. Ciprian Demeter, Christoph Thiele. On the two-dimensional bilinear Hilbert transform. American Journal of Mathematics 132 (2010), no. 1, 201-256. DOI: 10.1353/ajm.0.0101. https://doi.org/10.1353/ajm.0.0101 (arXiv:0803.1268, https://arxiv.org/abs/0803.1268).
  11. Vjekoslav Kovač. Boundedness of the twisted paraproduct. Revista Matemática Iberoamericana 28 (2012), no. 4, 1143-1164. DOI: 10.4171/RMI/707. https://doi.org/10.4171/rmi/707 (arXiv:1011.6140, https://arxiv.org/abs/1011.6140).
  12. Vjekoslav Kovač, Christoph Thiele, Pavel Zorin-Kranich. Dyadic triangular Hilbert transform of two general functions and one not too general function. Forum of Mathematics, Sigma 3 (2015), e25. DOI: 10.1017/fms.2015.25. https://doi.org/10.1017/fms.2015.25 (arXiv:1506.00861, https://arxiv.org/abs/1506.00861).
  13. Michael Christ, Polona Durcik, Joris Roos. Trilinear smoothing inequalities and a variant of the triangular Hilbert transform. Advances in Mathematics 390 (2021), 107863. DOI: 10.1016/j.aim.2021.107863. https://doi.org/10.1016/j.aim.2021.107863 (arXiv:2008.10140, https://arxiv.org/abs/2008.10140).
  14. Olli Saari, Christoph Thiele. Paraproducts for bilinear multipliers associated with convex sets. Mathematische Annalen 385 (2023), no. 3-4, 2013-2036. DOI: 10.1007/s00208-022-02397-6. https://doi.org/10.1007/s00208-022-02397-6 (arXiv:2102.10305, https://arxiv.org/abs/2102.10305).
  15. Valentina Ciccone. A note on bilinear multipliers with convex singularities. arXiv:2505.04517 (2025). https://arxiv.org/abs/2505.04517.
  16. Jianghao Zhang. Non-degenerate multilinear singular multipliers with fractional rank. arXiv:2503.14635 (2025). https://arxiv.org/abs/2503.14635.
  17. Lars Becker, Polona Durcik, Fred Yu-Hsiang Lin. On trilinear singular Brascamp-Lieb integrals. arXiv:2411.00141 (2024). https://arxiv.org/abs/2411.00141.
  18. Camil Muscalu, Terence Tao, Christoph Thiele. Multi-linear multipliers associated to simplexes of arbitrary length. arXiv:0712.2420 (2007). https://arxiv.org/abs/0712.2420. Published as Chapter 14 of Advances in Analysis: The Legacy of Elias M. Stein, Princeton University Press (2014), pp. 346-401. DOI: 10.1515/9781400848935-015. https://doi.org/10.1515/9781400848935-015.
  19. Fred Yu-Hsiang Lin, Lenka Slavíková. Rough averages of triangular Hilbert transforms. arXiv:2607.20206 (2026). https://arxiv.org/abs/2607.20206.