% ---- preamble ---- \documentclass[12pt,reqno]{amsart} \usepackage[utf8]{inputenc} \usepackage{amsmath,amssymb,amsthm,mathtools,mathrsfs,esint} \usepackage{mathalpha} \usepackage{enumitem} \usepackage{libertinus} \usepackage{microtype} \usepackage{booktabs,longtable,array} \usepackage[a4paper,top=25mm,bottom=26mm,left=29mm,right=29mm,headsep=7mm]{geometry} \usepackage[dvipsnames]{xcolor} \usepackage[ breaklinks=true, pagebackref=true, hypertexnames=false ]{hyperref} \hypersetup{ colorlinks=true, pdfpagemode=UseNone, citecolor=ForestGreen, linkcolor=MidnightBlue, urlcolor=RoyalBlue, pdfstartview=FitH, pdftitle={The lower-class escape rate of planar favorite sites}, pdfsubject={Planar simple random walk, favorite sites, and lower-class escape rate} } \usepackage[nameinlink,noabbrev]{cleveref} % Page geometry and paragraph spacing from the template. \linespread{1.05} \normalsize \linespacing=\baselineskip \setlength{\parindent}{1em} \setlength{\parskip}{3pt} \makeatletter % First-page title, author block, abstract, and separate AI usage statement. \def\@settitle{% \begin{center}% \uppercasenonmath\@title \Large\@title \end{center}} \def\@setauthors{% \vskip21pt plus6pt minus3pt \begin{center}% {\tiny BY\par}\medskip {\normalsize\scshape\authors\par}\medskip \def\address##1##2{\par ##2\par}% \def\email##1##2{\par e-mail:~##2\par}% {\small\itshape\addresses\par}% \end{center}} \def\enddoc@text{} \renewenvironment{abstract}{% \global\setbox\abstractbox=\vtop\bgroup \vskip10pt \normalfont\tiny\centering\MakeUppercase{\abstractname}\par \Small \list{}{% \leftmargin3pc \rightmargin3pc \labelwidth0pt \listparindent\normalparindent \itemindent0pt \parsep0pt plus1pt \let\fullwidthdisplay\relax }% \item[] }{% \vspace*{-10pt} \endlist\egroup } \newbox\aiusagebox \newenvironment{aiusage}{% \global\setbox\aiusagebox=\vtop\bgroup \normalfont\Small \list{}{% \leftmargin3pc \rightmargin3pc \topsep0pt \partopsep0pt \parsep0pt \itemsep0pt \listparindent0pt \itemindent0pt }% \item[]% \nobreak\vspace{4pt}% {\normalfont\tiny\centering\MakeUppercase{AI usage}\par}% \nobreak\vspace{2pt}% \noindent\ignorespaces }{% \par\endlist\egroup } \let\original@setabstracta\@setabstracta \def\@setabstracta{% \original@setabstracta \ifvoid\aiusagebox\else \vskip10pt \box\aiusagebox \prevdepth\z@ \fi } % Section headings. \let\@secnumfont\@empty \def\section{\@startsection{section}{1}{0pt}% {20pt plus4pt minus4pt} {.5\linespacing} {\normalfont\bfseries}} \def\subsection{\@startsection{subsection}{2}{0pt}% {.5\linespacing plus.7\linespacing} {-.5em} {\normalfont\scshape}} \def\subsubsection{\@startsection{subsubsection}{3}{0pt}% {.5\linespacing plus.7\linespacing} {-.5em} {\normalfont\itshape}} \let\@afterindenttrue\@afterindentfalse \@afterindentfalse % Theorem and proof styles. \newtheoremstyle{plain} {.5\linespacing}{.5\linespacing} {\itshape}{1em}{\scshape}{.}{.5em} {\thmname{#1}\thmnumber{ #2}\thmnote{\normalfont{} (#3)}} \theoremstyle{plain} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{corollary}[theorem]{Corollary} \newtheoremstyle{definition} {.5\linespacing}{.5\linespacing} {\upshape}{1em}{\scshape}{.}{.5em} {\thmname{#1}\thmnumber{ #2}\thmnote{\normalfont{} (#3)}} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{assumption}[theorem]{Assumption} \newtheoremstyle{remark} {.5\linespacing}{.5\linespacing} {\upshape}{1em}{\scshape}{.}{.5em} {\thmname{#1}\thmnumber{ #2}\thmnote{\normalfont{} (#3)}} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} \renewenvironment{proof}[1][\proofname]{% \par \pushQED{\qed}% \normalfont \topsep6pt plus6pt\relax \trivlist \item[\hskip\labelsep\slshape #1\@addpunct{.}]% \ignorespaces }{% \popQED \endtrivlist \@endpefalse } \DeclareRobustCommand{\qed}{% \leavevmode\unskip\penalty9999\hbox{}\nobreak \ifmmode \makebox[0pt][l]{\qquad\hbox{\qedsymbol}}% \else \qquad\hbox{\qedsymbol}% \fi } \renewcommand{\qedsymbol}{\rule[.47pt]{4pt}{6pt}} \makeatother \setlength{\emergencystretch}{1.8em} \numberwithin{equation}{section} \allowdisplaybreaks[1] \DeclareMathOperator{\supp}{supp} \DeclareMathOperator{\Var}{Var} \DeclareMathOperator{\Cov}{Cov} \DeclareMathOperator{\osc}{osc} \DeclareMathOperator{\sech}{sech} \title[The escape rate of favorite sites]{On the escape rate of favorite sites of planar random walks} \author{GPT-5.6 Sol \and Heng Ma} \thanks{Only Heng Ma takes responsibility for this work.} \date{} \begin{document} \begin{abstract} For planar simple random walk, the favorite sites at time $n$ are the sites whose local time at time $n$ is maximal. We prove that, almost surely, for every $\gamma>1/2$ and every $c>0$, all favorite sites lie outside the ball centered at the origin with radius $c\sqrt n/(\log n)^\gamma$ for all sufficiently large $n$. At the critical exponent $\gamma=1/2$, almost surely, for every $c>0$, the entire favorite sites lies within distance $c\sqrt{n/\log n}$ of the origin infinitely often. \end{abstract} \begin{aiusage} This manuscript was substantially generated and subsequently checked by AI systems. At the time of this version, the arguments and citations have not yet been independently verified in full by the human contributor. \textbf{A digested and more streamlined version will replace this preliminary version on both Hexagon and arXiv in the coming months.} \end{aiusage} \maketitle \begingroup \raggedbottom \vspace*{-20pt} \setcounter{tocdepth}{1} \tableofcontents \clearpage \endgroup % ---- introduction ---- \section{Introduction}\label{sec:introduction} A favorite site of a random walk is a site at which the accumulated local time is maximal. Although the planar walk returns to the origin infinitely often, its favorite sites need not remain near the origin. The question considered here is how small a neighborhood of the origin can still contain a favorite at arbitrarily large times. Because several sites can tie for the maximum, there are two distinct observables: the distance of the nearest favorite and the distance of the farthest favorite. Let $(S_n)_{n\ge0}$ be simple symmetric random walk on $\mathbb Z^2$, started at the origin. Define \begin{equation}\label{eq:local-time-definition} L_n(x)=\sum_{j=0}^n\mathbf1_{\{S_j=x\}},\qquad M_n=\max_{x\in\mathbb Z^2}L_n(x),\qquad \mathcal F_n=\{x\in\mathbb Z^2:L_n(x)=M_n\}. \end{equation} The set $\mathcal F_n$ is finite and nonempty. Put \begin{equation}\label{eq:radii-definition} R_n^- =\min_{x\in\mathcal F_n}\|x\|_2,\qquad R_n^+ =\max_{x\in\mathcal F_n}\|x\|_2, \qquad a_\gamma(n)=\frac{\sqrt n}{(\log n)^\gamma},\quad n\ge3. \end{equation} For example, along the path $(0,e_1,0,e_1)$ the two sites tie, so $R_3^-=0$ and $R_3^+=1$. Thus localizing an arbitrarily chosen favorite does not by itself localize the whole favorite set. \begin{theorem}[Lower-class escape rate]\label{main:precise-escape-rate} Almost surely, simultaneously for every real $\gamma$, \begin{align} \liminf_{n\to\infty}\frac{R_n^-}{a_\gamma(n)} &=\infty, &&\gamma>\tfrac12,\label{eq:main-escape}\\ \liminf_{n\to\infty}\frac{R_n^+}{a_\gamma(n)} &=0, &&\gamma\le\tfrac12.\label{eq:main-recurrence} \end{align} \end{theorem} In particular, for each $\gamma>1/2$ and each fixed $A>0$, eventually every favorite lies outside $B(0,Aa_\gamma(n))$. At the critical exponent, for every $\varepsilon>0$ there are infinitely many times at which \emph{all} favorites lie in $B(0,\varepsilon\sqrt{n/\log n})$. The conclusion identifies the boundary among logarithmic powers of the diffusive scale. It does not assert an integral test for arbitrary slowly varying perturbations of the critical gauge. Dembo, Peres, Rosen and Zeitouni~\cite[Theorem~1.1]{DPRZ2001} proved that every planar favorite lies at distance $n^{1/2+o(1)}$. This first-order exponent leaves the logarithmic scales in Theorem~\ref{main:precise-escape-rate} undetermined. Erd\H{o}s and R\'ev\'esz \cite[Question~14]{ErdosRevesz1987} asked whether planar favorite sites escape to infinity. R\'ev\'esz later asked for their rate of escape \cite[Chap.~19]{Revesz1990}, as recorded by Dembo~\cite[p.