Non-Degenerate Multilinear Singular Multipliers with Fractional Rank
Jianghao Zhang · 2025
Read Non-Degenerate Multilinear Singular Multipliers with Fractional Rank by Jianghao Zhang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-14635.txt.
Introduction
Jianghao Zhang, writing alone in 2025, proves bounds for a class of multi-input singular operators in higher dimensions. The operators act on groups of functions, and their defining rule is well behaved away from a flat subspace. The hard part is that proof gets tougher as the size of that subspace grows relative to the whole space.
Zhang focuses on the case where that size ratio falls in a narrow, awkward band between two landmarks. In that band he establishes the norm bounds the field seeks. The paper is short, dense, and aimed at specialists in harmonic analysis. Its value lies in closing a gap in the known range of cases for these operators.
Catalog record
| CREATOR | Jianghao Zhang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 149 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-14635
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Non-Degenerate Multilinear Singular Multipliers with Fractional Rank Jianghao Zhang Abstract We establish L p L^{p} estimates for multilinear multipliers acting on ( n − 1 ) (n-1) -tuples of functions on ℝ d \mathbb{R}^{d} . We assume that the multiplier satisfies symbol estimates outside a linear subspace of dimension m m . The difficulty of proving L p L^{p} bounds increases with the rank m d \frac{m}{d} . Such bounds are unachievable using existing tools when m d ≥ n 2 \frac{m}{d}\geq\frac{n}{2} . And when ⌈ m d ⌉ < n 2 \lceil\frac{m}{d}\rceil<\frac{n}{2} , our results do not improve upon those in [ 22 ] . Therefore, our focus is on the fractional rank case m d < n 2 ≤ ⌈ m d ⌉ \frac{m}{d}<\frac{n}{2}\leq\lceil\frac{m}{d}\rceil proposed in [ 4 ] . We extend the results of [ 4 ] by covering a broader range of Lebesgue exponents. Additionally, we develop the tree counting argument, thereby including cases with larger m d \frac{m}{d} compared to [ 4 ] . In doing so, we also identify and correct an inaccuracy in [ 4 ] . 1 Introduction We use \xi_{j}=(\xi_{j,1},\cdots,\xi_{j,d})\in\mathbb{R}^{d} and \xi:=(\xi^{{}^{\prime}},\xi_{n}):=(\xi_{1},\cdots,\xi_{n})\in(\mathbb{R}^{d})^{n} for frequency variables. We will consider the boundedness of multilinear multipliers of the form: \int_{\Gamma_{0}}\mathfrak{m}(\xi_{1},\cdots,\xi_{n})\widehat{f_{1}}(\xi_{1})\cdots\widehat{f_{n}}(\xi_{n}). (1.1) Here the function \mathfrak{m} , defined on \Gamma_{0}:=\{\xi\in(\mathbb{R}^{d})^{n}:\sum_{j=1}^{n}\xi_{j}=0\} , is singular on a subspace \Gamma\subset\Gamma_{0} . The precise definition is given as follows: Definition 1 . Let \Gamma\subset\Gamma_{0} be a linear subspace and \mathfrak{m}:\Gamma_{0}\rightarrow\mathbb{C} be a function satisfying |\partial^{\alpha}\mathfrak{m}(\xi)|\lesssim_{\alpha}d(\xi,\Gamma)^{-|\alpha|},\ \forall\alpha:|\alpha|\leq N_{2} (1.2) for some large N_{2} . Let f_{1},\cdots,f_{n-1}\in\mathcal{S}(\mathbb{R}^{d}) . Define \begin{split}T_{\mathfrak{m}}&(f_{1},\cdots,f_{n-1})(x)\\ &:=\int_{(\mathbb{R}^{d})^{n-1}}e^{2\pi ix\cdot(\sum_{j=1}^{n-1}\xi_{j})}\mathfrak{m}(\xi_{1},\cdots,\xi_{n-1},-\sum_{j=1}^{n-1}\xi_{j})\widehat{f_{1}}(\xi_{1})\cdots\widehat{f_{n-1}}(\xi_{n-1})d\xi_{1}\cdots d\xi_{n-1}.