2025-11-18T15:46:13.504176

Recurrence and transience of the critical random walk snake in random conductances

Legrand, Sabot, Schapira
In this paper we study the recurrence and transience of the $\mathbb{Z}^d$-valued branching random walk in random environment indexed by a critical Bienaymé-Galton-Watson tree, conditioned to survive. The environment is made either of random conductances or of random traps on each vertex. We show that when the offspring distribution is non degenerate with a finite third moment and the environment satisfies some suitable technical assumptions, then the process is recurrent up to dimension four, and transient otherwise. The proof is based on a truncated second moment method, which only requires to have good estimates on the quenched Green's function.
academic

Recurrence and transience of the critical random walk snake in random conductances

Basic Information

  • Paper ID: 2406.17622
  • Title: Recurrence and transience of the critical random walk snake in random conductances
  • Authors: Alexandre Legrand, Christophe Sabot, Bruno Schapira
  • Classification: math.PR (Probability Theory)
  • Publication Date: June 2024 (arXiv preprint)
  • Paper Link: https://arxiv.org/abs/2406.17622

Abstract

This paper investigates the recurrence and transience properties of Zd\mathbb{Z}^d-valued branching random walks in random environments, indexed by critical Bienaymé-Galton-Watson trees conditioned on survival. The environment consists of random conductances or random traps at each vertex. The study demonstrates that when the offspring distribution is non-degenerate with finite third moment and the environment satisfies appropriate technical assumptions, the process is recurrent in dimensions four and below, and transient otherwise. The proof is based on a truncated second moment method requiring only good estimates of the quenched Green function.

Research Background and Motivation

Problem Background

  1. Research Object: The critical random walk snake, a Zd\mathbb{Z}^d-valued branching random walk process indexed by critical Bienaymé-Galton-Watson trees
  2. Core Problem: Determining recurrence versus transience of this process in random environments
  3. Environment Types: Two types of random environments are considered: random conductances and random traps

Research Significance

  1. Theoretical Importance: Extends understanding of critical branching processes in random environments
  2. Technical Challenges: Critical cases are more difficult to analyze than supercritical cases, receiving limited attention in existing literature
  3. Methodological Contribution: Develops a truncated second moment method relying only on quenched Green function estimates

Limitations of Existing Methods

  1. General criteria for supercritical offspring distributions exist, but critical cases are understudied
  2. Previous approaches, such as those by Benjamini and Curien, based on unimodularity and mass transport techniques, are technically complex
  3. High requirements for Green function estimates in random environments

Core Contributions

  1. Establishes a Dimension Threshold Theorem: Proves that the critical random walk snake is recurrent for d≤4d \leq 4 and transient for d≥5d \geq 5
  2. Develops a New Proof Method: Proposes a proof framework based on the truncated second moment method requiring only quenched Green function estimates
  3. Provides a 0-1 Law: Under appropriate assumptions, establishes a 0-1 law for recurrence
  4. Handles Two Environment Types: Separately analyzes random conductance and random trap environments with different technical assumptions

Detailed Methodology

Task Definition

Input:

  • Critical offspring distribution q=(qk)k≥0q = (q_k)_{k \geq 0} satisfying ∑k≥0kqk=1\sum_{k \geq 0} kq_k = 1 and ∑k≥0k3qk<∞\sum_{k \geq 0} k^3q_k < \infty
  • Random environment ω=(ωx,y)x,y∈Zd\omega = (\omega_{x,y})_{x,y \in \mathbb{Z}^d}

Output: Determine recurrence or transience of the critical random walk snake ST∞:T∞→ZdS_{T_\infty}: T_\infty \to \mathbb{Z}^d

Constraints:

  • Kesten tree T∞T_\infty (critical BGW tree conditioned on survival)
  • Environment satisfies stationarity and ergodicity assumptions

Core Technical Framework

1. Environment Definition

The paper considers two classes of environments:

Random Conductances: For x∼yx \sim y, ωx,y=ωy,x>0\omega_{x,y} = \omega_{y,x} > 0; otherwise ωx,y=0\omega_{x,y} = 0