~6]{Dembo2005}. Dembo's ICM survey~\cite[Open Problem~2.1]{Dembo2006} asks more broadly about the evolution of favorite sites in dimensions $d\ge2$. Hao, Li, Okada and Zheng~\cite{HLOZ2025} recently proved that almost surely three favorite sites occur simultaneously infinitely often and eventually there is no simultaneous occurrence of four favorite sites. In \cite[Remark~1.4(1)]{HLOZ2025} they also highlight the planar escape-rate question. In one dimension the corresponding theorem has critical exponent $1$; see Lifshits and Shi~\cite[Theorem~1.1]{LifshitsShi2004} for the simple random walk and Bass~\cite[Theorem~2.2]{Bass2023} for both Brownian motion and the walk. Our focus is the location of the full set of planar maximizers. The two assertions in Theorem~\ref{main:precise-escape-rate} therefore retain the distinction between its nearest and farthest members throughout the proof. \subsection{Notation and the two local-time estimates} All logarithms are natural. We write $P_x$ and $E_x$ for the law and expectation of a walk started at $x$, and omit the subscript when $x=0$. Its transition operator is \[ Pf(x)=\frac14\sum_{y:\|y-x\|_2=1}f(y). \] By symmetry the same notation acts on lattice measures. For $r>0$ let $D(x,r)=\{y\in\mathbb Z^2:\|y-x\|_20$, $c>2$ and $A>0$, \begin{equation}\label{eq:intro-localized-upper} P\left(\max_{\|x\|_2\le e^JJ^{-\beta}} L_{\tau_{e^J}}(x)\ge u_{J,c}\right) \le J^{-2\beta+c-2+o(1)}+O(J^{-A}). \end{equation} For every fixed $\eta>0$, there is $p_\eta>0$ such that \begin{equation}\label{eq:intro-disk-lower} P(M_{\tau_{e^J}}\ge u_{J,2+\eta})\ge p_\eta \quad\text{for all sufficiently large }J. \end{equation} Since $a_\gamma(e^{2J})=e^J/(2J)^\gamma$, the disk in \eqref{eq:intro-localized-upper} is at the scale relevant to the theorem. Taking $\beta=\gamma$ and $c=2+2\delta$ makes its main term $J^{-2\gamma+2\delta+o(1)}$, summable over integer $J$ when $\gamma>1/2$ and $\delta>0$ is small enough. The lower estimate is amplified over independent segments to give an eventual almost-sure bound at deterministic exponential times. Comparing it with the upper estimate excludes a favorite near the origin throughout each time block. More precisely, if $N_j=\lfloor e^{2j}\rfloor$ and \[ \max_{x\in B(0,C e^{j+1}j^{-\gamma})}L_{N_{j+1}}(x)1$. The theorem needs $\beta>1/2$. The localized estimate \eqref{eq:intro-localized-upper} gains the missing factor $J^{-1}$ by keeping the traversal event and the terminal occupation in the same cell. Cover the target disk by cells centered at $y$, with terminal radius $M=e^h$ and core $V_y=D(y,M/2)$; write $L=J-h$. Let $T_j$ count the completed inward traversals around a cell at logarithmic level $j$. The upper argument uses $h=(C_h\log J)^4$, large enough that the first-crossing estimate applies through level $L$. Fix a positive barrier intercept $z$; from this point $z$ has this meaning in the traversal argument. The upper argument keeps the square-root traversal profile below $B_j=\rho_Jj+z+(j\wedge(J-j))^{1/4}$, where $\rho_J=2-(\log J)/J$. The slope records the $J\log J$ shift in an extreme traversal count; the curved buffer permits a summable bound on early barrier crossings. Fix a cell in source shell $\beta\log J+O(1)\le k\le C_0\log J$. For a fixed unit endpoint band $[q,q+1]$ in the central range $\epsilon_0L\le q\le B_L+1$, let $p=B_L+1-q\ge0$ be its deficit from the terminal barrier. Then the source-inclusive ballot estimate is \begin{equation}\label{eq:ideas-ballot} P(\text{good prefix},\ q\le\sqrt{2T_L}1$ makes the terminal disk large enough for the lattice comparison, while $\xi<2$ makes the moment screen and the error in \eqref{eq:ideas-terminal-lower} fit within the $J\log J$ margin. A conditional failure bound for each candidate cell, followed by a union bound, turns coarse-root existence into the positive probability estimate \eqref{eq:intro-disk-lower} without selecting a root using other bridge interiors. Finally, independent walk segments amplify disk success to an almost-sure lower bound at deterministic exponential times. Comparing that lower bound at the left end of a time block with the localized upper bound at its right end gives a block error with main term $j^{-2\gamma+2\delta+o(1)}$ and a summable exit error. For any $\gamma>1/2$, choose $\delta>0$ small enough to make the main term summable. This proves escape of $R_n^-$ throughout the blocks and completes the other half of the theorem. % ---- elementary ---- \section{Elementary estimates}\label{sec:elementary} For a finite lattice domain $D$, write \[ G_D(x,y)=E_x\sum_{j<\tau_{D^c}}\mathbf1_{\{S_j=y\}}, \qquad \tau_{D^c}=\inf\{j\ge0:S_j\notin D\}. \] The following estimates fix the normalization used throughout the paper. \begin{lemma}[Elementary walk estimates]\label{lem:elementary-walk-estimates} There are constants \(c,C\in(0,\infty)\) with the following properties. \begin{enumerate}\item If \(\tau_R=\inf\{n:\lVert S_n\rVert_2\ge R\}\), then \[ \sup_{\lVert x\rVertmCR^2)\le 2^{-m}. \] \item For all \(N,R\ge1\), \[ P\left(\max_{k\le N}\lVert S_k\rVert_2\ge R\right) \le C\exp\{-cR^2/N\}. \] \item Uniformly for \(1\ll r<\lVert x\rVert0\) such that, whenever \(2\le f\le N^{1/8}\), \begin{equation} \mu\left(B\left(0,C\frac{\sqrt N}{f}\right)\right) \ge c\beta\,\frac{1+\log f}{f^2}. \label{eq:rec-2} \end{equation} \end{lemma} \begin{proof} We test the resolvent identity against a function whose maximum is at the origin. This makes the unknown positive measure $\nu$ contribute with the needed sign in \eqref{eq:rec-3}. Let \(Q_R=[-R,R]^2\cap\mathbb Z^2\) and set \[ w_R:=|Q_R|^{-1}{\bf1}_{Q_R}*{\bf1}_{-Q_R}. \] Then \(0\le w_R\le1\), \(w_R(0)=1\), and \(w_R\) is supported in \(B(0,3R)\). Put \[ G_z:=\sum_{n\ge0}z^nP^n,\qquad \phi_R:=G_z*w_R. \] Fourier inversion gives \[ \widehat{\phi_R}(\theta)= \frac{|\widehat{{\bf1}_{Q_R}}(\theta)|^2/|Q_R|} {1-z\widehat P(\theta)}\ge0. \] Consequently \(\phi_R(y)\le\phi_R(0)\) for every \(y\). Pairing \eqref{eq:rec-1} with \(\phi_R\) and using \((I-zP)\phi_R=w_R\) yields \begin{equation} \begin{split} \mu(w_R) &=\alpha\phi_R(0)-\nu(\phi_R)\\ &\ge(\alpha-\nu(\mathbb Z^2))\phi_R(0) =(1-z)\beta\phi_R(0). \label{eq:rec-3} \end{split} \end{equation} For $\|\theta\|\le c/R$, the numerator in the Fourier transform is at least $cR^2$, while $1-z\widehat P(\theta)\asymp N^{-1}+\|\theta\|^2$. Hence, for $R^2\le N/4$, polar integration gives \begin{equation} \begin{aligned} \phi_R(0) &\ge cR^2\int_0^{c/R}\frac{t\,dt}{N^{-1}+t^2}\\ &\ge cR^2\left(1+\log\frac N{R^2}\right). \end{aligned} \label{eq:rec-4} \end{equation} Since \(\mu(w_R)\le\mu(B(0,3R))\), take \(R=\lfloor\sqrt N/(4f)\rfloor\) in \eqref{eq:rec-3}--\eqref{eq:rec-4}. This proves \eqref{eq:rec-2}, after changing the absolute radius constant. \end{proof} \begin{lemma}[Localization of a selected favorite]\label{lem:plateau-bijection} Let $0V_0\}},\qquad y\ne0. \label{eq:rec-7} \end{equation} Indeed, a better-priority site must be strictly below the root's value \(r+1\), while a worse-priority site may tie it. We now fix all pieces except the terminal arm $C$, together with the translated priorities: \[ \mathcal{B}=(r,E_1,\ldots,E_r,B,V). \] The local time still available to $C$ at $y\ne0$ is the capacity \begin{equation} c_\mathcal{B}(y):= r+{\bf1}_{\{V_y>V_0\}}-H_r(y)-N_B(y),\qquad y\ne0. \label{eq:rec-8} \end{equation} Only tuples with $c_{\mathcal B}(y)\ge0$ for every $y\ne0$ can occur. For such a tuple, run a new walk $C$ from zero and let $\tau_c$ be its first positive return to zero or its first time at which some occupation count $N_C(y)$ exceeds $c_{\mathcal B}(y)$. Condition \eqref{eq:rec-7} says exactly that the admissible terminal arms are the prefixes $(C_0,\ldots,C_q)$ with $q<\tau_c$. Conversely, a feasible base tuple and one such prefix reconstruct the original path: concatenate \eqref{eq:rec-6}, reverse it, translate by \(-a=x\), and append the translated suffix \(B\). Inequality \eqref{eq:rec-7} makes \(x\) a favorite and excludes every better-priority tie; avoidance of zero by \(B\) makes \(\ell\) its last visit. The endpoint \(a\) recovers the sector \(x=-a\), so both the path and the priority translation are unique. The forward and inverse maps are therefore genuine bijections, including \(r=0\). \smallskip\noindent\underline{\textit{Step 2.}}\quad Identify the stopped occupation measure.\nopagebreak[4] Write $q=|C|$ and $k=|B|$ for the terminal-arm and suffix lengths, so that $n=\rho_r+q+k$. Reversal and symmetry preserve the path weight, and \begin{equation} z^{\rho_r}4^{-\rho_r}\, z^q4^{-q}\,(1-z)z^k4^{-k} =(1-z)z^n4^{-n}. \label{eq:rec-9} \end{equation} For each fixed tuple, the factors outside $z^q4^{-q}$ are fixed. Summing over its admissible terminal arms therefore gives their stopped occupation measure, with no multiplicity or extra normalization. For a fixed capacity let \[ g_c(u)=E_0\sum_{q<\tau_c}z^q{\bf1}_{\{C_q=u\}}. \] By the Markov property, \[ zPg_c(u) =E_0\sum_{q<\tau_c}z^{q+1}{\bf1}_{\{C_{q+1}=u\}} =g_c(u)-{\bf1}_{\{u=0\}} +E_0[z^{\tau_c}{\bf1}_{\{C_{\tau_c}=u\}}]. \] Thus the occupation terms cancel except at the initial and stopped endpoints: \begin{equation} (I-zP)g_c=\delta_0-E_0[z^{\tau_c}\delta_{C_{\tau_c}}]. \label{eq:rec-10} \end{equation} The arm endpoint is $a=-x$. Thus the bijection and \eqref{eq:rec-9} identify the law of $-X_K^*$ as a positive mixture of the $g_c$'s, integrating also over the translated i.i.d. priorities. Mixing \eqref{eq:rec-10} gives \eqref{eq:rec-1}; its mass relation follows by summing over $\mathbb Z^2$. Restricting the tuples by any event depending only on $\mathcal B$ merely removes positive mixture weights, so the same identity holds for the restricted endpoint measure. \smallskip\noindent\underline{\textit{Step 3.