\end{split} The associated multilinear form is given by \begin{split}\Lambda_{\mathfrak{m}}(f_{1},\cdots,f_{n})&:=\langle T_{\mathfrak{m}}(f_{1},\cdots,f_{n-1}),f_{n}\rangle\\ &=\int_{(\mathbb{R}^{d})^{n-1}}\mathfrak{m}(\xi_{1},\cdots,\xi_{n-1},-\sum_{j=1}^{n-1}\xi_{j})\widehat{f_{1}}(\xi_{1})\cdots\widehat{f_{n-1}}(\xi_{n-1})\widehat{f_{n}}(-\sum_{j=1}^{n-1}\xi_{j})d\xi_{1}\cdots d\xi_{n-1},\end{split} where the last line also serves as the definition of ( 1.1 ). Remark . When integrating over \Gamma_{0} , we always use the normalized Lebesgue measure as above. There has been a long interest in studying the boundedness of T_{\mathfrak{m}} . The decay condition ( 1.2 ) makes T_{\mathfrak{m}} have modulation symmetry along \Gamma , so tools from time-frequency analysis are particularly useful. This field can be traced back to the classical works on pointwise convergence of Fourier series [ 3 ] [ 11 ] . Then seminal breakthroughs [ 16 ] [ 17 ] , made by Lacey and Thiele, established boundedness of the bilinear Hilbert transform. It is the first non-trivial example of T_{\mathfrak{m}} with \dim(\Gamma)>0 . Their method also led to a new short proof of Carleson’s theorem [ 18 ] . See [ 8 ] , [ 26 ] , [ 6 ] , [ 14 ] [ 10 ] , [ 2 ] , [ 20 ] [ 28 ] , and references therein for some recent developments. The systematic study for general T_{\mathfrak{m}} with d=1 originated in [ 22 ] . For general d , the quantity \frac{m}{d} measures the complexity of our operators. We call this quantity rank, and it can be a fractional number when d\geq 2 . We only consider non-degenerate \Gamma such that the rank is strictly smaller than \frac{n}{2} , because this is the range where our existing time-frequency tools work well. Now let’s give some definitions about the non-degeneracy of \Gamma . There are two types of non-degenerate conditions. The first can be traced back to [ 22 ] , which is sufficient for the case \lceil\frac{m}{d}\rceil<\frac{n}{2} . The second was introduced in [ 4 ] to handle fractional rank operators such that \frac{m}{d}<\frac{n}{2}\leq\lceil\frac{m}{d}\rceil . Since the \lceil\frac{m}{d}\rceil<\frac{n}{2} case was essentially solved in [ 22 ] , we will mainly consider the range \frac{m}{d}<\frac{n}{2}\leq\lceil\frac{m}{d}\rceil . This case requires both non-degenerate conditions. Definition 2 . Given \Theta\subset[n]\times[d] , consider the projection \Pi_{\Theta} from \Gamma_{0} to the coordinates (\xi_{j,k})_{(j,k)\in\Theta} . Let v\in\Gamma_{0} . Write (v)_{\Theta} for \Pi_{\Theta}v . For A\subset[n] , consider \Pi_{\Theta} with \Theta:=A\times[d] , let P_{A} denote \Pi_{\Theta} restriced on \Gamma . Remark . Using this notation, We interpret ( 1.2 ) as follows: Regard \mathfrak{m} as a function of \xi^{{}^{\prime}}:=(\xi_{1},\cdots,\xi_{n-1}) , and ( 1.2 ) means |\partial^{\alpha}\mathfrak{m}(\xi^{{}^{\prime}})|\lesssim_{\alpha}d(\xi^{{}^{\prime}},P_{[n-1]}\Gamma)^{-|\alpha|},\ \forall\alpha:|\alpha|\leq N. (1.3) In fact, assume ( 1.3 ) holds. Then the same type of estimates holds if we regard \mathfrak{m} as the function of any n-1 variables. Thus we use \Gamma_{0},\Gamma and write ( 1.3 ) as ( 1.2 ) to emphasize that \xi_{1},\cdots,\xi_{n} are at symmetric positions. Definition 3 (Type I) . Assume \frac{m}{d}<\frac{n}{2} . We say \Gamma satisfies Type I non-degenerate condition if for any A\subset[n],\#A=\lceil\frac{m}{d}\rceil , we have \dim\Big(P_{A}(\Gamma)\Big)=m. Remark . This means that \Gamma is the