I have just uploaded to the arXiv my paper “Sharp bounds for multilinear curved Kakeya, restriction and oscillatory integral estimates away from the endpoint“, submitted to Mathematika. In this paper I return (after more than a decade’s absence) to one of my first research interests, namely the Kakeya and restriction family of conjectures. The starting point is the following “multilinear Kakeya estimate” first established in the non-endpoint case by Bennett, Carbery, and myself, and then in the endpoint case by Guth (with further proofs and extensions by Bourgain-Guth and Carbery-Valdimarsson:

Theorem 1 (Multilinear Kakeya estimate) Let ${\delta > 0}$ be a radius. For each ${j = 1,\dots,d}$, let ${\mathbb{T}_j}$ denote a finite family of infinite tubes ${T_j}$ in ${{\bf R}^d}$ of radius ${\delta}$. Assume the following axiom:

• (i) (Transversality) whenever ${T_j \in \mathbb{T}_j}$ is oriented in the direction of a unit vector ${n_j}$ for ${j =1,\dots,d}$, we have

$\displaystyle \left|\bigwedge_{j=1}^d n_j\right| \geq A^{-1}$

for some ${A>0}$, where we use the usual Euclidean norm on the wedge product ${\bigwedge^d {\bf R}^d}$.

Then, for any ${p \geq \frac{1}{d-1}}$, one has

$\displaystyle \left\| \prod_{j=1}^d \sum_{T_j \in \mathbb{T}_j} 1_{T_j} \right\|_{L^p({\bf R}^d)} \lesssim_{A,p} \delta^{\frac{d}{p}} \prod_{j \in [d]} \# \mathbb{T}_j. \ \ \ \ \ (1)$

where ${L^p({\bf R}^d)}$ are the usual Lebesgue norms with respect to Lebesgue measure, ${1_{T_j}}$ denotes the indicator function of ${T_j}$, and ${\# \mathbb{T}_j}$ denotes the cardinality of ${\mathbb{T}_j}$.

The original proof of this proceeded using a heat flow monotonicity method, which in my previous post I reinterpreted using a “virtual integration” concept on a fractional Cartesian product space. It turns out that this machinery is somewhat flexible, and can be used to establish some other estimates of this type. The first result of this paper is to extend the above theorem to the curved setting, in which one localises to a ball of radius ${O(1)}$ (and sets ${\delta}$ to be small), but allows the tubes ${T_j}$ to be curved in a ${C^2}$ fashion. If one runs the heat flow monotonicity argument, one now picks up some additional error terms arising from the curvature, but as the spatial scale approaches zero, the tubes become increasingly linear, and as such the error terms end up being an integrable multiple of the main term, at which point one can conclude by Gronwall’s inequality (actually for technical reasons we use a bootstrap argument instead of Gronwall). A key point in this approach is that one obtains optimal bounds (not losing factors of ${\delta^{-\varepsilon}}$ or ${\log^{O(1)} \frac{1}{\delta}}$), so long as one stays away from the endpoint case ${p=\frac{1}{d-1}}$ (which does not seem to be easily treatable by the heat flow methods). Previously, the paper of Bennett, Carbery, and myself was able to use an induction on scale argument to obtain a curved multilinear Kakeya estimate losing a factor of ${\log^{O(1)} \frac{1}{\delta}}$ (after optimising the argument); later arguments of Bourgain-Guth and Carbery-Valdimarsson, based on algebraic topology methods, could also obtain a curved multilinear Kakeya estimate without such losses, but only in the algebraic case when the tubes were neighbourhoods of algebraic curves of bounded degree.

Perhaps more interestingly, we are also able to extend the heat flow monotonicity method to apply directly to the multilinear restriction problem, giving the following global multilinear restriction estimate:

Theorem 2 (Multilinear restriction theorem) Let ${\frac{1}{d-1} < p \leq \infty}$ be an exponent, and let ${A \geq 2}$ be a parameter. Let ${M}$ be a sufficiently large natural number, depending only on ${d}$. For ${j \in [d]}$, let ${U_j}$ be an open subset of ${B^{d-1}(0,A)}$, and let ${h_j: U_j \rightarrow {\bf R}}$ be a smooth function obeying the following axioms:

• (i) (Regularity) For each ${j \in [d]}$ and ${\xi \in U_j}$, one has

$\displaystyle |\nabla_\xi^{\otimes m} \otimes h_j(\xi)| \leq A \ \ \ \ \ (2)$

for all ${1 \leq m \leq M}$.