}}\quad Compare the deterministic and random clock scales.\nopagebreak[4] Lemma~\ref{lem:stopped-resolvent} gives a radius proportional to $\sqrt N/f$. To obtain $\sqrt K/f$, we restrict the mixture to an event of fixed positive probability on which $K\ge N/8$. This restriction must depend only on $\mathcal B$, so that the resolvent identity still applies. Whole-path reversal, translation to the new starting point, and transport of the i.i.d. priorities preserve the joint law and send \[ (\sigma,\ell)\longmapsto(n-\ell,n-\sigma). \] Conditionally on $K=n$, the pair has this symmetry. In particular, $E[\sigma+\ell\mid K=n]=n$. Since $\sigma\le\ell$ and $0\le n-\sigma\le n$, it follows that \begin{equation} E[n-\sigma\mid K=n]=E[\ell\mid K=n]\ge n/2, \qquad P(n-\sigma\ge n/4\mid K=n)\ge1/3. \label{eq:rec-11} \end{equation} Also \(P(K\ge N/2)\ge1/4\) for all large \(N\). Hence the restriction \[ \Gamma=\{n-\sigma\ge N/8\} \] has mixture mass \(\beta\ge1/12\). In the tuple, \[ n-\sigma=(\ell-\sigma)+(n-\ell)=\rho_r+|B|, \] so $\Gamma$ depends only on the fixed tuple. Apply Lemma~\ref{lem:stopped-resolvent} to this restricted endpoint measure, whose mass is $\beta\ge1/12$. It gives probability at least $c(1+\log f)/f^2$ in a disk of radius $C_1\sqrt N/f$. On $\Gamma$ we have $K\ge N/8$, so this radius is at most $\sqrt8 C_1\sqrt K/f$. Since $\|-X_K^*\|_2=\|X_K^*\|_2$, absorbing the fixed factor proves \eqref{eq:rec-5}. \end{proof} \subsection{Removing ties and absorbing the past} \begin{lemma}[Localization with a unique favorite]\label{lem:singleton-geometric-localization} For the same geometric clock, uniformly for \(2\le f\le N^{1/8}\), \begin{equation} P\left(|\mathcal F_K|=1,\ R_K^+\le C\frac{\sqrt K}{f}\right) \ge c\,\frac{1+\log f}{f^2}. \label{eq:rec-12} \end{equation} The conclusion remains valid, up to a universal change of \(c\), after restricting the output clock to \begin{equation} aN\le K\le N\log N,\qquad a=\frac c8\,\frac{1+\log f}{f^2}. \label{eq:rec-13} \end{equation} \end{lemma} \begin{proof} For every atom counted in \eqref{eq:rec-5}, insert immediately after the last visit to its priority-selected favorite \(x\) the fixed loop \begin{equation} x,\ x+e_1,\ x,\ x+e_2,\ x. \label{eq:rec-14} \end{equation} The local time at \(x\) rises by two, those at the two neighbors rise by one, and all other local times are unchanged. Thus \(x\) is the unique favorite of the new path. The old suffix avoids \(x\), so in the image the last visit to the unique favorite is the final vertex of \eqref{eq:rec-14}. Removing the four preceding prescribed steps is therefore the unique inverse. On geometrically weighted path atoms the image-to-source weight ratio is \(z^4/4^4\), bounded below for \(z\) close to one. This proves \eqref{eq:rec-12}. Now $P(KN\log N)\le\exp\{-(\log N)/2\}$. With the constant in \eqref{eq:rec-13} chosen after the constant in \eqref{eq:rec-12}, these discarded masses total at most half the right-hand side of \eqref{eq:rec-12}. Four added steps are absorbed by changing the universal constants. \end{proof} \begin{lemma}[Adding an inherited local-time profile]\label{lem:inherited-profile-dichotomy} Let \(H\ge0\) be any finitely supported integer profile and let \(\ell\ge0\) be a second integer profile with a unique maximizer \(y\). Every maximizer of \(H+\ell\) lies in \(\operatorname{supp}H\cup\{y\}\). \end{lemma} \begin{proof} For every $x\notin\operatorname{supp}H\cup\{y\}$, uniqueness of the maximum of $\ell$ gives \[ (H+\ell)(x)=\ell(x)<\ell(y)\le(H+\ell)(y). \] Such a site cannot maximize $H+\ell$. \end{proof} \subsection{Infinitely many successful blocks} \begin{proposition}[Recurrence at the critical scale]\label{prop:critical-recurrence} Almost surely, \begin{equation} \liminf_{n\to\infty} \frac{R_n^+}{\sqrt{n/\log n}}=0. \label{eq:rec-15} \end{equation} Consequently, for every \(\gamma\le\tfrac12\), \begin{equation} \liminf_{n\to\infty}\frac{R_n^+}{a_\gamma(n)}=0. \label{eq:rec-16} \end{equation} \end{proposition} \begin{proof} \smallskip\noindent\underline{\textit{Step 1.}}\quad Choose independent fresh segments.\nopagebreak[4] Fix $\varepsilon>0$. We need fresh segments long enough that the old range is small compared with $\sqrt{K_k/\log K_k}$, while each segment still ends before the next deterministic block. Choose $J_1$ sufficiently large and define \[ J_{k+1}=J_k+10\log J_k,\qquad s_k=\lfloor e^{J_k}\rfloor, \] Here $J_k$ is the logarithmic block time. For the fresh block beginning at $s_k$, choose the geometric time scale and localization divisor as \[ N_k=\frac{e^{J_{k+1}}}{J_{k+1}^2},\qquad f_k=C\varepsilon^{-1}\sqrt{J_{k+1}}. \] Choose $C$ large enough that localization gives the radius below; for all large $k$, $2\le f_k\le N_k^{1/8}$. Apply Lemma~\ref{lem:singleton-geometric-localization} to the fresh increment walk, with an independent geometric clock with parameter $1/N_k$. It gives a unique fresh favorite \(Y_k\) satisfying \begin{equation} \lVert Y_k\rVert\le\frac{\varepsilon}{8} \sqrt{\frac{K_k}{J_{k+1}}} \label{eq:rec-17} \end{equation} with conditional probability at least \begin{equation} d_k=c_\varepsilon\frac{\log J_{k+1}}{J_{k+1}}, \label{eq:rec-18} \end{equation} even after \begin{equation} \frac{d_k}{8}N_k\le K_k\le J_{k+1}N_k. \label{eq:rec-19} \end{equation} Since $e^{J_{k+1}}=e^{J_k}J_k^{10}$ and $J_{k+1}\sim J_k$, \eqref{eq:rec-18}--\eqref{eq:rec-19} give, uniformly over the clock restriction, \[ c'_{\varepsilon}e^{J_k}J_k^7\log J_k \le K_k\le\frac{e^{J_{k+1}}}{J_{k+1}} \] for all sufficiently large $k$. The lower bound makes the fresh segment dominate the past, while the upper bound keeps it within its deterministic block: $K_k/s_k\to\infty$ and $s_k+K_k0\) proves \eqref{eq:rec-15}. Finally, \[ \frac{a_{1/2}(n)}{a_\gamma(n)} =(\log n)^{\gamma-1/2}\le1 \] eventually for \(\gamma\le1/2\), which proves \eqref{eq:rec-16}. \end{proof} % ---- continuum_transfer ---- \section{Continuum barriers and lattice traversal counts} \label{sec:continuum-transfer} Traversal counts separate the large-scale accumulation of excursions from occupation inside the last annulus. We first record the continuum barrier estimates in the form needed below. We then compare rare upper events by multiplying one-step lattice probabilities. For the lower event, a single strong coupling transfers the existence of a suitable Brownian center. The two comparisons have different purposes: the first preserves probabilities on their exponential scale, while the second preserves a positive probability of existence. Together they give the three lattice inputs used later: an upper bound on barrier crossings, a joint bound for a barrier-respecting traversal endpoint, and a positive probability of finding a lower root. The first two enter the localized upper bound; the third enters the lower bound for the disk maximum. \subsection{Spherical traversal estimates} We use Rosen's spherical traversal construction \cite{Rosen2022}. Write $d$ for geodesic distance on the sphere. We denote the geodesic radii by $\vartheta_j$ (Rosen's $h_j$), reserving $h$ for the terminal logarithmic radius $\log M$. The upper and lower arguments use separate fixed choices of the initial radius. We keep the customary notation $r_0$ in each argument and state the change explicitly; the two choices are never used in the same pathwise comparison. Let $X$ be spherical Brownian motion started at $v$, stopped at its first hit $\tau$ of $\partial B_d(v,r_*)$. When defining source stopping times we extend it by an independent Brownian continuation; the safety event below will imply that every portion actually used lies before $\tau$. Put \begin{equation} r_j=r_0e^{-j},\qquad \vartheta_j=2\arctan(r_j/2). \label{eq:ct-external-23} \end{equation} For the upper estimates, choose the fixed radii as in \cite[Section~2, (2.2)]{Rosen2022}, in particular $2r_*\le \vartheta_0$. Let $F_j$ be a maximal $d_0\vartheta_j$-separated set, augmented if necessary to a $d_0\vartheta_j$-cover; thus $|F_j|\asymp e^{2j}$. For $y\in F_L$, $T_{y,j}$ is the number, before $\tau$, of completed excursions from $\partial B_d(y,\vartheta_{j-1})$ to $\partial B_d(y,\vartheta_j)$, with the initial and final incomplete pieces omitted. Its source shell is \begin{equation} k_y=\inf\{k:y\notin B_d(v,\vartheta_k)\}.