graph over the coordinates (\xi_{j,k})_{(j,k)\in\Theta} for some \Theta\subset A\times[d] such that \#\Theta=m . When \frac{m}{d}<\frac{n}{2}\leq\lceil\frac{m}{d}\rceil , there are two possibilities. The first is the case \frac{m}{d}\leq\frac{n-1}{2} . This situation was handled in [ 4 ] , and we call it the medium rank case. The second is the case \frac{n-1}{2}<\frac{m}{d} . Call this situation the large rank case. The large rank case was claimed to be solved in [ 4 ] , but the non-degenerate condition there holds for no \Gamma . For example, consider (41) in [ 4 ] . Let all terms on the right to be zero and \overrightarrow{\xi}^{(1)}=\cdots=\overrightarrow{\xi}^{(d)} . We see the number of remaining variables are more than the number of remaining equations, so there exist non-zero solutions. We will extend the result for the medium rank case in [ 4 ] and also tackle the large rank case. However, for the large rank case, Our argument only works well when n is large in terms of d . If n is small, the operator can satisfy a large group of symmetry like the triangular Hilbert transform. See Example 5 for more details. Definition 4 (Type II) . Assume \frac{m}{d}<\frac{n}{2}\leq\lceil\frac{m}{d}\rceil . Consider the linear map \mathfrak{L} defined on \Gamma^{L} which maps (v^{(1)},\cdots,v^{(L)}) to \Big(\big(P_{A}(v^{(1)})-P_{A}(v^{(k)})\big)_{2\leq k\leq L},\big(P_{B^{(k)}}(v^{(k)})\big)_{1\leq k\leq L}\Big). (1.4) We define L , A,B^{(1)},\cdots,B^{(L)} and Type II non-degeneracy as follows: • When \frac{m}{d}\leq\frac{n-1}{2} , let L:=2 . Note that n must be an even integer. Write n=2n^{{}^{\prime}}+2 . Let A,B^{(1)},B^{(2)} be any disjoint subsets of [n] such that \#A=1,\#B^{(1)}=\#B^{(2)}=n^{{}^{\prime}} . We say \Gamma satisfies Type II non-degenerate condition if the linear map \mathfrak{L} satisfies \ker(\mathfrak{L})=\{0\} for any possible choice of A,B^{(1)},\cdots,B^{(L)} . • When \frac{n-1}{2}<\frac{m}{d} , let L:=d . Write m=ld+r,0\leq r\leq d-1 . Further, write l+1-(d-r)=ad+b,0\leq b\leq d-1 . Let A,U_{1},\cdots,U_{2d-1},W be any disjoint subsets of [n] such that \#A=d-r,\#U_{1}=\cdots=\#U_{2d-1}=a,\#W=b . Define B^{(k)}:=(U_{k}\cup\cdots\cup U_{k+d-1})\cup W,\ 1\leq k\leq d. We say \Gamma satisfies Type II non-degenerate condition if l+1-(d-r)\geq 0 and the linear map \mathfrak{L} satisfies \ker(\mathfrak{L})=\{0\} for any possible choice of A,B^{(1)},\cdots,B^{(L)} . These conditions look complicated, but they hold for almost all \Gamma in the following sense. Definition 5 . Fix m\in\mathbb{N}_{+} . Consider \Gamma_{0}^{m} and the product of normalized Lebesgue measures on it. If a property about tuples (v_{1},\cdots,v_{m})\in\Gamma_{0}^{m} is true outside a zero-measure set, then we say it holds for generic tuples. Proposition 1 . Let n,d,m\in\mathbb{N}_{+},n\geq 3 such that \frac{m}{d}<\frac{n}{2} . Then for generic tuples (v_{1},\cdots,v_{m})\in\Gamma_{0}^{m} , the linear space \Gamma:=\spn\{v_{1},\cdots,v_{m}\} has dimension m and satisfies Type I non-degenerate condition. Moreover, assume \frac{m}{d}<\frac{n}{2}\leq\lceil\frac{m}{d}\rceil . When \frac{m}{d}\leq\frac{n-1}{2} , \Gamma also satisfies Type II non-degenerate condition for generic tuples (v_{1},\cdots,v_{m})\in\Gamma_{0}^{m} . When \frac{n-1}{2}<\frac{m}{d} and n>10d^{3} , \Gamma also satisfies Type II non-degenerate condition for generic tuples (v_{1},\cdots,v_{m})\in\Gamma_{0}^{m}