A note on limsup sets of annuli
Mumtaz Hussain, Benjamin Ward · 2025
Read A note on limsup sets of annuli by Mumtaz Hussain, Benjamin Ward (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-2502-09807.txt.
Introduction
Mumtaz Hussain and Benjamin Ward measure the size of point sets caught by infinitely many shrinking rings around rationals, in work from twenty twenty-five. Each ring, or annulus, sits in the max-norm around a rational centre. Points landing in endless rings form the limsup set studied. Jarnik-style theorems give its Hausdorff dimension in exact terms. Ring thickness and outer radius decay rates drive the answer.
Slowly shrinking outer radii with fast-thinning rings push dimension near n minus one. Rectangular and mixed-norm ring variants receive parallel treatment. A scaling lemma and a rectangles transference principle power proofs. Number theory's fine geometry gains new sharp benchmarks.
Catalog record
| CREATOR | Mumtaz Hussain, Benjamin Ward |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 43 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-09807
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A note on limsup sets of annuli Mumtaz Hussain, Benjamin Ward Abstract . We consider the set of points in infinitely many max-norm annuli centred at rational points in ℝ n {\mathbb{R}}^{n} . We give Jarník-Besicovitch type theorems for this set in terms of Hausdorff dimension. Interestingly, we find that if the outer radii are decreasing sufficiently slowly, dependent only on the dimension n n , and the thickness of the annuli is decreasing rapidly then the dimension of the set tends towards n − 1 n-1 . We also consider various other forms of annuli including rectangular annuli and quasi-annuli described by the difference between balls of two different norms. Our results are deduced through a novel combination of a version of Cassel’s Scaling Lemma and a generalisation of the Mass Transference Principle, namely the Mass transference principle from rectangles to rectangles due to Wang and Wu (Math. Ann. 2021). 1. Introduction A fundamental set in Diophantine approximation is the set of \psi -approximable points. Throughout let \psi:{\mathbb{N}}\to{\mathbb{R}}_{+} be a monotonically decreasing function such that \psi(q)\to 0 as q\to\infty . Define the set of \psi -approximable points to be W_{n}(\psi):=\left\{\mathbf{x}\in[0,1]^{n}:\left\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\right\|<\tfrac{\psi(q)}{q}\quad\text{ for i.m. }\,(\mathbf{p},q)\in{\mathbb{Z}}^{n}\times{\mathbb{N}}\right\}\,, where \|\cdot\| denotes the maximum norm on {\mathbb{R}}^{n} and ‘i.m.’ denotes ‘infinitely many’. It is well known that W_{n}(\psi) can be written as a \limsup set of balls centred at rational points \tfrac{\mathbf{p}}{q} and with radii \tfrac{\psi(q)}{q} . Much is known on the set W_{n}(\psi) , for example Jarník [ 13 ] and Besicovitch [ 5 ] independently proved that, for \psi(q)=q^{-\tau_{\psi}} with \tau\geq\tfrac{1}{n} , \dimh W_{n}(\psi)=\frac{n+1}{1+\tau_{\psi}}\,. Here \dimh denotes the Hausdorff dimension, see [ 10 ] for the definition and properties of Hausdorff dimension and measure. We consider the following variation of W_{n}(\psi) . Let \phi:{\mathbb{N}}\to[0,1] and define the set of points W_{n}(\psi,\phi):=\left\{\mathbf{x}\in[0,1]^{n}:(1-\phi(q))\frac{\psi(q)}{q}<\left\|\mathbf{x}-\frac{\mathbf{p}}{q}\right\|<\frac{\psi(q)}{q}\,\text{ for i.m. }\,(\mathbf{p},q)\in{\mathbb{Z}}^{n}\times{\mathbb{N}}\right\}\,. It is straightforward to see that W_{n}(\psi,\phi)\subset W_{n}(\psi) . In parallel to the set of \psi -approximable points, it can be seen that W_{n}(\psi,\phi) can be written as a \limsup set of annuli, or n -dimensional hyperspherical shells. Generally, we refer to these types of sets as quasi-annuli. Denoting by A(\mathbf{x};r_{1},r_{2}):=B(\mathbf{x},r_{1})\backslash B(\mathbf{x},r_{2}) the annulus with centre \mathbf{x}\in{\mathbb{R}}^{n} and inner and outer radii r_{2},r_{1} respectively, we may write W_{n}(\psi,\phi)=\limsup_{q\to\infty}\bigcup_{0\leq\|\mathbf{p}\|\leq q}A\left(\frac{\mathbf{p}}{q};\frac{\psi(q)}{q},(1-\phi(q))\frac{\psi(q)}{q}\right). In this note, we prove the following Jarník-Besicovitch type theorem in the setting of approximation by annuli. Theorem 1.1 . Take n\in{\mathbb{N}} and let \psi(q)=q^{-\tau_{\psi}} and \phi(q)=q^{-\tau_{\phi}} for \tau_{\psi},\tau_{\phi}\in{\mathbb{R}}_{+} with \tau_{\psi}\geq\frac{1}{n} . Then \dimh W_{n}(\psi,\phi)=\min\left\{\frac{n+1}{1+\tau_{\psi}},\frac{n+1+(n-1)\tau_{\phi}}{1+\tau_{\psi}+\tau_{\phi}}\right\}\,. The set W_{n}(\psi,\phi) can be related to sets of exact approximation order. These have recently gained significant interest, in chronogloical order see [ 3 , 6 , 7 , 8 , 11 , 12 , 19 , 15 , 1 ] . Define Exact_{n}(\psi):=\left\{\mathbf{x}\in[0,1]^{n}:\begin{cases}\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|<\frac{\psi(q)}{q}\quad\text{ for i.m. }\,\,q\in{\mathbb{N}}\\ \text{For each }\,0<c<1,\,\,\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|>c\frac{\psi(q)}{q}\quad\text{for all sufficiently large}\,\,q\in{\mathbb{N}}\,.\end{cases}\right\}. Bandi and De Saxce [ 1 ] recently obtained a Jarník-Besicovitch type statement in this setting. Namely, they showed that \dimh Exact_{n}(\psi)=\frac{n+1}{1+\tau_{\psi}} , where \psi is taken as in Theorem 1.1 and \tau_{\psi}>\tfrac{1}{n} . Note the case of n=1 had been proven prior in [ 8 ] . If \mathbf{x}\in Exact_{n}(\psi) it can be seen that there exists a sequence (c_{q})_{q\geq 1} with c_{q}\to 1 as q\to\infty such that \begin{cases}\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|<\frac{\psi(q)}{q}\quad\text{ for i.m. }\,\,q\in{\mathbb{N}}\\ \|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|>c_{q}\frac{\psi(q)}{q}\quad\text{ for all}\,\,q\in{\mathbb{N}}\,.\end{cases} A natural question to ask on the set Exact_{n}(\psi) is if one can be more specific on the rate at which the sequence (c_{q})_{q\geq 1} tends to one. To study this take a function f:{\mathbb{N}}\to[0,1] with f(q)\to 1 as q\to\infty , and define the set of points of (f,\psi) -Exact approximation order by Exact_{n}(f,\psi):=\left\{\mathbf{x}\in[0,1]^{n}:\begin{cases}\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|<\frac{\psi(q)}{q}\quad\text{ for i.m. }\,\,q\in{\mathbb{N}}\\ \|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|>f(q)\frac{\psi(q)}{q}\quad\text{ for all sufficiently large }\,\,q\in{\mathbb{N}}\,.\end{cases}\right\}\,. Clearly Exact_{n}(f,\psi)\subseteq Exact_{n}(\psi) , and so we have the trivial upper bound \dimh Exact_{n}(f,\psi)\leq\dimh Exact_{n}(\psi)\,. Using the set of study in this note it can also be seen that Exact_{n}(q\mapsto(1-\phi(q)),\psi)\subseteq W_{n}(\psi,\phi)\,. Note the following curious property on \dimh W_{n}(\psi,\phi) : for n\geq 2 it can be verified that \tau_{\psi}\leq\tfrac{2}{n-1}\quad\text{ if and only if }\quad\frac{n+1}{1+\tau_{\psi}}\geq\frac{n+1+(n-1)\tau_{\phi}}{1+\tau_{\psi}+\tau_{\phi}}\,. Thus, if \tau_{\psi}\leq\tfrac{2}{n-1} and \tau_{\phi}\to\infty , then \dimh W_{n}(\psi,\phi)=n-1\,. Whereas, if \tau_{\psi}\geq\tfrac{2}{n-1} , then regardless of our choice of \tau_{\phi} , we have that \dimh W_{n}(\psi,\phi)=\frac{n+1}{1+\tau_{\psi}}\,. This phenomenon is also seen in the case of weighted Diophantine approximation, see the comments proceeding [ 18 , Theorem 1] , and Remark 2 for more details. Combining our two observations above, Theorem 1.1 tells us that, for n\geq 2 , if \tau_{\psi}<\tfrac{2}{n-1} and f(q)=1-q^{-\varepsilon} for some \varepsilon>0 then \dimh Exact_{n}(f,\psi)\leq\frac{n+1+(n-1)\varepsilon}{1+\tau_{\psi}+\varepsilon}<\dimh Exact_{n}(\psi)\,. Conversely, if \tau_{\psi}>\tfrac{2}{n-1} , then it may be possible that \dimh Exact(f,\psi)=\dimh Exact(\psi) regardless of the rate at which f tends