\label{eq:ct-external-24} \end{equation} Pairs have separation shell $k$ when $2\vartheta_k\alpha_{z,+}(l). \] Rosen's representative construction is deterministic. With fixed tie-breaking, every fine center is sent to its closest point of the predetermined coarser net at scale $\vartheta_{l+\frac14 (l\wedge(L-l))^{1/4}}$; equations (2.26)--(2.33) give the buffered ball inclusions and domination of its traversal count. Proposition 2.5, specifically equations (2.34)--(2.35), then proves the complete estimates \begin{equation} \#\{\hbox{representatives with source }k\} \le C\exp\{2(l-k)+\tfrac12(l\wedge(L-l))^{1/4}\},\label{eq:ct-external-26} \end{equation} \begin{equation} P(U(y,l,z))\le C\exp\{-2l-(l\wedge(L-l))^{1/4}\}e^{-2z}.\label{eq:ct-external-27} \end{equation} They hold pointwise for $1\le k\le l\le L-(4\log L)^4$. Only the choice $F_L\cap B_d(v,\vartheta_{\log L})^c$, and not the proof of \eqref{eq:ct-external-26}--\eqref{eq:ct-external-27}, imposed $k\le\log L$. Hence, for any $s\le C_0\log L$, summing just the representatives whose source shell is at least $s$ gives the bound below. Indeed, the product of \eqref{eq:ct-external-26} and \eqref{eq:ct-external-27} is $Ce^{-2k}e^{-(l\wedge(L-l))^{1/4}/2}e^{-2z}$, and \[ \sum_{k\ge s}e^{-2k} \sum_{l=k}^{L-(4\log L)^4}e^{-(l\wedge(L-l))^{1/4}/2} \le Ce^{-2s}. \] Here the inner sum is uniformly bounded by the two endpoint tails. Consequently, \begin{equation} P\big(\exists y:k_y\ge s,\ \exists l\le L-(4\log L)^4: U(y,l,z)\big) \le C e^{-2s+o(\log L)}e^{-2z}.\label{eq:ct-external-28} \end{equation} \subsubsection{The endpoint estimate for complete centered excursions} The endpoint calculation must be applied to a complete centered excursion forest. To specify that law, start spherical Brownian motion at $v$ with $\vartheta_{k_y}\le d(v,y)<\vartheta_{k_y-1}$ and continue it until its first exit from $B_d(y,\vartheta_0)$. Let $T^{\mathrm c}_{y,j}$ count its completed traversals at level $j$, and let $\mathcal{G}^{\mathrm c}_{y,L,z}$ require the upper barrier in \eqref{eq:ct-external-25} from $m=k_y+1$ through $L-1$. Put $Y_y^{\mathrm c}=\sqrt{2T^{\mathrm c}_{y,L}}$ and $B_L=\rho_LL+z$. Uniformly for $k_y\le C_0\log L$, every fixed $\epsilon_0>0$, $\epsilon_0L\le q\le B_L+1$, and fixed $z$, the source law in \cite[(2.8)]{Rosen2022} and its Appendix Theorem~9.1(a) give \begin{equation} P\big(\mathcal{G}^{\mathrm c}_{y,L,z},Y_y^{\mathrm c}\in[q,q+1]\big) \le C(k_y+1)^C(2+B_L-q)L^{-2}e^{-q^2/(2L)}.\label{eq:ct-external-29} \end{equation} This assertion concerns the centered stopping rule just defined. We do not identify the descendants of a population stopped at the off-center boundary $\partial B_d(v,r_*)$ with an independent Galton--Watson forest. The later lattice endpoint estimate will be proved directly for its centered continuation. The following version of \cite[Appendix, Theorem~9.1(a)]{Rosen2022} is sufficient: if $T$ is critical Galton--Watson with $P(\xi=m)=2^{-m-1}$, $T_0=x^2/2$, and \[ \begin{gathered} \delta>0,\quad C\ge0,\quad\eta>1,\quad\epsilon\in(0,1/2),\\ 2\le x,y\le\eta L,\quad x^2/2\in\mathbb N,\\ 0\le x\le a,\quad0\le y\le b, \end{gathered} \] \begin{equation} \begin{split} P\big(&\sqrt{2T_j}\le a+(b-a)j/L +C(j\wedge(L-j))^{1/2-\epsilon},\ 1\le jN/2$ use the right-end distance directly. Thus every parameter in \eqref{eq:ct-external-30} has been specified, including $x\le a$, $q\le b$, and $x,q\le\eta N$. The sole positive start not covered by the hypothesis $x\ge2$ is $t=1$. Couple a process from two ancestors as the sum of two independent one-ancestor processes. On the event, of probability $1/2$, that the second ancestor has no child in the first generation, the total process agrees from generation $1$ onward with the first process. The barrier in \eqref{eq:ct-external-30} is imposed precisely at generations $1,\ldots,N-1$, so the $t=1$ contribution is at most twice the same event for initial population two, to which \eqref{eq:ct-external-30} applies. The case $t=0$ cannot have the positive endpoint $q\ge\epsilon_0L$. The source-height cost and the descendant cost combine through the identity \begin{equation} \frac{x^2}{m}+\frac{(q-x)^2}{N} =\frac{q^2}{L}+ \frac{L(x-mq/L)^2}{mN}.\label{eq:ct-external-31} \end{equation} In particular, their product always contains $e^{-q^2/(2L)}$; the remaining exponential factor is at most one. Since $q\ge\epsilon_0L$, $N\asymp L$, and $x\le Cm$, the square-root factor in \eqref{eq:ct-external-30}, together with its explicit $N^{-1}$, is at most $Cm^{1/2}L^{-2}$. Multiplying \eqref{eq:ct-external-30a} by \eqref{eq:ct-external-30} and summing the $O(m)$ possible unit source-height bands therefore gives \[ P(\mathcal G^{\mathrm c}_{y,L,z},Y_y^{\mathrm c}\in[q,q+1]) \le C m^C(2+B_L-q)L^{-2}e^{-q^2/(2L)}, \] which is \eqref{eq:ct-external-29}. All source and band-counting factors are absorbed in $m^C$. No estimate of the form \eqref{eq:ct-external-29} is asserted at $q=0$: extinction there has order-one probability. In the lattice upper bound below, $q<\epsilon_0L$ is instead disposed of by the saturated terminal Chernoff factor. We shall also use the following truncated version. Set $L=J-h$ and $N=L-m$, with $m\le C_0\log J$. For the barrier \[ B_l=\rho_Jl+z+(l\wedge(J-l))^{1/4}, \] the estimate \eqref{eq:ct-external-29} holds using $B_l$ in the no-crossing event and $B_L$ on its right-hand side, for $\epsilon_0L\le q\le B_L+1$. To verify the only new point, put $\phi(l)=(l\wedge(J-l))^{1/4}$. Its two Hölder bounds from $m$ and $L$ give \begin{equation} \phi(m+j)\le(1-j/N)\phi(m)+(j/N)\phi(L) +C(j\wedge(N-j))^{1/4}.\label{eq:ct-external-31a} \end{equation} The affine part $\rho_Jl+z$ interpolates exactly, so \eqref{eq:ct-external-30} applies with the arbitrary endpoints $a=B_m,b=B_L+1$. In particular, no $\rho_L$-barrier theorem is being assumed. \subsubsection{A fixed number of source excursions} For the lower event, now choose $r_0$ sufficiently small that $4\vartheta_{-1}0$ from that section so that \begin{equation} m_0:=x_*^2\in\mathbb N.\label{eq:ct-external-32a} \end{equation} (This $x_*$ is not the variable called $x$ in Appendix Theorem 9.1; the latter equals $\sqrt{2m_0}$ when that theorem is started with $m_0$ particles.) Put \[ F_L^0=F_L\cap\{\vartheta_1/40\le d(v,y)\le \vartheta_1/20\}. \] This fixed annulus, instead of Rosen's full ball, retains a fixed positive area fraction; the one- and two-point sums in Rosen's proof are therefore unchanged up to constants, and every source index is $O(1)$. We spell out the superscript in Rosen (3.3), because it is important for the stopping-time comparison below. Write $\Gamma_{y,j}=\partial B_d(y,\vartheta_j)$ and set \[ A_{y,1}=H_{\Gamma_{y,1}},\qquad B_{y,i}=\inf\{t>A_{y,i}:X_t\in\Gamma_{y,0}\}, \] \begin{equation} A_{y,i+1}=\inf\{t>B_{y,i}:X_t\in\Gamma_{y,1}\}, \qquad 1\le iC_{i,l,q}:X_t\in\Gamma_{y,l}\},\qquad C_{i,l,q+1}=\inf\{t>D_{i,l,q}:X_t\in\Gamma_{y,l-1}\}, \] and put \begin{equation} T_{y,l}^{1,m_0}=\sum_{i=1}^{m_0} \#\{q:D_{i,l,q}0$. Thus the safe connector kernel dominates $\kappa_*$ times the unrestricted kernel pointwise. Iterating the strong Markov factorization, with the whole later path functional left in the last factor, gives \begin{equation} P(I_{y,z}\cap S_y)\ge\kappa_*^{m_0-1}P(I_{y,z}).\label{eq:ct-external-38b} \end{equation} The pointwise kernel bound can be integrated against every nonnegative function of the later path, so it preserves the barrier and angular conditions in $I_{y,z}$ together. Before endpoint truncation put \[ \widetilde Z_L^{\rm sf}:=\sum_{y\in F_L^0}{\bf1}_{I_{y,z_0}\cap S_y}. \] Equation \eqref{eq:ct-external-38b} preserves its first moment, whereas $I_{y,z_0}\cap S_y\subset I_{y,z_0}$ preserves the pair upper bound \eqref{eq:ct-external-36}. Since $|F_L^0|\asymp e^{2L}$, the one-center lower bound gives $E\widetilde Z_L^{\rm sf}\ge c$. For the second moment, group distinct pairs by separation shell $k$. There are at most $Ce^{4L-2k}$ pairs in shell $k$, so \eqref{eq:ct-external-36} gives \[ E[(\widetilde Z_L^{\rm sf})^2] \le C+C\sum_{k=0}^{L-d_*} e^{-c(k\wedge(L-k))^{1/4}}\le C. \] The last $d_*$ shells contribute only $O(1)$ by the one-center bound. Paley--Zygmund now gives \begin{equation} E\widetilde Z_L^{\rm sf}\le C,\qquad P(\widetilde Z_L^{\rm sf}>0)\ge c_1>0.\label{eq:ct-external-38c} \end{equation} Rosen chooses $r_0$ so that $4\vartheta_{-1}0$ such that \begin{equation} B_d(y,\vartheta_{-1})\subset B_d(v,r_*-\chi).\label{eq:ct-external-38d} \end{equation} Every source block and every connector on $S_y$ lies in the left side, hence ends before $\tau$. This is the exact stopped-source certificate used in the coupling below. In particular, on $S_y$, \begin{equation} T_{y,L}^{1,m_0}\le T_{y,L}^{\tau}.\label{eq:ct-external-38e} \end{equation} Rosen's Theorem~1.6 gives the following upper tail on $F_L$; extending its bound to bounded nonnegative $z$ only enlarges the constant: \begin{equation} P\left(\sup_{y\in F_L}\sqrt{2T_{y,L}^{\tau}}\ge\rho_LL+z\right) \le C(1+z)e^{-2z},\qquad0\le z\le\log L.