• (ii) (Transversality) One has

$\displaystyle \left| \bigwedge_{j \in [d]} (-\nabla_\xi h_j(\xi_j),1) \right| \geq A^{-1}$

whenever ${\xi_j \in U_j}$ for ${j \in [d]}$.

Let ${U_{j,1/A} \subset U_j}$ be the sets

$\displaystyle U_{j,1/A} := \{ \xi \in U_j: B^{d-1}(\xi,1/A) \subset U_j \}. \ \ \ \ \ (3)$

Then one has

$\displaystyle \left\| \prod_{j \in [d]} {\mathcal E}_j f_j \right\|_{L^{2p}({\bf R}^d)} \leq A^{O(1)} \left(d-1-\frac{1}{p}\right)^{-O(1)} \prod_{j \in [d]} \|f_j \|_{L^2(U_{j,1/A})}$

for any ${f_j \in L^2(U_{j,1/A} \rightarrow {\bf C})}$, ${j \in [d]}$, extended by zero outside of ${U_{j,1/A}}$, and ${{\mathcal E}_j}$ denotes the extension operator

$\displaystyle {\mathcal E}_j f_j( x', x_d ) := \int_{U_j} e^{2\pi i (x' \xi^T + x_d h_j(\xi))} f_j(\xi)\ d\xi.$

Local versions of such estimate, in which ${L^{2p}({\bf R}^d)}$ is replaced with ${L^{2p}(B^d(0,R))}$ for some ${R \geq 2}$, and one accepts a loss of the form ${\log^{O(1)} R}$, were already established by Bennett, Carbery, and myself using an induction on scale argument. In a later paper of Bourgain-Guth these losses were removed by “epsilon removal lemmas” to recover Theorme 2, but only in the case when all the hypersurfaces involved had curvatures bounded away from zero.

There are two main new ingredients in the proof of Theorem 2. The first is to replace the usual induction on scales scheme to establish multilinear restriction by a “ball inflation” induction on scales scheme that more closely resembles the proof of decoupling theorems. In particular, we actually prove the more general family of estimates

$\displaystyle \left\| \prod_{j \in [d]} E_{r}[{\mathcal E}_j f_j] \right\|_{L^{p}({\bf R}^d)} \leq A^{O(1)} \left(d-1 - \frac{1}{p}\right)^{O(1)} r^{\frac{d}{p}} \prod_{j \in [d]} \| f_j \|_{L^2(U_{j,1/A})}^2$

where ${E_r}$ denotes the local energies

$\displaystyle E_{r}[f](x',x_d) := \int_{B^{d-1}(x',r)} |f(y',x_d)|^2\ dy'$

(actually for technical reasons it is more convenient to use a smoother weight than the strict cutoff to the disk ${B^{d-1}(x',r)}$). With logarithmic losses, it is not difficult to establish this estimate by an upward induction on ${r}$. To avoid such losses we use the heat flow monotonicity method. Here we run into the issue that the extension operators ${{\mathcal E}_j f_j}$ are complex-valued rather than non-negative, and thus would not be expected to obey many good montonicity properties. However, the local energies ${E_r[{\mathcal E}_j f_j]}$ can be expressed in terms of the magnitude squared of what is essentially the Gabor transform of ${{\mathcal E}_j f_j}$, and these are non-negative; furthermore, the dispersion relation associated to the extension operators ${{\mathcal E}_j f_j}$ implies that these Gabor transforms propagate along tubes, so that the situation becomes quite similar (up to several additional lower order error terms) to that in the multilinear Kakeya problem. (This can be viewed as a continuous version of the usual wave packet decomposition method used to relate restriction and Kakeya problems, which when combined with the heat flow monotonicity method allows for one to use a continuous version of induction on scales methods that do not concede any logarithmic factors.)

Finally, one can combine the curved multilinear Kakeya result with the multilinear restriction result to obtain estimates for multilinear oscillatory integrals away from the endpoint. Again, this sort of implication was already established in the previous paper of Bennett, Carbery, and myself, but the arguments there had some epsilon losses in the exponents; here we were able to run the argument more carefully and avoid these losses.