Critical Self-Similar Markov Trees
Nicolas Curien, Xingjian Hu, Dongjian Qian · 2026
Read Critical Self-Similar Markov Trees by Nicolas Curien, Xingjian Hu, Dongjian Qian (2026) 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-2603-16408.txt.
Introduction
Curien, Hu, and Qian study critical self-similar Markov trees. These random decorated trees generalize the fragmentation tree. Earlier work set the critical case aside. The authors borrow tools from branching random walk. A recent result by Aïdékon, Hu, and Shi anchors the method. They construct the critical trees and compute their fractal dimension. Spinal decomposition then yields their harmonic and length measures.
Catalog record
| CREATOR | Nicolas Curien, Xingjian Hu, Dongjian Qian |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 127327 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.16408
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Critical Self-Similar Markov Trees Nicolas Curien, Xingjian Hu, Dongjian Qian Abstract Recently introduced and studied in [ 7 ] , a self-similar Markov tree (ssMt) is a random decorated tree that vastly generalises the fragmentation tree. We study here the critical case that was left aside in [ 7 ] . Borrowing techniques from branching random walk, in particular the recent result of Aïdékon–Hu–Shi [ 3 ] , we can complete the picture by constructing critical ssMt, computing their fractal dimension and studying their associated harmonic and length measures using spinal decomposition. 1 Introduction Self-similar Markov trees have recently been introduced in [ 7 ] . They can be seen as generalisations of the fragmentation trees of Haas & Miermont [ 22 ] , or as the genealogical trees underlying the Lamperti transformation of branching Lévy processes [ 8 ] . They encompass and unify various models including Aldous’ famous Brownian Continuum Random tree [ 4 ] and its stable generalisations [ 21 ] as well as more exotic trees such as the Brownian Cactus [ 19 ] sitting inside the Brownian sphere [ 25 , 6 ] , or the scaling limits of peeling or parking trees [ 5 , 17 ] . They are conjectured to describe scaling limits of multi-type Bienaymé–Galton–Watson trees, exactly as Lamperti’s self-similar Markov processes describe the scaling limits of positive Markov chains on the integers. Their central role in the theory of random tree sparked a recent interest, see e.g. [ 9 , 18 ] for their intrinsic studies or [ 1 , 25 , 6 , 20 ] for their connections with various models of random planar geometry. Formally, a self-similar Markov tree (ssMt) is a family of laws (\mathbb{Q}_{x}:x>0) of decorated continuous random \mathbb{R} -trees \mathtt{T}=(T,d_{T},\rho,g) where \rho is the root of the tree and g:T\to\mathbb{R}_{+} is a real decoration which is upper semi-continuous (usc) on T and positive on its skeleton. Under the law \mathbb{Q}_{x} , the random decorated tree \mathtt{T} starts from the initial decoration g(\rho)=x and enjoys the following two eponymic properties: 1. Markov property. For each h>0 , conditioned on the subtree truncated at height h , the decorated subtrees above h are independent of each other and have law \mathbb{Q}_{y} if the decoration at its root is y . 2. Self-similarity. There exists \alpha>0 , the self-similar index , such that for each x>0 , the tree (T,d_{T},\rho,g) under \mathbb{Q}_{x} has the same law as (T,x^{\alpha}d_{T},\rho,x\cdot g) under \mathbb{Q}_{1} . Following [ 7 , Section 2] , a self-similar Markov tree can be described by its characteristic quadruplet (\sigma^{2},\mathrm{a},\boldsymbol{\Lambda};\alpha) , which through a Lamperti transformation encapsulates the law of their underlying branching Lévy processes. In particular, the generalised Lévy measure \boldsymbol{\Lambda} is a (possibly infinite) measure on the space \mathcal{S}=\{\mathbf{u}=(u_{0},u_{1},...):u_{0}\in\mathbb{R}\mbox{ and }u_{1}\geq u_{2}\geq...\in\mathbb{R}\cup\{-\infty\}\} which describes the splitting rules: Informally, the ssMt is the genealogical tree of a system of individuals evolving independently of each other, and where an individual of decoration x>0 sees its decoration instantaneously moved to x\cdot\mathrm{e}^{y_{0}} while giving rise to a family of new individuals with decorations x\cdot\mathrm{e}^{y_{1}},x\cdot\mathrm{e}^{y_{2}},... (which are interpreted as the birth of new individuals) \displaystyle x\to(x\cdot\mathrm{e}^{y_{0}},(x\cdot\mathrm{e}^{y_{1}},x\cdot\mathrm{e}^{y_{2}},\cdots))\quad\mbox{ at a rate }\quad x^{-\alpha}\cdot\boldsymbol{\Lambda}(\mathrm{d}\mathbf{y}), (1.1) where \alpha>0 is the self-similarity parameter. The