\label{eq:ct-external-38f} \end{equation} Choose $C_*$ so large that the right side at $z=z_0+C_*$ is below $c_1/2$, and now define $I_y^\circ$ by \eqref{eq:ct-external-37}. Equations \eqref{eq:ct-external-38c}--\eqref{eq:ct-external-38f} give, with \[ Z_L^{\rm sf}:=\sum_{y\in F_L^0}{\bf1}_{I_y^\circ\cap S_y}, \] \begin{equation} EZ_L^{\rm sf}\le C,\qquad P(Z_L^{\rm sf}>0)\ge c_1/2.\label{eq:ct-external-38} \end{equation} The constant $C_*$ and the safe count are now fixed. \begin{lemma}[Continuum barrier inputs]\label{lem:external-barrier-input} In the upper choice of radii, the first-crossing estimate \eqref{eq:ct-external-28} holds. For the complete centered traversal process, the endpoint estimate \eqref{eq:ct-external-29} holds, including the truncated barrier described in \eqref{eq:ct-external-31a}. In the lower choice of radii, on $I_y^\circ\cap S_y$ all first $m_0$ source excursions finish before $\tau$, and the number of centers with this event satisfies \eqref{eq:ct-external-38}. \end{lemma} \begin{proof}[Proof of Lemma~\ref{lem:external-barrier-input}] The representative estimates are the packing and one-center bounds of \cite[Proposition~2.5, (2.34)--(2.35)]{Rosen2022}, summed over the indicated source shells. The complete centered source calculation above proves \eqref{eq:ct-external-29}, and \eqref{eq:ct-external-31a} checks the change of barrier at a truncated horizon. The lower construction combines \cite[Lemmas~3.3--3.4]{Rosen2022} with the connector-kernel domination \eqref{eq:ct-external-38b} and the terminal truncation. All the first- and second-moment assertions \eqref{eq:ct-external-35}--\eqref{eq:ct-external-38} remain in Rosen's original spherical geometry. In particular, we do \emph{not} infer an event-level Euclidean two-center theorem from Rosen's occupation-measure argument in Section 8. The next lemma instead maps the actual spherical interfaces into the plane, so the spherical event and its pair law are preserved exactly until the one-sided lattice coupling is made. \end{proof} \subsection{Comparison of smooth interface kernels} The word comparison requires relative errors for the probabilities of successive radial decisions. We obtain them from a one-interface estimate: compare harmonic boundary payoffs, take the payoff equal to one, and then multiply the resulting row-mass bounds along a complete word. Fix $\alpha\in(0,1)$. Let $(\Gamma_R^-,\Gamma_R^0,\Gamma_R^+)$ range over triples of disjoint Jordan curves which, after translation, rotation, and scaling by $R^{-1}$, together with their tubular-coordinate charts, belong to a bounded $C^{4,\alpha}$ family. Assume consecutive curves are separated by distances comparable to $R$, and that a specified conformal coordinate sends each triple to three concentric logarithmic levels. The interfaces used below are circles or uniformly Möbius-transformed circles, so this hypothesis entails no loss. For a boundary function put \begin{equation} \|f\|_{4,R}:=\sum_{j=0}^4R^j\|\partial_\tau^jf\|_\infty +R^{4+\alpha}[\partial_\tau^4f]_\alpha .\label{eq:ct-smooth-39a} \end{equation} A lattice starting state is a directed edge whose endpoints lie on opposite sides of $\Gamma_R^0$; the lattice query starts from its post-crossing endpoint and the Brownian query from the normal projection of that endpoint onto $\Gamma_R^0$. The side of a vertex is determined by membership in the bounded component of the complement (vertices lying on the curve are assigned to the exterior). Let $Q_R^B$ be the Brownian selected-side subprobability kernel, stopped at first contact with either competing curve, and let $Q_R^S$ be the lattice selected-side subprobability kernel, stopped at the first directed edge whose endpoint-side indicators differ for one of the two competing curves. A tangency, or any intersection of an interpolated edge whose two endpoints remain on the same side, is not a lattice crossing. Write $\pi_0,\pi_+$ for normal projection of the digital source and selected target states. For a function $f$ on the selected target curve and a function $g$ on the middle curve, pull them back to the corresponding digital states by \[ U_+f=f\circ\pi_+, \qquad U_0g=g\circ\pi_0. \] \begin{lemma}[Comparison of interface kernels] \label{lem:smooth-interface-discretization} As an estimate from functions on the selected target curve to functions on the digital source-edge space, \begin{equation} \|Q_R^SU_+f-U_0(Q_R^Bf)\|_\infty \le CR^{-1}\|f\|_{4,R}.\label{eq:ct-smooth-39b} \end{equation} If the Brownian side probability $p_R=Q_R^B1$ is constant on the middle curve and lies in $[c,1-c]$, then, uniformly in the starting angle, \begin{equation} Q_R^S1=p_R(1+O(R^{-1}))\label{eq:ct-smooth-39c} \end{equation} and the corresponding row-normalized angular kernels differ by $O(R^{-1})$ on $C^4$ tests. Consequently, let a chronological radial word consist of $N$ ordinary decisions between neighboring concentric interfaces at scales $R_i\ge M$. The next queried triple may depend on the already exposed radial prefix, but not on an unexposed angular mark. For a fixed word, let $E_i$ be the directed-edge state space after its first $i$ letters and let $K^S_{i,w_{i+1}}(e,de')$ be the subkernel realizing the next prescribed side. If $p_i$ is the corresponding Brownian side probability, then, for every current edge, \begin{equation} e^{-C/R_i}p_i\le K^S_{i,w_{i+1}}1(e) \le e^{C/R_i}p_i .\label{eq:ct-smooth-39cprime} \end{equation} After summing all compatible directed-edge refinements, \begin{equation} e^{-C\sum_iR_i^{-1}}P^B(w) \le P^S(w)\le e^{C\sum_iR_i^{-1}}P^B(w).\label{eq:ct-smooth-39d} \end{equation} The same inequality holds after integrating the initial edge against any entrance law and after summing a disjoint family of complete words. A radial-prefix stopping rule is padded by cemetery letters of row mass one. No assertion is made after conditioning on an edge list generated later by the word. Source-root and source-death letters are handled explicitly in Lemma \ref{lem:buffered-transfer}. \end{lemma} \begin{proof}[Proof of Lemma~\ref{lem:smooth-interface-discretization}] There are two errors to control: the discrete walk accumulates a small harmonicity defect before exit, and its crossing endpoints lie within one lattice unit of the smooth interfaces. The first error is $O(R^{-2})$; the second gives the stated $O(R^{-1})$ bound. Let $\Omega_R$ be the annulus between the two competing curves and let $u$ be the continuum Dirichlet solution with boundary data $f$ on the selected component and zero on the other. Uniform rescaled boundary regularity gives \begin{equation} \|D^ju\|_{L^\infty(\overline\Omega_R)} \le CR^{-j}\|f\|_{4,R},\qquad0\le j\le4.\label{eq:ct-smooth-39e} \end{equation} The compact rescaled family has a tubular radius $\rho R$. In a signed normal chart $\Phi(s,t)$, with $t\ge0$ on $\Omega_R$, choose fixed numbers $c_1,\ldots,c_5$ satisfying \[ \sum_{a=1}^5c_a(-a)^q=1, \qquad q=0,1,2,3,4, \] and, for $-\rho R/5\eta_i:S_{t-1}\notin D_{y,j},\ S_t\in D_{y,j}\},\\ \eta_{i+1}&=\inf\{t>\sigma_i:S_{t-1}\in D_{y,j-1},\ S_t\notin D_{y,j-1}\}. \end{aligned}\label{eq:ct-transfer-39h} \end{equation} A state records the directed crossing edge; interpolation is used only to assign its projected $\chi_y$-angle. Only portions belonging to the first $m_0$ source intervals $[A_{y,i},B_{y,i}]$ in \eqref{eq:ct-external-32b} contribute traversal counts or terminal bridges. Inter-source connectors contribute only their guarded coarse boundary decisions, never their interior occupation. The whole segment from $\sigma_i$ to $\eta_{i+1}$, including further contacts with the inner interface, is one outward piece. Recursing inside each parent excursion gives the standard nested excursion tree. The lower \emph{radial word} $W_y=(W_{y,1},\ldots,W_{y,m_0})$ is a fixed collection of $m_0$ centered words, one for each source interval, with all angles erased; the complete guard record additionally contains the finitely many connector boundary decisions. If $\ell_{y,i}\in\{1,-1\}$ records whether the connector after $B_{y,i}$ first hits $\Gamma_{y,1}$ or $\Gamma_{y,-1}$, write \begin{equation} \overline W_y=(W_{y,1},\ell_{y,1},W_{y,2},\ldots, \ell_{y,m_0-1},W_{y,m_0})\label{eq:ct-transfer-39hprime} \end{equation} for the complete conditioned guard word. Its traversal coordinate $T^S_{y,j}$ is the sum over these words of completed crossings from $\Gamma_{y,j-1}$ to $\Gamma_{y,j}$. At the last fixed number of levels, the interfaces are replaced through one unit-width digital collar by concentric circles about the nearest lattice point to $\Psi_J(y)$, of radii $M_y,eM_y$ with $M_y\in[cM,CM]$. We relabel $M_y$ as $M$; every later estimate is uniform under this fixed-factor change and $\log M_y=h+O(1)$. Define $\mathcal{R}_y$ by the chronological erasure map which replaces each completed terminal segment, from its inward directed crossing of radius $M$ through its next outward directed crossing of radius $eM$, by the two endpoint edges. It retains $\overline W_y$, all radial letters and all crossing edges and projected angles outside the erased segments, but it does not retain an erased segment's duration, interior crossings, path, or occupation. Inward connector pieces outside these erased segments are exposed. This quotient sigma-field is the intrinsic coarse sigma-field. The upper application is instead defined directly on the lattice and is entirely centered. Write $N=e^J$, $\tau_N=\tau_{D(0,N)}$, and, for each candidate cell center with $\|y\|<2N$, set \begin{equation} \sigma_y=\tau_{D(y,3N)}.