arXiv subjects
Publications and source records attributed to Tuomas Orponen.
I construct a compact subset of the plane whose visible parts are $\tfrac{3}{2}$-dimensional in all directions. This disproves the visibility conjecture. The value $\tfrac{3}{2}$ cannot be increased, as shown in recent collaboration with A. Rutar.
We disprove the visibility conjecture in the plane and prove the sharp upper bound for the almost-sure dimension of visible parts. Precisely, let $K \subset \mathbb{R}^{2}$ be a compact set. For $\sigma \in S^{1}$, let $\mathrm{Vis}_{\sigma}(K) \subset K$ be the visible part of $K$ in direction $\sigma$. We prove that $\operatorname{dim}_{\mathrm{H}} \mathrm{Vis}_{\sigma}(K) \leq \tfrac{3}{2}$ for $\mathcal{H}^{1}$ almost every $\sigma \in S^{1}$. This is sharp: we construct a compact set $K \subset \mathbb{R}^{2}$ such that $\operatorname{dim}_{\mathrm{H}} \mathrm{Vis}_{\sigma}(K)\geq \tfrac{3}{2}$ for all $\sigma \in S^1$.
We prove sharp $\delta$-discretised versions of some variants of the Furstenberg set problem under weaker or different non-concentration assumptions compared to previous works.
We study the planar Nikod\'ym maximal operator $\mathcal{N}_{\Theta;\delta}$ associated to a direction set $\Theta \subset \mathbb{S}^{1}$. We show that the quasi-Assouad dimension $s := \dim_{\mathrm{qA}} \Theta$ characterises the essential $L^{p}$-boundedness of $\mathcal{N}_{\Theta;\delta}$ in the following sense. If $s \in [\tfrac{1}{2},1]$, then $\mathcal{N}_{\Theta;\delta}$ is essentially bounded on $L^{p}(\mathbb{R}^{2})$ for $p \geq 1 + s$, and essentially unbounded for $p < 1 + s$. Here essential boundedness means $L^{p}$-boundedness with constant $O_{\epsilon}(\delta^{-\epsilon})$. We also show that the characterisation described above fails for $s < \tfrac{1}{2}$. More precisely, there exists a set $\Theta \subset \mathbb{S}^{1}$ with $\dim_{\mathrm{qA}} \Theta = \tfrac{1}{3}$ such that $\mathcal{N}_{\Theta;\delta}$ is essentially unbounded on $L^{p}(\mathbb{R}^{2})$ for all $p < \tfrac{3}{2}$. As an application, we show there exists a convex domain with affine dimension $\tfrac{1}{6}$ such that the $\alpha$-order Bochner-Riesz means converge in $L^6$ for all $\alpha>0$.
I prove two variants of the $ABC$ sum-product theorem for $\delta$-separated sets $A,B,C \subset [0,1]$ satisfying Katz-Tao spacing conditions. The main novelty is that the cardinality of the sets $B,C$ need not match their non-concentration exponent. The new $ABC$ theorems are sharp under their respective hypotheses, and imply the previous one.
Uniformly perfect measures are a common generalisation of Ahlfors regular measures, self-conformal measures on the line, and their push-forwards under sufficiently regular maps. We show that every uniformly perfect measure $\sigma$ on a strictly convex planar $C^{2}$-graph is $L^{2}$-flattening. That is, for every $\epsilon>0$, there exists $p = p(\epsilon,\sigma) \geq 1$ such that $$\|\hat{\sigma}\|_{L^{p}(B(R))}^{p} \lesssim_{\epsilon,\sigma} R^{\epsilon}, \qquad R \geq 1.$$
Let $t \in [1,2)$ and $p > 2/(2 - t)$. I construct a $t$-Frostman Borel measure $\mu$ on $[0,1]^{2}$ such that $\pi_{\theta}\mu \notin L^{p}$ for every $\theta \in S^{1}$. This answers a question of Peres and Schlag.
We prove an extension of the Furstenberg set theorem to families of graphs satisfying a transversality condition. We apply the result to derive bounds on $L^{p}$-norms of Fourier transforms of fractal measures supported on plane curves.
Let $\mu$ be the logarithmic equilibrium measure on a compact set $\gamma \subset \mathbb{R}^{d}$. We prove that $\mu$ is absolutely continuous with respect to the length measure on the part of $\gamma$ which can be locally expressed as the graph of a $C^{1,\alpha}$-function $\mathbb{R} \to \mathbb{R}^{d - 1}$, $\alpha > 0$. For $d = 2$, at least in the case where $\gamma$ is a compact $C^{1,\alpha}$-graph, our result can also be deduced from the classical fact that $\mu$ coincides with the harmonic measure of $\Omega =\mathbb{R}^{2} \, \setminus \, \gamma$ with pole at $\infty$. For $d \geq 3$, however, our result is new even for $C^{\infty}$-graphs. In fact, up to now it was not even known if the support of $\mu$ has positive dimension.