projection \Lambda_{0} of \boldsymbol{\Lambda} on its first coordinate is required to be a Lévy measure, and its projection \boldsymbol{\Lambda}_{1} onto the second coordinate satisfies a mild integrability assumption (see ( 2.1 )). The coefficient \mathrm{a}\in\mathbb{R} encodes the drift term while \sigma^{2} controls the Brownian part of the evolution of the decoration along branches. See Section 2.1 or [ 7 ] for details. When the starting decoration x>0 is fixed, the law \mathbb{Q}_{x} is the distribution of the above genealogical tree started with an individual of decoration x . A crucial quantity to consider is the cumulant function defined by \begin{split}\kappa(\gamma)&=\frac{1}{2}\sigma^{2}\gamma^{2}+\mathrm{a}\gamma+\int_{\mathcal{S}}\boldsymbol{\Lambda}(\mathrm{d}\mathbf{u})\left(\mathrm{e}^{\gamma u_{0}}-1-\gamma u_{0}\mathbf{1}_{\{|u_{0}|\leq 1\}}+\sum_{i=1}^{\infty}\mathrm{e}^{\gamma u_{i}}\right),\\ &=\psi(\gamma)+\int_{\mathcal{S}}\boldsymbol{\Lambda}(\mathrm{d}\mathbf{u})\bigg(\sum_{i=1}^{\infty}\mathrm{e}^{\gamma u_{i}}\bigg)\end{split} (1.2) where \psi(\gamma) is the Laplace exponent of the Lévy process with characteristics (\sigma^{2},\mathrm{a},\Lambda_{0}) . This function \kappa also appears as the Biggins transform of the underlying branching Lévy process. In particular, if \kappa takes strictly negative values then the types of individuals is decaying over generations in expectation, and the genealogical tree of the above system of particles is compact and provides a ssMt. While if \kappa only takes positive values then the tree explodes locally, see [ 10 ] . The case when \kappa touches 0 while remaining non-negative was left aside in [ 7 ] and is the context of this paper: Assumption 1.1 (Criticality I) . Suppose that for \gamma\geq 0 we have \kappa(\gamma)\geq 0 and there exists \omega_{-}\geq 0 such that \kappa(\omega_{-})=0 and \kappa is twice differentiable at \omega_{-} . Furthermore, There exists \gamma_{1}>\omega_{-} such that \psi(\gamma_{1})<0 and \kappa(\gamma_{1})<\infty . We shall have a stronger technical 3.5 analogue to the Cramér assumption in [ 7 ] needed to study fine properties of ssMt. This assumption is omitted in the introduction for readability. Figure 1: Illustration of the criticality on the cumulant function. Theorem 1.2 (Construction of the critical ssMt) . Under 1.1 the construction in [ 7 ] can indeed be performed and it yields a family of laws (\mathbb{Q}_{x}) of compact decorated random trees satisfying the Markov and self-similarity property. Furthermore, under 3.5 , for any x>0 the Hausdorff dimension of the leaves \partial T of a tree T under \mathbb{Q}_{x} is a.s. \mathrm{dim_{H}}(\partial T)=\frac{\omega_{-}}{\alpha}. Compared to [ 7 ] , the construction of self-similar Markov trees in the critical case relies on the very recent result of Aïdékon–Hu–Shi [ 3 ] which was in fact motivated by the above application. A few critical ssMt were already considered in [ 7 ] but their existence was proved case-by-case using specific features of the models. This was notably the case for the ssMt of Aïdékon and Da Silva [ 1 ] which arose from half-planar Brownian excursion. With the critical case at hand, the family of ssMt, which is now complete, is conjectured to describe all possible scaling limits of multi-type Bienaymé–Galton–Watson trees. We now move on to discuss the properties of critical self-similar Markov trees and their natural measures following the same strategy as for the subcritical case [ 7 ] . Measures and spine decomposition. Any decorated tree \mathtt{T}=(T,g) carries a natural Lebesgue measure \lambda_{T} on its skeleton, and the length measures are obtained by using the decoration g as density. Formally, for any \gamma\geq 0 , we define the \gamma -length measure on T as \mathrm{d}\lambda^{\gamma}:=g^{\gamma-\alpha}\cdot\mathrm{d}\lambda_{T}. In the subcritical case, those measures are defined as soon as \gamma>\omega_{-} and Proposition 2.12 of [ 7 ] even ensures that \lambda^{\gamma} has finite expected mass when \kappa(\gamma)<0 . In the critical case, we prove in Section 3 that \lambda^{\gamma} is still a finite measure when \gamma>\omega_{-} a.s. but with infinite expected length . Let us now turn to the equivalent of the harmonic measure of [ 7 , Chapter 2.3.3] . Recall from Assumption 1.1 that \kappa(\omega_{-})=0 . In this case, the process W_{n}=\sum_{|u|=n}(\chi(u))^{\omega_{-}}, where \chi(u) are the initial decorations of the individuals appearing in the genealogical tree (indexed by the Ulam’s tree, see the construction of ssMt in Section 2.1 ) is a martingale. In the context of