\label{eq:ct-transfer-39k} \end{equation} The pathwise inclusions \begin{equation} D(0,N)\subset D(y,3N),\qquad \tau_N\le\sigma_y, \qquad L_{\tau_N}(x)\le L_{\sigma_y}(x)\label{eq:ct-transfer-39l} \end{equation} hold for every $x$. For each one-cell probability we decompose the extended path up to $\sigma_y$ using only circles centered at $y$, and sum those probabilities only afterwards. In this upper application, $T^S_{y,j}$ counts all completed level-$j$ traversals before $\sigma_y$; in the lower application it counts traversals within the first $m_0$ source intervals defined above. Replacing $N$ by $3N$ changes every logarithmic index by $O(1)$. Use Rosen's formal horizon $J$, its barrier $\rho_Jj+z+(j\wedge(J-j))^{1/4}$, and retain only interfaces through $L$. Let $\mathcal{U}_{J,s}$ be the union of these centered lattice upper first-crossing events, with source shell at least $s$, and let $\mathcal{G}^{S,J}_{y,L,z}$ be the corresponding no-crossing event through $L-1$, and put \[ B_L^J=\rho_JL+z+(L\wedge(J-L))^{1/4}. \] On the detailed range $\|y\|>M$, its exact source index is \begin{equation} u_y=\log \frac{3e^J}{\|y\|},\qquad k_y=\lceil u_y\rceil,\qquad m_y=k_y+1;\label{eq:ct-transfer-39m} \end{equation} centers with $\|y\|\le M$ belong to the separately treated near-start range. Thus $u_y\in(k_y-1,k_y]$ and $m_y-u_y\in[1,2)$. For the lower application, use the horizon-$L$ slope $\rho_L$ and its endpoint buffer. Let $I_y^{S,\circ}$ be the \emph{radial} lattice version of \eqref{eq:ct-external-37}: it retains the traversal barrier and the bounded terminal band but drops every angular screen. Angular screens are used only in the spherical Brownian certificate that supplies the second moment. \begin{lemma}[Transfer of traversal events]\label{lem:buffered-transfer} With the definitions above, \begin{equation} P(\mathcal{U}_{J,s})\le J^{-2s/\log J+o(1)}+O(J^{-A}),\label{eq:ct-transfer-40} \end{equation} \begin{equation} P\big(\mathcal{G}^{S,J}_{y,L,z}, \sqrt{2T^S_{y,L}}\in[a,a+1]\big) \le C(k_y+1)^C(2+B_L^J-a)L^{-2}e^{-a^2/(2L)},\label{eq:ct-transfer-41} \end{equation} uniformly for $k_y\le C_0\log J$ and every fixed central range $\epsilon_0L\le a\le B_L^J+1$. No such assertion is made in the extinction range $a<\epsilon_0L$. Most importantly, if \[ Z_L^S=\sum_{y\in F_L^0}{\bf1}_{I_y^{S,\circ}}, \] then \begin{equation} EZ_L^S\le C,\qquad P(Z_L^S>0)\ge c.\label{eq:ct-transfer-42} \end{equation} Only the Brownian proof of \eqref{eq:ct-transfer-42} uses the two-center estimate \eqref{eq:ct-external-36}. No two-center lattice estimate is claimed or used. \end{lemma} \begin{proof}[Proof of Lemma~\ref{lem:buffered-transfer}] There are two logically separate comparisons. Rare upper and endpoint events are compared directly, word by word, using Lemma \ref{lem:smooth-interface-discretization}. The global strong coupling is used only in the safe direction ``strict Brownian lower root implies loose lattice lower root.'' At no point is a Brownian bad-subkernel estimate multiplied by an arbitrary future functional of the coupled lattice path. Throughout, $L$ is the last retained logarithmic level. The upper barrier has formal horizon $J$ and carries a superscript $J$; the lower barrier has horizon $L$. \smallskip\noindent \underline{\textit{Step 1.}}\quad Geometry and one global coupling.\nopagebreak[4] The conformal image retains the original spherical excursion structure. Stereographic projection is conformal and sends spherical Brownian motion to planar Brownian motion under a continuous increasing time change. Moreover, by the definition of $\chi_y$, \[ \Psi_J\bigl(\partial B_d(y,\vartheta_j)\bigr) =\Gamma_{y,j}. \] Thus hitting orders, the complete nested excursion tree, all traversal counts, and all $\chi_y$-angular marks are literally Rosen's spherical objects. In particular, \eqref{eq:ct-external-36} will be applied only before any lattice comparison. For the strong approximation we use Zaitsev, Estimates for the Strong Approximation in Multidimensional Central Limit Theorem, \cite{Zaitsev2003}, Theorem 2 and Corollary 1. The form needed here is: for fixed dimension $d$, centered independent vectors with identity covariance whose laws lie in the analytic finite-exponential-moment class $\mathcal{A}_d(\tau)$ for a fixed $\tau\ge1$, one can couple them with independent standard Gaussian vectors so that, for $\Delta_N=\max_{k\le N}\lVert\sum_{i\le k}(X_i-Y_i)\rVert$, \[ E\exp\{c\Delta_N/(\tau d^{7/2}\log^*d)\} \le \exp\{C d^{9/4+\alpha}\log^*(N/\tau^2)\}, \] and hence \begin{equation} P(\Delta_N>C\log N+x)\le C e^{-cx},\qquad x\ge0,\label{eq:ct-transfer-30} \end{equation} for a fixed i.i.d. law. The planar walk increment multiplied by $\sqrt2$ has identity covariance and bounded support, so the hypotheses apply. Take the deterministic time cap \[ T_J=e^{2J}J^{20}. \] Lemma \ref{lem:elementary-walk-estimates} and its Brownian analogue show that the walk or Brownian motion remains inside the outer disk beyond $T_J$ with probability $e^{-cJ^{10}}$. By \eqref{eq:ct-transfer-30}, after changing Brownian time by the covariance factor $1/2$, there is a coupling for which \begin{equation} \sup_{t\le T_J}\lVert S^{\rm lin}_{2t}-B_t\rVert \le b_J:=CJ^8\label{eq:ct-transfer-31} \end{equation} except on an event of probability $e^{-cJ^8}$. Brownian interpolation between integer times has a smaller Gaussian error. The crossing convention is an intrinsic deterministic function of the finite lattice path. For a curve with bounded component $D$, a lattice crossing is recorded exactly when $\mathbf1_{\{S_{t-1}\in D\}}\ne\mathbf1_{\{S_t\in D\}}$; a curve vertex is assigned to the exterior, and same-side tangencies or chord contacts are ignored. The crossing angle is the normal projection of the post-crossing endpoint. At all used scales the endpoint lies in the uniform tubular collar, so this projection is unique. On the compact family of centers under consideration, the maps $\chi_y\circ\Psi_J^{-1}$ and their inverses have uniformly bounded scaled first two derivatives. Consequently, if two physical paths are within $b$, then every crossing of a nominal interface by one path is sandwiched between the crossings by the other path of the two interfaces obtained by changing its $\log|\chi_y|$-level by $Cb/R$, where $R$ is the physical scale. This is just the deterministic inclusion of the two $b$-neighborhoods of a smooth separating curve; applying it successively preserves chronological order. The crossed directed edge retains the digital-boundary overshoot, which costs at most two extra lattice units. Thus the only difference between the nominal Brownian word and the lattice word comes from collars of relative logarithmic width \begin{equation} \delta_J=C\frac{b_J+1}{M}=O\left(\frac{J^8}{M}\right).\label{eq:ct-transfer-33} \end{equation} \smallskip\noindent \underline{\textit{Step 2.}}\quad Preserving the lower radial word.\nopagebreak[4] The target is the pathwise inclusion \eqref{eq:ct-transfer-R4b}. It suffices to exclude a set $\mathcal D_y$ of collar failures whose conditional probability is small for each complete capped radial word. Outside $\mathcal D_y$, collar closeness must give the same ordered traversal forest, and hence the same counts $T_{y,j}^{1,m_0}$, for the two paths. The three guards below prevent a wrong neighboring hit, a retreat before the intended hit, and a reversal just after it, respectively. Uniform path closeness alone does not preserve an excursion tree: a path may enter a narrow collar, retreat, and thereby insert or delete a traversal. We exclude precisely these retreats and wrong-side approaches. Their total conditional probability will be small for every prescribed radial word. Work in the centered cylinder coordinate \[ U_t=-\log\bigl(|\chi_y(X_t)|/r_0\bigr), \] after its conformal time change, so that $U$ is one-dimensional Brownian motion and $\Gamma_{y,j}=\{U=j\}$. For a stopping time $s$, write $H_a[s]=\inf\{t\ge s:U_t=a\}$, with $\inf\varnothing=\infty$. Put \begin{equation} \eta_J=8\delta_J,\qquad N_0=J^4,\label{eq:ct-transfer-R1} \end{equation} and take $J$ large enough that $0<\delta_J<\eta_J/4$ and $\eta_J\le1/4$. \smallskip\noindent\emph{The truncated word.