Limits of equi-affine equi-distant loci of planar convex domains with two non-parallel asymptotes
Nikita Kalinin, Mikhail Shkolnikov · 2025
Read Limits of equi-affine equi-distant loci of planar convex domains with two non-parallel asymptotes by Nikita Kalinin, Mikhail Shkolnikov (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-2512-21154.txt.
Introduction
Nikita Kalinin and Mikhail Shkolnikov build geometric invariants by averaging over tropical structures. Their approach fixes the covolume and averages over all tropical structures of that size. It applies the result to the tropical distance series. That produces functions attached to convex domains. The functions do not change under equi-affine transformations and are expected to enjoy several striking traits. The level sets of these functions are the equi-distant loci of the title.
For bounded domains the limiting shape of the loci remains a conjecture. For unbounded domains opening toward two non-parallel asymptotes, the authors prove the analogue: the loci approach a hyperbola, and its affine curvature is universal, free of the domain's details. An explicit formula for the mean value at the center of the unit disk rounds out the note, anchoring the theory in a computable case.
Catalog record
| CREATOR | Nikita Kalinin, Mikhail Shkolnikov |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-21154
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Limits of equi-affine equi-distant loci of planar convex domains with two non-parallel asymptotes Nikita Kalinin, Mikhail Shkolnikov Abstract In this note we discuss the novel approach to define equi-affine invariants starting from tropical geometry, where one averages over the space of tropical structures of fixed co-volume. Applied to the tropical distance series, this gives a family of equi-affine invariant functions associated with convex domains which are expected to satisfy a number of remarkable properties. The present note contains a conjecture about the limit structure of corresponding level sets in the compact case, as well as the proof of an analogue of this conjecture for unbounded domains with two non-parallel asymptotes. In addition, we carry out an explicit computation for the arithmetic mean value at center of the unit disk. † † M.S. is supported by the Simons Foundation, grant SFI-MPS-T-Institutes-00007697, and the Ministry of Education and Science of the Republic of Bulgaria, grant DO1-239/10.12.2024. 1 Introduction An equi-affine plane is modeled by \mathbb{R}^{2} with the symmetry group generated by all translations and \operatorname{SL}_{2}(\mathbb{R}) acting in the usual way. This means that among classical geometric quantities only the area makes sense. In particular, there is no invariant notion of distance between a pair of points on such a plane. In this note, we introduce a family of functions assigning a positive number to a point inside a convex domain, which may be interpreted as a distance from the point to the boundary of the domain. The definition of such an “equi-affine distance function” stems from a suggestion of Conan Leung, who proposed to average the tropical distance function over the space of all tropical structures of fixed co-area of the ambient plane. The expectation is that this way we may recover the classical Monge-Ampére equation and the affine normal flow. See a popular exposition on how the optimal transport problem leads to the Monge-Ampére equation [ 2 ] and a polyhedral approximation for its solution by Pogorelov [ 14 , 13 , 7 , 3 , 8 ] . Remarkably, in contrast with the tropical distance function, whose non-zero levels are polygonal, equi-affine distance function seems to have smooth level sets, which we may call “equi-distant loci” of a domain, even if we start with a polygon (Figure 2 shows a plot on a square). Our purpose is to initiate the study of these level sets. A particular conjecture that we put forward is that for a fixed compact convex domain, these loci, after an appropriate rescaling, converge to an ellipse. In fact, as a few numerical simulations performed by Ernesto Lupercio suggest, these level sets have monotonically increasing Mahler area (which is a scale free affine invariant roughly measuring the “roundness” of a domain) approaching that of the global maximum in the space of convex domains, i.e. that of ellipses. Thus, one can expect that developing the theory in higher dimensions may give a tool to tackle Mahler conjecture, which is a fundamental unsolved problem in affine and convex geometries. At present, we do not know how to treat the maximal limit of equi-affine equi-distant loci in the compact case. However, for the analogous question, but in the non-compact case of a domain with two non-parallel asymptotes, it is relatively straightforward to show that the equi-affine equi-distant loci after a rescaling converge to branches of hyperbolas, which is the main proven result