We show that the "sharp Kaufman projection theorem" from 2023 is sharp in the class of Ahlfors $(1,\delta^{-\epsilon})$-regular sets. This is in contrast with a recent result of the first author, which improves the projection theorem in the class of Ahlfors $(1,C)$-regular sets.
This paper contains the following $\delta$-discretised projection theorem for Ahlfors regular sets in the plane. For all $C,\epsilon > 0$ and $s \in [0,1]$, there exists $\kappa > 0$ such that the following holds for all $\delta > 0$ small enough. Let $\nu$ be a Borel probability measure on $S^{1}$ satisfying $\nu(B(x,r)) \leq Cr^{\epsilon}$ for all $x \in S^{1}$ and $r > 0$. Let $K \subset B(1) \subset \mathbb{R}^{2}$ be Ahlfors $s$-regular with constant at most $C$. Then, there exists a vector $\theta \in \mathrm{spt\,} \nu$ such that $$|\pi_{\theta}(F)|_{\delta} \geq \delta^{\epsilon - s}$$ for all $F \subset K$ with $|F|_{\delta} \geq \delta^{\kappa - s}$. Here $\pi_{\theta}(z) = \theta \cdot z$ for $z \in \mathbb{R}^{2}$.
Let $s \in [0,1]$. We show that a Borel set $N \subset \mathbb{R}^{2}$ whose every point is linearly accessible by an $s$-dimensional family of lines has Hausdorff dimension at most $2 - s$.
Let $P \subset \mathbb{R}^{2}$ be a Katz-Tao $(\delta,s)$-set, and let $\mathcal{L}$ be a Katz-Tao $(\delta,t)$-set of lines in $\mathbb{R}^{2}$. A recent result of Fu and Ren gives a sharp upper bound for the $\delta$-covering number of the set of incidences $\mathcal{I}(P,\mathcal{L}) = \{(p,\ell) \in P \times \mathcal{L} : p \in \ell\}$. In fact, for $s,t \in (0,1]$, $$ |\mathcal{I}(P,\mathcal{L})|_{\delta} \lesssim_{\epsilon} \delta^{-\epsilon -f(s,t)}, \qquad \epsilon > 0,$$ where $f(s,t) = (s^{2} + st + t^{2})/(s + t)$. For $s,t \in (0,1]$, we characterise the near-extremal configurations $P \times \mathcal{L}$ of this inequality: we show that if $|\mathcal{I}(P,\mathcal{L})|_{\delta} \approx \delta^{-f(s,t)}$, then $P \times \mathcal{L}$ contains "cliques" $P' \times \mathcal{L}'$ satisfying $|\mathcal{I}(P',\mathcal{L}')|_{\delta} \approx |P'|_{\delta}|\mathcal{L}'|_{\delta}$, $$|P'|_{\delta} \approx \delta^{-s^{2}/(s + t)} \quad \text{and} \quad |\mathcal{L}'|_{\delta} \approx \delta^{-t^{2}/(s + t)}.$$
Let $s \in [0,1]$ and $t \in [0,\min\{3s,s + 1\})$. Let $\sigma$ be a Borel measure supported on the parabola $\mathbb{P} = \{(x,x^{2}) : x \in [-1,1]\}$ satisfying the $s$-dimensional Frostman condition $\sigma(B(x,r)) \leq r^{s}$. Answering a question of the first author, we show that there exists an exponent $p = p(s,t) \geq 1$ such that $$\|\hat{\sigma}\|_{L^{p}(B(R))} \leq C_{s,t}R^{(2 - t)/p}, \qquad R \geq 1.$$ Moreover, when $s \geq 2/3$ and $t \in [0,s + 1)$, the previous inequality is true for $p \geq 6$. We also obtain the following fractal geometric counterpart of the previous results. If $K \subset \mathbb{P}$ is a Borel set with $\dim_{\mathrm{H}} K = s \in [0,1]$, and $n \geq 1$ is an integer, then $$ \dim_{\mathrm{H}}(nK) \geq \min\{3s - s \cdot 2^{-(n - 2)},s + 1\}.$$
Let $t \in (1,2)$, and let $B \subset \mathbb{R}^{2}$ be a Borel set with $\dim_{\mathrm{H}} B > t$. I show that $$\mathcal{H}^{1}(\{e \in S^{1} : \dim_{\mathrm{H}} (B \cap \ell_{x,e}) \geq t - 1\}) > 0$$ for all $x \in \mathbb{R}^{2} \, \setminus \, E$, where $\dim_{\mathrm{H}} E \leq 2 - t$. This is the sharp bound for $\dim_{\mathrm{H}} E$. The main technical tool is an incidence inequality of the form $$\mathcal{I}_{\delta}(\mu,\nu) \lesssim_{t} \delta \cdot \sqrt{I_{t}(\mu)I_{3 - t}(\nu)}, \qquad t \in (1,2),$$ where $\mu$ is a Borel measure on $\mathbb{R}^{2}$, and $\nu$ is a Borel measure on the set of lines in $\mathbb{R}^{2}$, and $\mathcal{I}_{\delta}(\mu,\nu)$ measures the $\delta$-incidences between $\mu$ and the lines parametrised by $\nu$. This inequality can be viewed as a $\delta^{-\epsilon}$-free version of a recent incidence theorem due to Fu and Ren. The proof in this paper avoids the high-low method, and the induction-on-scales scheme responsible for the $\delta^{-\epsilon}$-factor in Fu and Ren's work. Instead, the inequality is deduced from the classical smoothing properties of the $X$-ray transform.