} Here is the finite depth-$L$ word belonging to \eqref{eq:ct-external-32b}. From $S_{i,0}=A_{y,i}$, $K_{i,0}=1$, recursively set, when $K_{i,r}S_{i,r}:U_t\in \{K_{i,r}-1,K_{i,r}+1\}\},\qquad K_{i,r+1}=U_{S_{i,r+1}}. \] When $K_{i,r}=L$, instead set \begin{equation} S_{i,r+1}=H_{L-1}[S_{i,r}],\qquad K_{i,r+1}=L-1;\label{eq:ct-transfer-R1a} \end{equation} this is a forced terminal return, and the motion below $\Gamma_{y,L}$ is deliberately not resolved. Stop at $N_i=\inf\{r\ge1:K_{i,r}=0\}$, so $B_{y,i}=S_{i,N_i}$. For $iB_{y,i}:U_t\in\{-1,1\}\}. \] On $S_y$, $U_{C_i}=1$, $C_i=A_{y,i+1}$, and the connector is the single prescribed sign $0\to1$. Concatenating the $m_0$ block words and these $m_0-1$ connector signs gives exactly $\overline W_y$ in \eqref{eq:ct-transfer-39hprime}. A block letter $j-1\to j$ pushes the next Ulam--Harris child, a letter $j\to j-1$ pops it, and \eqref{eq:ct-transfer-R1a} closes a terminal leaf. Hence \begin{equation} T_{y,1}^{1,m_0}=m_0,\qquad T_{y,j}^{1,m_0}=\#\{j-1\to j\hbox{ block letters}\}, \quad2\le j\le L.\label{eq:ct-transfer-R1b} \end{equation} All these times are recursively defined hitting times of closed sets; the completed $m_0$-block word and the cap $|\overline W_y|\le N_0$ are therefore stopping-line measurable. \smallskip\noindent\emph{The collar conditions.} Consider first a competitive letter beginning at a stopping time $s$, at level $ks:U_u\in\{k-1,k+1\}\},\qquad U_t=k+\epsilon. \] Define \[ \begin{array}{lll} p=H_{k+\epsilon(1-\eta_J)}[s], &w=H_{k-\epsilon(1-\eta_J)}[s], &r=H_{k+\epsilon\eta_J}[p],\\ e=H_{k+\epsilon(1+\eta_J)}[t], &o=H_{k+\epsilon\eta_J}[t]. \end{array} \] The bad events for this letter are \begin{equation} D^{\rm wrong}=\{w0. \] Thus $V$'s next nominal neighboring line is exactly $k+\epsilon$, and it has made no intervening push or pop. The forced return \eqref{eq:ct-transfer-R1a} uses the retreat and post guards only; deeper visits are ignored by both truncated words. At $A_{y,1}$, \eqref{eq:ct-transfer-R2a} first prevents a premature completed neighboring letter and then forces $V$ across the source line. The preceding letter argument initializes every later $A_{y,i}$ through its connector. Induction in chronological order on the active Ulam--Harris stack now gives the same push/pop operation at every symbol. At $B_{y,i}$ the same stack empties; for $iM$, define from the actual starting radius \begin{equation} u_y=\log \frac{R_0}{\|y\|},\qquad k_y=\lceil u_y\rceil,\qquad m=k_y+1.\label{eq:ct-transfer-45a0} \end{equation} Then $u_y\in(k_y-1,k_y]$, $m-u_y\in[1,2)$, and the deterministic start lies outside $D(y,R_{m-1})$. This index differs by only $O(1)$ from $\lfloor\log(e^J/\|y\|)\rfloor$, so it changes no shell exponent. Write \begin{equation} \begin{split} E_j^{\rm in}&=\{(a,b):a\notin D(y,R_j),\ b\in D(y,R_j),\ a\sim b\},\\ E_j^{\rm out}&=\{(a,b):a\in D(y,R_j),\ b\notin D(y,R_j),\ a\sim b\}. \end{split}\label{eq:ct-transfer-45a1} \end{equation} Every radial query retains its phase, level, Ulam--Harris stack and child counters, population vector, and current directed crossing edge. Potential-kernel optional stopping, uniformly in that edge, gives for $kR_l^+$, so every buffered annulus below is nonempty. Starting with the first entrance into $D(\widehat y,R_k^+)$, and stopping on exiting $D(\widehat y,R_0^+)$, let $\widetilde T_{\widehat y,l}^{S,k\to0}$ count completed crossings from $\partial D(\widehat y,R_{l-1}^-)$ to $\partial D(\widehat y,R_l^+)$. The inclusions \begin{equation} \begin{split} D(y,R_k)&\subset D(\widehat y,R_k^+),\\ D(y,R_l)&\subset D(\widehat y,R_l^+),\\ D(\widehat y,R_{l-1}^-)&\subset D(y,R_{l-1}),\\ D(y,R_0)&\subset D(\widehat y,R_0^+) \end{split}\label{eq:ct-transfer-45d0} \end{equation} show pathwise that \begin{equation} T_{y,l}^{S,k\to0} \le\widetilde T_{\widehat y,l}^{S,k\to0}.\label{eq:ct-transfer-45d1} \end{equation} Indeed every $y$-crossing begins outside the representative's buffered outer circle, ends inside its buffered inner circle, and occurs after the representative start and before its stop. The same inclusions cover $k=l-1$. Area comparison for the mesh in a shell of diameter $O(R_k)$ gives, for every $k\le l-1$, \begin{equation} \#\{\hbox{representatives at }(k,l)\} \le C e^{2(l-k)}\phi(l)^2.\label{eq:ct-transfer-45d} \end{equation} The buffered inclusions in Rosen (2.26)--(2.33) do not impose $k\le\log J$. For the four modified radii $(R_k^+,R_{l-1}^-,R_l^+,R_0^+)$, the same potential-kernel calculation gives, uniformly in all directed entrance edges, \begin{equation} \widetilde q_{k,l}^S\le C\frac{k+1}{l},\qquad \widetilde p_l^S \le1-\frac{1-C/\phi(l)-C/M}{l}.\label{eq:ct-transfer-45d2} \end{equation} Consequently \eqref{eq:ct-transfer-45b} holds with $T$ replaced by $\widetilde T_{\widehat y,l}^{S,k\to0}$, the additional exponent being $CN/(l\phi(l))+CN/(lM)$. Taking the upper horizon-$J$ barrier level \[ N=\frac12\bigl(\rho_Jl+z+(l\wedge(J-l))^{1/4}\bigr)^2 \] there gives \begin{equation} P(\widetilde T_{\widehat y,l}^{S,k\to0}\ge N) \le C e^{-2l-(l\wedge(J-l))^{1/4}}e^{-2z}.\label{eq:ct-transfer-45e} \end{equation} On the used range $l\le J-(4\log J)^4$, the elementary exponent calculation gives $l/\phi(l)=O(1)$ (the bounded early values are absorbed into $C$), and $N/(lM)=o(1)$. Multiplying \eqref{eq:ct-transfer-45d} and \eqref{eq:ct-transfer-45e} and summing first in $l$, then in $k\ge s$, proves \eqref{eq:ct-transfer-40}, including the range $\log J0\right)\ge c_* .\label{eq:ct-transfer-46} \end{equation} The screens occur only in this Brownian second-moment certificate; the lattice target discards all angular screens. Let $\widehat I_y^B$ be the Brownian event obtained from $I_y^\circ$ by deleting every angular screen and retaining only the radial barrier, the bounded endpoint band, and the fixed $m_0$-source definition. Rosen Appendix Theorem 9.1(a), followed by the same source and barrier calculation as \eqref{eq:ct-external-29}, gives \begin{equation} P(\widehat I_y^B)\le Ce^{-2L}, \qquad \sum_yP(\widehat I_y^B)\le C.\label{eq:ct-transfer-47} \end{equation} Since $I_y^{\rm sf}\cap \mathcal D_y\subseteq \widehat I_y^B\cap S_y\cap \mathcal D_y$, estimate \eqref{eq:ct-transfer-R3} may be summed over complete safe radial words before any angular screen is imposed. Hence \begin{equation} \sum_yP(I_y^{\rm sf}\cap \mathcal D_y) \le\varepsilon_J\sum_yP(\widehat I_y^B\cap S_y) \le\varepsilon_J\sum_yP(\widehat I_y^B)=o(1).\label{eq:ct-transfer-47a} \end{equation} Let $\mathcal{K}_J$ be the single global coupling/time-cap event \eqref{eq:ct-transfer-31}. The pathwise implication \eqref{eq:ct-transfer-R4b} gives \[ \{Z_L^{\rm sf}>0\}\cap\mathcal{K}_J \setminus\bigcup_y(I_y^{\rm sf}\cap \mathcal D_y) \subseteq\{Z_L^S>0\}. \] Consequently, paying the coupling failure only once, \begin{equation} P(Z_L^S>0)\ge P(Z_L^{\rm sf}>0) -\sum_yP(I_y^{\rm sf}\cap \mathcal D_y)-P(\mathcal{K}_J^c) \ge c_*-o(1).\label{eq:ct-transfer-47b} \end{equation} Source safety and \eqref{eq:ct-external-38d} keep every block, connector, and final exposed post-closing collar a fixed positive distance inside the stopping boundary, so the close lattice path is not killed before the transferred blocks finish. No Brownian angular screen is transferred. This is the only use of the two-center estimate \eqref{eq:ct-external-36}, and it occurs entirely on the spherical Brownian side. For the lattice first moment, extend the path if necessary and decompose its first $m_0$ source blocks, treating every connector as an integrated subkernel of row mass at most one. For a fixed collection of block words, apply the edge-marginal induction \eqref{eq:ct-smooth-39iprime} to all competitive letters, uniformly over the entrance law supplied by each preceding connector. Forced terminal returns have row mass one. Thus, if $q$ is the number of competitive decisions, \[ P_S(\hbox{these block words})\le e^{CJ^3/M}2^{-q}. \] Summing the disjoint words obeying the lower barrier and terminal band gives the critical geometric Galton--Watson probability with exactly $m_0$ ancestors. Rosen Appendix Theorem 9.1(a), followed by the fixed-source calculation already used in \eqref{eq:ct-transfer-47}, gives \begin{equation} P(I_y^{S,\circ})\le Ce^{-2L},\qquad EZ_L^S\le C.