Random Traps: There exist ρx∈[0,1)\rho_x \in [0,1) such that

\rho_x/(1-\rho_x) & \text{if } x = y \\ 1/(2d) & \text{if } x \sim y \\ 0 & \text{otherwise} \end{cases}$$ #### 2. Dynamical Systems Approach Through a change of measure from particle perspective $Q \sim P$: $$\frac{dQ}{dP}(\omega) = \frac{\pi_\omega(0)}{E[\pi_\omega(0)]}$$ Constructs an ergodic dynamical system $(A, \mathcal{A}, \tilde{P}, R)$ to prove the 0-1 law. #### 3. Green Function Estimates Utilizes estimates from reference [3]: **Upper Bound**: For $d \geq 5$, when $1/p + 1/q < 2/d$, $$P_n^\omega(0,0) \leq c_2 n^{-d/2}$$ **Lower Bound**: For $d \geq 3$, $$g^\omega(x,y) \geq c_5|x-y|^{2-d}$$ ### Technical Innovations 1. **Truncated Second Moment Method**: Does not require complete heat kernel estimates, relying only on Green function estimates 2. **Box Constraint Technique**: Controls variance by constraining the process within large boxes $\Lambda_m$ 3. **Improved Treatment of Trap Environments**: Uses double truncation (box boundaries and deep traps) for random trap environments ## Main Results ### Theorem 1.5 (Random Conductance Environment) **(i) Transience** ($d \geq 5$): When $1/p + 1/q < 2/d$ and $\omega \in L^p(P), \omega^{-1} \in L^q(P)$, the critical snake is almost surely transient. **(ii) Recurrence** ($d \leq 4$): There exists a constant $p \in [1,\infty)$ such that when the environment satisfies Assumptions 1.2, 1.4 and $\omega, \omega^{-1} \in L^p(P)$, the critical snake is almost surely recurrent. ### Theorem 1.7 (Random Trap Environment) **(i) Transience** ($d \geq 5$): When $\sup_{x \in \mathbb{Z}^d} E\pi_\omega(x) < \infty$, the critical snake is almost surely transient. **(ii) Recurrence** ($d \leq 4$): Under Assumptions 1.2, 1.6 and $E\pi_\omega(0) < \infty$, the critical snake is almost surely recurrent. ## Proof Strategy ### Transience Proof ($d \geq 5$) Through normalized local time: $$L_\infty(0) := \frac{\#S_{T_\infty}^{-1}(\{0\})}{\pi_\omega(0)}$$ Proves $E^\omega[L_\infty(0)] < \infty$, utilizing: $$E^\omega[L_\infty(0)] = \sigma^2 \sum_{n \geq 0} (n+1)P_n^\omega(0,0)$$ ### Recurrence Proof ($d \leq 4$) Uses the second moment method on the process $L_m$ constrained to box $\Lambda_m$: **Lemma 4.3**: $E[L_m] \asymp \begin{cases} m & \text{if } d=3 \\ \log m & \text{if } d=4 \end{cases}$ **Lemma 4.4**: $E[L_m^2] \leq c_2 E[L_m]^2$ By the Paley-Zygmund inequality, obtains $P(L_\infty(0) = \infty) > 0$, and combined with the 0-1 law yields recurrence. ## Technical Difficulties and Innovations ### 1. Refined Green Function Estimates - Utilizes results from Andres-Halberstam to obtain upper and lower bound estimates in random conductance environments - For random trap environments, proves $g^\omega(x,y) = g(x,y)$ (homogeneous case) ### 2. Variance Control Controls second moments through the many-to-two formula: $$E^\omega[(L_{x,0}^{1,m})^2] \leq c \sum_{y \in \Lambda_m} g^m(x,y)g^m(y,0)^2\pi_\omega(y)\pi_\omega(0)^2$$ ### 3. Double Truncation for Trap Environments Defines the deep trap set $A_\omega = \{x: \pi_\omega(x) \geq R|x|^2\}$, simultaneously truncating the process at boundaries and deep traps. ## Related Work 1. **Benjamini-Curien (2012)**: Used unimodularity and mass transport techniques to prove results in homogeneous environments 2. **Comets-Popov (2007), Gantert-Müller (2006)**: General criteria for supercritical cases 3. **Andres-Halberstam (2021)**: Green function estimates in random conductance environments 4. **Classical Results**: Kesten's concentration results for snakes in balls ## Conclusions and Discussion ### Main Conclusions Establishes a complete dimension threshold for recurrence of critical random walk snakes: recurrent for $d \leq 4$, transient for $d \geq 5$. ### Limitations 1. Integrability assumptions in the random conductance case are strong, with optimal $p$ values not provided 2. Finite range dependence assumptions could be relaxed but require more complex techniques 3. Complete treatment of mixed environments (conductances + traps) requires additional work ### Future Directions 1. **Heat Kernel Decay**: Study the validity of $\sum_n nP_0^\omega(X_{2n} = 0) < \infty$ across different dimensions 2. **Hitting Probabilities**: Extend to hitting probability estimates for finite BRWRE 3. **More General Environments**: Handle combined environments of conductances and traps ## In-Depth Evaluation ### Strengths 1. **Theoretical Completeness**: Establishes a complete dimension threshold theorem for the critical case 2. **Methodological Innovation**: The truncated second moment method avoids complex unimodularity techniques 3. **Technical Depth**: The combination of Green function estimates and dynamical systems methods demonstrates sophisticated technical mastery 4. **Precise Results**: Provides exact conditions for both environment types ### Weaknesses 1. **Strong Assumptions**: Particularly the integrability requirements in the random conductance case 2. **Non-optimal Constants**: Does not pursue optimal integrability indices 3. **Limited Applications**: Primarily theoretical results with limited practical applicability ### Impact 1. **Theoretical Contribution**: Fills an important gap in critical branching random walk theory 2. **Methodological Value**: The truncated second moment method may apply to other related problems 3. **Technical Impact**: Provides new tools for studying branching processes in random environments ### Applicable Scenarios Primarily applicable to probability theory research on branching processes, random walks, and random environment theory, with theoretical guidance for phase transition phenomena in statistical physics. ## References The paper cites 24 important references covering classical and recent results in branching processes, random walks, and random environments, reflecting the authors' deep understanding of related fields.