The superposition principle for local 1-dimensional currents
L. Ambrosio, F. Renzi, F. Vitillaro · 2025
Read The superposition principle for local 1-dimensional currents by L. Ambrosio, F. Renzi, F. Vitillaro (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-2503-18157.txt.
Introduction
Luigi Ambrosio, Federico Renzi, and Federico Vitillaro prove a result about one-dimensional currents in metric spaces. Their 2025 paper works in a branch of math that treats curves and flows in spaces where only distance is given. A current is a way to add up motion along many curves at once.
The team shows that each well-behaved local current of this kind can be written as a neat sum over curves. Those curves may run without bound but keep a finite length on each bounded part. The result widens an older theorem that held for a narrower class. It holds in Polish spaces and in complete spaces when the support is tight.
Catalog record
| CREATOR | L. Ambrosio, F. Renzi, F. Vitillaro |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 94 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-18157
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The superposition principle for local 1-dimensional currents L. Ambrosio, F. Renzi, F. Vitillaro Abstract . We prove that every one-dimensional locally normal metric current, intended in the sense of U. Lang and S. Wenger, admits a nice integral representation through currents associated to (possibly unbounded) curves with locally finite length, generalizing the result shown by E. Paolini and E. Stepanov in the special case of Ambrosio-Kirchheim normal currents. Our result holds in Polish spaces, or more generally in complete metric spaces for 1-currents with tight support. 1. Introduction Superposition principles are very useful tools in Analysis and Probability, as they allow to factorize a problem into lower dimensional ones. Classical examples are the coarea formula for BV functions, the slicing theory of currents and the Optimal Transport problem with cost=distance (in the latter case, the decomposition being given by transport rays). In the context of 1-dimensional currents, it is particularly interesting to extend the classical S. Smirnov’s [ Smi93 ] superposition principle even to the case when the ambient space is a metric space, having in mind the well-posedness of ODE’s in metric measure spaces (see [ AT14 ] , [ ST17 ] ) or the theory of Optimal Transport, see [ PS12 ] and [ PS13 ] . In this paper we prove that every one-dimensional locally normal metric current admits a decomposition, thus extending the results of E. Paolini and E. Stepanov in [ PS12 ] and [ PS13 ] for Ambrosio-Kirchheim normal currents. A notion of local metric current was first introduced by U. Lang in [ Lan11 ] . Later, he gave, alongside S. Wenger, a different, non equivalent definition, in [ LW11 ] . We will work with the latter (recalled in Section 2.1 below). Roughly speaking, this notion generalizes Ambrosio-Kirchheim metric currents by requiring finite mass only on bounded sets. In Section 3 of this paper, we provide the natural concepts of “subcurrent” and “acyclic current” for this class of objects. As it is customary in this field, we will work with the requirement that the mass measures of currents are tight, which is not restrictive, for instance, when the ambient metric space is Polish (see also Remark 2.1 ). Let us define, preliminarly, the classes of curves we will be dealing with. As in [ PS12 ] and [ PS13 ] we consider compact curves with finite length (which always admit a Lipschitz parameterization) and the distance induced by the sup norm when we consider the invariance under reparameterization: Definition 1.1 . Let (E,d) be a complete metric space. We equip the set of Lipschitz curves \theta:[0,1]\rightarrow E with the distance (1.1) d_{\Theta}(\theta_{1},\theta_{2}):=\inf\left\{\max_{t\in[0,1]}d(\theta_{1}(t),\theta_{2}(\phi(t)))\,:\,\phi\colon[0,1]\to[0,1]\mbox{ bijective increasing}\right\}, and call two curves \theta_{i}:[0,1]\to E , i=1,2 , equivalent, if d_{\Theta}(\theta_{1},\theta_{2})=0 . The set of equivalence classes of Lipschitz curves equipped with the distance d_{\Theta} will be denoted by \Theta(E) . Remark 1.2 . (\Theta(E),d_{\Theta}) can be written as a countable union of complete subspaces of the form \Theta_{L}=\{[\theta]\in\Theta(E):\ell(\theta)\leq L\} , L\in\mathbb{N} . For example, if E is separable (and then \Theta(E) is separable as well), this gives that any finite Borel measure \eta on \Theta(E) is tight. A notable distinction from the classical case is that, in order to maintain the transport property of the representation of acyclic currents (which comes from ( 1.4 ) below), highlighted in [ PS12 ] , it is necessary to consider curves which are not necessarily of bounded length. This can be easily understood by looking at the basic example of the current