to 1 . The rest of the note is as follows. Firstly, we generalise the above theorem in several instances. Precisely we consider various forms of quasi-annuli, including annuli formed from the difference of balls described by different norms (Theorem 1.2 ), and rectangular annuli (Theorem 1.3 ). Many of these results stem from a variation of inhomogeneous Diophantine approximation, which we describe in § 2 . In § 3 we recall auxiliary results required in the proofs of our main theorems. Here, we use a variation of Cassel’s Scaling Lemma [ 9 ] and a generalisation of the Mass Transference Principle [ 4 ] to deduce a “shifted mass transference principle”, Theorem 3.2 and Theorem 3.4 . These theorems are crucial in proving our main results, which are done in § 3 - 4 . 1.1. Annuli defined by different norms We can consider the annuli described by taking the difference between balls described by different norms. Let \|\cdot\|_{\rho} denotes the \rho -norm on {\mathbb{R}}^{n} , \|\mathbf{x}\|_{\rho}=\left(\sum_{i=1}^{n}x_{i}^{\rho}\right)^{\tfrac{1}{\rho}} and define the set W_{n}(\psi,\|\cdot\|,\|\cdot\|_{\rho}):=\left\{\mathbf{x}\in[0,1]^{n}:\begin{cases}\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|<\tfrac{\psi(q)}{q},\\ \|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|_{\rho}>\tfrac{\psi(q)}{q},\end{cases}\quad\text{for i.m.}\,q\in{\mathbb{N}}\right\}\,. For \rho_{1},\rho_{2}\in{\mathbb{R}}_{+}\cup\{\infty\} , \mathbf{x}\in[0,1]^{n} and r>0 let A^{\rho_{1}}_{\rho_{2}}\left(\mathbf{x};r\right):=B_{\rho_{1}}(\mathbf{x},r)\backslash B_{\rho_{2}}(\mathbf{x},r)\,. Then we may write W_{n}(\psi,\|\cdot\|,\|\cdot\|_{\rho})=\limsup_{q\to\infty}\bigcup_{0\leq\|\mathbf{p}\|\leq q}A^{\infty}_{\rho}\left(\tfrac{\mathbf{p}}{q};\tfrac{\psi(q)}{q}\right)\,. We have the following result. Theorem 1.2 . Let \psi(q)=q^{-\tau_{\psi}} for \tau_{\psi}\geq\tfrac{1}{n} . Then, for any \rho -norm, \dimh W_{n}(\psi,\|\cdot\|,\|\cdot\|_{\rho})=\frac{n+1}{1+\tau_{\psi}}\,. Remark 1 . What will become clear in the proof of Theorem 1.2 is that we can take any norm that allows us to construct a ball of radius comparable, up to a multiplicative constant, with \frac{\psi(q)}{q} within the resulting quasi-annulus. The following is a natural open problem akin to the sets of exact approximation order as defined above. Question 1 . Define Exact^{*}_{n}(\psi,\|\cdot\|,\|\cdot\|_{\rho}):=\left\{\mathbf{x}\in[0,1]^{n}:\begin{cases}\|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|<\tfrac{\psi(q)}{q},\quad\text{ for i.m. }q\in{\mathbb{N}}\\ \|\mathbf{x}-\tfrac{\mathbf{p}}{q}\|_{\rho}>\tfrac{\psi(q)}{q},\quad\text{for all sufficiently large}\,q\in{\mathbb{N}}\end{cases}\right\}\,. What can be said on the size of Exact^{*}_{n}(\psi,\|\cdot\|,\|\cdot\|_{\rho}) ? One can use Khintchine’s Theorem and the equivalence of norms to deduce that Exact^{*}_{n}(\psi,\|\cdot\|,\|\cdot\|_{\rho}) is a Lebesgue nullset. Notice that it is not true that Exact_{n}(\psi)\subseteq Exact_{n}^{*}(\psi,\|\cdot\|,\|\cdot\|_{\rho}) , so the Hausdorff dimension result does not follow immediately from that of [ 1 ] . However, it may be possible using the techniques of [ 1 ] , or [ 15 ] , to obtain the Hausdorff dimension of Exact_{n}^{*}(\psi,\|\cdot\|,\|\cdot\|_{\rho}) provided one can impose some additional “directional dependence” on the constructed Cantor sets in the respective papers. It would seem reasonable to conjecture that \dimh Exact_{n}^{*}(\psi,\|\cdot\|,\|\cdot\|_{\rho})=\tfrac{n+1}{1+\tau_{\psi}}. 1.2. Approximation by rectangular annuli Weighted Diophantine approximation generalises the set W_{n}(\psi) by considering \limsup sets of rectangles, rather than \limsup set of balls. These sets have been well-studied, in particular, an analogue of the Jarník-Besicovitch theorem in the weighted case was proven by Rynne [ 18 ] . Similar to our setup in the introduction, we can also consider a weighted version