of this note, a first step toward a more general theory. The prominence of ellipses and hyperbolae also relates to affine isoperimetric principles. In affine geometry [ 1 , 11 ] , ellipses often extremize or simplify affine-invariant functionals. For example, the affine isoperimetric inequality states that among all closed convex curves of a given affine length, the ellipse encloses the maximal area. Similarly, the affine curvature flows tend to “round out” convex shapes into ellipses. In fact, affine curve shortening flow and affine normal flow provide a dynamic illustration: Sapiro and Tannenbaum [ 9 ] showed that evolving a convex plane curve under motion by its affine normal causes the curve to become asymptotically ellipsoidal. 2 Definition, conjecture, theorem An (integrable) tropical structure on the plane is a two-dimensional lattice \Lambda in the tangent space of \mathbb{R}^{2} at some point, which is extended to all other points via translations. The co-area of tropical structure \Lambda is the area of the quotient torus \mathbb{R}^{2}/\penalty\Lambda . The standard tropical structure \mathbb{Z}^{2}\subset\mathbb{R}^{2} has co-area one. The group \operatorname{SL}_{2}(\mathbb{R}) acts transitively on the space of tropical structures of co-area one. The stabilizer of the standard tropical structure is \operatorname{SL}_{2}(\mathbb{Z}), thus the space of all co-area one tropical structures on \mathbb{R}^{2} is identified with the quotient \operatorname{SL}_{2}(\mathbb{Z})\backslash\operatorname{SL}_{2}(\mathbb{R}) which has a finite volume with respect to a natural measure \mu (this volume is actually \zeta(2)=\frac{\pi^{2}}{6} ). Let \Omega be a convex domain on \mathbb{R}^{2} with tropical structure \Lambda. We call it \Lambda -admissible if every line it contains (there might be none) goes in the direction of some vector in \Lambda. Note that admissibility of \Omega is equivalent to asking that among numbers c_{\lambda}^{\Omega}=-\inf_{p\in\Omega}\lambda\cdot p for \lambda\in\Lambda there exist a non-infinite one. For a \Lambda -admissible \Omega, define the tropical distance function as a tropical series \mathcal{F}_{\Lambda}^{\Omega}(p)=\inf_{\lambda\in\Lambda\backslash\{0\}}(c_{\lambda}^{\Omega}+\lambda\cdot p),\hskip 15.0pt\mathcal{F}_{\Lambda}^{\Omega}:\Omega\rightarrow[0,+\infty). A lot can be said about such tropical series, see for instance [ 5 , 6 , 4 ] . One important geometric property following from elementary number theory is that (\mathcal{F}_{\Lambda}^{\Omega})^{-1}(0)=\partial\Omega. It is an important conceptual challenge to figure out if the tropical distance can be defined for a non-convex \Omega, see [ 10 ] . A convex domain \Omega\subset\mathbb{R}^{2} is called strongly admissible if it is admissible for every tropical structure on the plane, i.e. if it doesn’t contain any lines. Definition . For a real number h>0 we define the h -equi-affine distance function \mathcal{A}_{h}^{\Omega}:\Omega\rightarrow[0,+\infty] an h -Hölder average of the tropical distance series over the space of tropical structures of co-area one, i.e. as \mathcal{A}_{h}^{\Omega}(p)=\frac{6}{\pi^{2}}\Big(\int_{[\Lambda]\in\operatorname{SL}_{2}(\mathbb{Z})\backslash\operatorname{SL}_{2}(\mathbb{R})}(\mathcal{F}^{\Omega}_{\Lambda}(p))^{h}d\mu[\Lambda]\Big)^{h^{-1}} We will prove a bit later that \mathcal{A}_{h}^{\Omega} is finite at every point. If we assume this, then it is clear from the definition that this notion is equi-affine invariant and homogeneous of degree one, i.e. \mathcal{A}_{h}^{A(\Omega)}(A(p))=\mathcal{A}_{h}^{\Omega}(p) and \mathcal{A}_{h}^{r\Omega}(rp)=r\mathcal{A}_{h}^{\Omega}(p) for any A\in\operatorname{SL}_{2}(\mathbb{R}),\,p\in\Omega, and r>0. Denote by \mathcal{C} the space of simple continuous curves (embedded circles or lines, unbounded in the latter case) on the plane considered up to homothety and translations, with its elements denoted by [C] . We put forward the following conjecture: Conjecture . Let \Omega be a compact convex domain with non-empty interior. For h>0 denote by \mathbb{m}_{h}^{\Omega} the maximal value of \mathcal{A}^{\Omega}_{h}. Then there exists an ellipse E^{h}_{\Omega} such that when t approaches \mathbb{m}_{h}^{\Omega} from the left, the classes of level sets [(\mathcal{A}^{\Omega}_{h})^{-1}(t)] converge to [E^{\Omega}_{h}]\in\mathcal{C}. An interesting supplementary problem is to understand how fast the level sets degenerate, i.e. what is an asymptotic for the area of (\mathcal{A}^{\Omega}_{\Lambda})^{-1}[t,\mathbb{m}_{h}^{\Omega}] as t approaches the maximal value \mathbb{m}_{h}^{\Omega} (Figure 2 suggests that it is quadratic). It is also likely that the class of an ellipse [E^{\Omega}_{h}] wouldn’t depend on h. In the next section, we prove the following analogue of the conjecture above. Theorem . Let \Omega be a convex planar domain with two non-parallel asymptotes. Denote by H^{\Omega} a branch of a