associated to an unbounded simple oriented curve with a single boundary point (e.g. a half-line in \mathbb{R}^{d} ), where, as it is somehow natural, the decomposition must be given by the curve itself. Clearly, the transport property has to be meant in a wider sense, involving the possibility of exchanging mass “with infinity” (see Remark 4.19 ). Therefore in the context of local currents, it is more convenient to consider, instead, open ended curves with locally finite length in E , according to the following definition: Definition 1.3 . Consider X:=\bigl\{\beta\in\operatorname{Lip}_{\rm loc}((0,1);E)\mid\ell(\beta\mathbin{\vrule height=6.88889pt,depth=0.0pt,width=0.55974pt\vrule height=0.55974pt,depth=0.0pt,width=5.59721pt}U)<\infty\,\,\text{for any bounded open set $U\subseteq E$}\bigr\}, where \ell(\beta\mathbin{\vrule height=6.88889pt,depth=0.0pt,width=0.55974pt\vrule height=0.55974pt,depth=0.0pt,width=5.59721pt}U)=\int_{\beta^{-1}(U)}|\dot{\beta}|\,\mathrm{d}t and |\dot{\beta}| is the metric derivative of \beta . We define the set \Gamma(E) as the quotient {\mathchoice{\raisebox{3.41666pt}{$\displaystyle{X}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-1.83437pt}{$\displaystyle{\sim}$}}{\raisebox{3.41666pt}{$\textstyle{X}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-1.83437pt}{$\textstyle{\sim}$}}{\raisebox{2.39166pt}{$\scriptstyle{X}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-1.34427pt}{$\scriptstyle{\sim}$}}{\raisebox{1.70833pt}{$\scriptscriptstyle{X}$}\mkern-5.0mu\diagup\mkern-4.0mu\raisebox{-0.99101pt}{$\scriptscriptstyle{\sim}$}}} , where \beta_{1}\sim\beta_{2} iff there exists \phi:(0,1)\rightarrow(0,1) increasing and bijective s.t. \beta_{1}=\beta_{2}\circ\phi . Notice that the difference between \Theta(E) and \Gamma(E) is essentially due to the requirement of locally finite length, since any open ended curve with finite length has a unique extension to a curve in \Theta(E) . Actually, the local finiteness of the length of \beta\in\Gamma(E) gives the following dichotomy: either \lim_{t\to 0}\beta(t) exists, or \lim_{t\to 0}d(\beta(t),\bar{x})=\infty , for any \bar{x}\in E (same for t\to 1 ). In addition, since all concepts we will be dealing with are invariant under reparameterization, we will occasionally consider other open of closed intervals as domains of our curves. In Section 4.5 we introduce, mainly for measure-theoretic purposes, the topology \tau_{\Gamma} induced by the map \beta\mapsto T_{\beta} in ( 2.22 ), canonically associating to \beta\in\Gamma(E) the local 1-dimensional current in E . In this connection, notice that this topology is not Hausdorff in \Gamma(E) since T_{\beta}=T_{\tilde{\beta}} does not imply that \beta is a reparameterization of \tilde{\beta} (for example, concatenating \beta with two copies with opposite orientation of a curve \alpha with \alpha(0)=\beta(1) does not change T_{\beta} ). Our main result is the following (here \mathbf{N}_{1,b}(E) denotes the class of local and normal 1 -dimensional currents in E , see Definition 2.3 and Definition 2.12 ). Theorem 1.4 . Let (E,d) be a complete metric space and let T\in\mathbf{N}_{1,b}(E) . Then there exists a positive Borel measure \eta over \Gamma(E) such that (1.2) \displaystyle T=\int_{\Gamma(E)}T_{\gamma}\,\mathrm{d}\eta(\gamma), (1.3) \displaystyle\|T\|=\int_{\Gamma(E)}\|T_{\gamma}\|\,\mathrm{d}\eta(\gamma). Moreover, T can be decomposed as T=A+C , where C is a cycle of T and A is acyclic. Finally, denoting by \eta_{A} the measure associated to A as in ( 1.2 ) and ( 1.3 ), we have that \eta_{A} -almost every curve is injective, and that (1.4) \|\partial T\|=\|\partial A\|=\int_{\Gamma(E)}\|\partial T_{\gamma}\|\,\mathrm{d}\eta_{A}(\gamma). In this generality, Theorem 1.4 represents a new result even in the Euclidean setting. However, our proof-technique is intrinsically metric, in the sense that it relies on the validity of the statement for normal currents in general metric spaces, due to Paolini and Stepanov, even when our space E is Euclidean. Indeed, we exploit a simple trick which consists in using a \|T\| -dependent “conformally” modified distance \delta (see Section 4.3 ) of the distance d , turning T into a current \overline{T} of finite mass with respect to \delta in the completion \overline{E} of E with respect to \delta . Therefore, when the mass of \partial T is assumed to be already finite, we reduce ourselves to an application of Paolini and Stepanov’s result, provided we are able to show that still \partial\overline{T} has finite mass (this is not obvious, see Example 4.13 , but it can be obtained at the expense of embedding isometrically our spaces into \ell^{\infty} ). We recall here Paolini and Stepanov’s result for further use (cfr. [ PS12 , Theorem 5.1] and [ PS13 , Corollary 3.3] ). Theorem 1.5 . Let T\in\mathbf{N}_{1}(E) . Then there exists a finite positive Borel measure