One formulation of Marstrand's slicing theorem is the following. Assume that $t \in (1,2]$, and $B \subset \mathbb{R}^{2}$ is a Borel set with $\mathcal{H}^{t}(B) < \infty$. Then, for almost all directions $e \in S^{1}$, $\mathcal{H}^{t}$ almost all of $B$ is covered by lines $\ell$ parallel to $e$ with $\dim_{\mathrm{H}} (B \cap \ell) = t - 1$. We investigate the prospects of sharpening Marstrand's result in the following sense: in a generic direction $e \in S^{1}$, is it true that a strictly less than $t$-dimensional part of $B$ is covered by the heavy lines $\ell \subset \mathbb{R}^{2}$, namely those with $\dim_{\mathrm{H}} (B \cap \ell) > t - 1$? A positive answer for $t$-regular sets $B \subset \mathbb{R}^{2}$ was previously obtained by the first author. The answer for general Borel sets turns out to be negative for $t \in (1,\tfrac{3}{2}]$ and positive for $t \in (\tfrac{3}{2},2]$. More precisely, the heavy lines can cover up to a $\min\{t,3 - t\}$ dimensional part of $B$ in a generic direction. We also consider the part of $B$ covered by the $s$-heavy lines, namely those with $\dim_{\mathrm{H}} (B \cap \ell) \geq s$ for $s > t - 1$. We establish a sharp answer to the question: how much can the $s$-heavy lines cover in a generic direction? Finally, we identify a new class of sets called sub-uniformly distributed sets, which generalise Ahlfors-regular sets. Roughly speaking, these sets share the spatial uniformity of Ahlfors-regular sets, but pose no restrictions on uniformity across different scales. We then extend and sharpen the first author's previous result on Ahlfors-regular sets to the class of sub-uniformly distributed sets.
We prove the following. Let $\mu_{1},\ldots,\mu_{n}$ be Borel probability measures on $[-1,1]$ such that $\mu_{j}$ has finite $s_j$-energy for certain indices $s_{j} \in (0,1]$ with $s_{1} + \ldots + s_{n} > 1$. Then, the multiplicative convolution of the measures $\mu_{1},\ldots,\mu_{n}$ has power Fourier decay: there exists a constant $\tau = \tau(s_{1},\ldots,s_{n}) > 0$ such that \[ \left| \int e^{-2\pi i \xi \cdot x_{1}\cdots x_{n}} \, d\mu_{1}(x_{1}) \cdots \, d\mu_{n}(x_{n}) \right| \leq |\xi|^{-\tau} \] for sufficiently large $|\xi|$. This verifies a suggestion of Bourgain from 2010. We also obtain a quantitative Fourier decay exponent under a stronger assumption on the exponents $s_{j}$.
For $0 \leq s \leq 1$ and $0 \leq t \leq 3$, a set $F \subset \mathbb{R}^{2}$ is called a circular $(s,t)$-Furstenberg set if there exists a family of circles $\mathcal{S}$ of Hausdorff dimension $\dim_{\mathrm{H}} \mathcal{S} \geq t$ such that $$\dim_{\mathrm{H}} (F \cap S) \geq s, \qquad S \in \mathcal{S}.$$ We prove that if $0 \leq t \leq s \leq 1$, then every circular $(s,t)$-Furstenberg set $F \subset \mathbb{R}^{2}$ has Hausdorff dimension $\dim_{\mathrm{H}} F \geq s + t$. The case $s = 1$ follows from earlier work of Wolff on circular Kakeya sets.