\label{eq:ct-transfer-47cprime} \end{equation} This proves \eqref{eq:ct-transfer-42}; in particular, the lower first moment does not use the upper source-population estimate \eqref{eq:ct-transfer-45c}. Finally we justify the terminal switch used in the definition of $\mathcal{R}_y$. On the compact center family, with $F_y=\sigma\circ\chi_y^{-1}$, \[ F_y(w)=F_y(0)+a_yw+b_yw^2+O(|w|^3), \qquad00\), after the centered extension \eqref{eq:ct-transfer-39k}. The case \(b=0\) is trivial and is henceforth excluded. The reset construction below assigns every initial/final suffix and connector exactly once, so there is no unrecorded future weight. We separate a subkernel into its row mass and its normalized transition. The row masses will retain the probability of the complete word; the normalized transitions will provide contraction of endpoint observables. Put \[ H(a,e)=P_a\big((S_{\tau_{D_y^c}-1},S_{\tau_{D_y^c}})=e\big). \] Write \(b_g\) for the number of completed pieces in block \(g\), so \(\sum_gb_g=b\). For \(i0\) completed pieces as above. The following estimates compare its endpoint law, concentrate its first two bridge moments, and control its occupation. The exact density of the complete word is \begin{equation} \frac{dP_w}{d\widetilde P_w} =\frac{\displaystyle\prod_{g=1}^{G_0}\prod_{i=1}^{b_g} \sigma_{g,i}(e_{g,i})}{\displaystyle E_{\widetilde P_w} \prod_{g=1}^{G_0}\prod_{i=1}^{b_g} \sigma_{g,i}(e_{g,i})}, \qquad e^{-2\eta_w^*}\le \frac{dP_w}{d\widetilde P_w} \le e^{2\eta_w^*}. \label{eq:tu-48d} \end{equation} The product includes the final no-more-piece suffix and every connector likelihood. The endpoint averages concentrate about their reference expectations: for every \(t>0\), \begin{equation} \begin{split} P_w\Bigg(&\max_{f\in\mathscr{F}_M}\left| b^{-1}\sum_{i\le b}f(\omega_i)-c_f\right|\\ &>t+C\{G_0/b+\varepsilon_M+Q/M\}\Bigg) \le2|\mathscr{F}_M|\exp\{2\eta_w^*-cbt^2\}. \end{split} \label{eq:tu-49} \end{equation} For \(x\in V_y\), \(G=G_{D_y}\), and \begin{equation} r_\theta(x):= \frac{e^\theta-1} {1-(G(x,x)-1)(e^\theta-1)}, \label{eq:tu-39} \end{equation} the logarithmic occupation moment satisfies, uniformly for \(0\le r_\theta\le c/h\), \(h=\log M\), \begin{equation} \log E\left[\exp\left\{\theta\sum_{i=1}^bX_i(x)\right\} \,\middle|\,w\right] =bg\,r_\theta+O\left(br_\theta^2+ r_\theta\{G_0+b\varepsilon_M+bQ/M\} +\eta_w^*\right). \label{eq:tu-50} \end{equation} The leading term is the number of pieces times the reference mean \(g\), expressed in the tilt parameter \(r_\theta\). The errors record the quadratic expansion, the initial laws of the \(G_0\) blocks, the Green and return-word comparisons, and the full-word likelihood. The estimate is uniform in \(x,w,b\) and concerns completed pieces. In both applications below every visit to \(V_y\) lies in a completed piece, as verified at the end of the proof. For the upper-bound choice from Lemma~\ref{lem:buffered-transfer}, \[ h=(C_h\log J)^4,\qquad L=J-h+O(1),\qquad U=u_{J,c},\quad c>2, \] put \(Y=\sqrt{2b}\). On a radial word satisfying the coefficient-two coarse barrier below, the upper occupation tail satisfies \begin{equation} \begin{split} P\left(\max_{x\in V_y}\sum_iX_i(x)\ge U\mid w,b\right) &\le C e^{-2L}J^c e^{b/L}\\ &\quad{}\times \exp\left\{-\kappa\min\left( \frac{(p-p_*)^2}{h},\frac{J^2}{h}\right)\right\}, \end{split} \label{eq:tu-51} \end{equation} in the central endpoint range \(\epsilon L\le Y\le B_L+1\), for one fixed small \(\epsilon>0\). Here \[ p=B_L+1-Y,\qquad B_L=\rho_JL+z_0+h^{1/4},\qquad p_*=O(\log J). \] The gap \(p\) measures the distance of the endpoint below the barrier, and \(p_*\) is the gap at the preferred endpoint of the terminal tilt. The coefficient-two barrier is \begin{equation} \sqrt{2T^S_{y,j}}\le B_j:=\rho_Jj+z_0+(j\wedge(J-j))^{1/4} \quad(k_y0\), \begin{equation} \sup_e\left|\frac{\sigma_{g,i}(e)}{s_{g,i}}-1\right| \le \frac{Cq_i}{M}. \label{eq:tu-58c} \end{equation} This estimate holds for the complete word, since \eqref{eq:tu-TP13} retains the row masses of all its letters. In the common logarithmic conformal coordinate, every normalized Brownian letter is the rotation convolution \eqref{eq:tu-TP5}, and the terminal disk exit is \eqref{eq:tu-TP4}. Apply Lemma~\ref{lem:typed-centered-endpoint-product} to the exact exit-edge averages \eqref{eq:tu-43a}--\eqref{eq:tu-43b}. For consecutive internal transitions \(r\le s\) in one block and every \(f\in\mathscr{F}_M\), it gives \begin{equation} \left|\Lambda_M \widetilde K_{g,r}^w\widetilde K_{g,r+1}^w\cdots \widetilde K_{g,s}^wf-c_f\right| \le C\left\{\varepsilon_M+ \frac{1}{M}\sum_{i=r}^s(q_i+1)\right\}. \label{eq:tu-58e} \end{equation} The two terms are compared separately with the same Haar integral of the disk-Green traces \eqref{eq:tu-TP6}--\eqref{eq:tu-TP9}. The estimate therefore controls complete prefixes even though \(\Lambda_M\) need not be invariant for a single discrete transition. In a lower source interval the common coordinate is \(\chi_y\); its terminal chart error is included in the displayed sum. An inter-source connector ends this comparison. To remove dependence on the initial law of each block, we also need a uniform contraction in total variation. The elliptic Harnack inequality \cite[Theorem~1.1]{AthreyaEtAl2022} states that for every \(d\ge1\) there is \(C_d\) such that a nonnegative function harmonic in \(B(x_0,R)\) satisfies \begin{equation} \max_{B(x_0,R/2)}u\le C_d\min_{B(x_0,R/2)}u. \label{eq:tu-58a} \end{equation} Starting one step inside the outer boundary is too close to use \eqref{eq:tu-58a} directly. Insert instead the deterministic intermediate circle \[ C_M=\partial D(y,e^{1/2}M), \] which every successful outer-to-inner crossing must hit. For a target atom \(a\in A_y\), put \[ h_a(z)=P_z(S_{H_{A_y}}=a,\ H_{A_y}<\tau_{D(y,eM)}), \qquad s(z)=\sum_{a\in A_y}h_a(z),\qquad z\in C_M. \] A bounded chain of lattice balls of radius \(cM\), all a fixed positive fraction from both boundary components, applies \eqref{eq:tu-58a} to \(h_a\) and \(s\). Uniformly in \(a,z,z_0\in C_M\), \[ ch_a(z_0)\le h_a(z)\le Ch_a(z_0),\qquad cs(z_0)\le s(z)\le Cs(z_0). \] After division by the total successful-leg probability this gives \begin{equation} J_{\rm in}(z,\cdot)\ge\alpha J_{\rm in}(z_0,\cdot) \label{eq:tu-58b} \end{equation} where \(J_{\rm in}(z,\cdot)=h_{\cdot}(z)/s(z)\) is the entrance law conditional on reaching the inner boundary, and \(\alpha>0\) is uniform in \(M,z,z_0\). Precomposition by all earlier within-block word factors and postcomposition by \(H\) preserve this row-independent minorization. Writing \(\operatorname{osc}(f)=\sup f-\inf f\), we obtain \(\operatorname {osc}(\widetilde K_{g,i}^wf)\le(1-\alpha) \operatorname {osc}(f)\) for every bounded \(f\) at every within-block transition. Write \(\rho=1-\alpha<1\). A reset transition is used exactly and has Dobrushin coefficient at most one; there are only \(G_0-1\) of them. \smallskip\noindent \underline{\textit{Step 3.}}\quad Retain the full likelihood at resets. A connector can have an arbitrary normalized angular kernel. What is needed is a bound on the oscillation of its row mass, uniform even when the prescribed future event is rare. To obtain it, factor each reset through a deterministic separator. Fix \begin{equation} \lambda_*=e^{3/2}\in(e,e^2),\qquad \Sigma_y=\partial D(y,\lambda_*M). \label{eq:tu-59a0} \end{equation} The next active outer interface after the terminal \(eM\)-circle has scale \(e^2M\). The fixed-gap geometry and \eqref{eq:tu-47c} put \(\Sigma_y\) strictly between them and keep \(D(y,\lambda_*M)\) inside the applicable source interval and global killing domain. Let \(e=(u,v)\in E_y\) be the last exit edge of a nonempty block, so the exterior kernel starts at \(v=\operatorname{out}(e)\) and \(|v-y|=eM+O(1)\). Put \[ \theta_-=H_{D(y,M)},\qquad \theta_+=\tau_{D(y,\lambda_*M)}, \] and on \(\{\theta_+<\theta_-\}\) let \(F_+\) be the directed outward edge at \(\theta_+\). For \(gbt\right) \le2e^{-cbt^2}. \label{eq:tu-61} \end{equation} To center this concentration bound at \(bc_f\), let \(\mu\) be the law of the first endpoint of a block, and consider its \(k\)-th endpoint. Comparing the same transition product started from \(\mu\) and from \(\Lambda_M\) costs \(2\rho^{k-1}\) by contraction. The reference product is controlled by \eqref{eq:tu-58e}, so \begin{equation} \begin{split} \left|\mu\widetilde K_{g,1}^w\cdots \widetilde K_{g,k-1}^wf-c_f\right| &\le2\rho^{k-1}\\ &\quad+C\left\{\varepsilon_M+ \frac{1}{M}\sum_{j0\). If \(\epsilon L\le Y\le B_L+1\) and \(|Y-Y_*|\le\delta J\), Taylor's formula, \eqref{eq:tu-51a}, and the displacement \[ s=\frac{F_b'(r_0)}{C_1J^2h} \] (with \(C_1\) enlarged once) give \begin{equation} \sup_rF_b(r)-F_b(r_0) \ge\kappa\frac{(Y-Y_*)^2}{h}. \label{eq:tu-51d} \end{equation} This displacement is feasible. If it is negative, the barrier and \(h^{1/4}=O(\log J)\) imply \(Y-Y_*=O(\log J)\), whence \(|s|=O(\log J/(Jh))=o(r_0)\). If it is positive, choosing \(\delta\) small ensures \(s0\), with \(a_0\) small compared with \(\delta/C_1\). Equations \eqref{eq:tu-51a}--\eqref{eq:tu-51c} then give \begin{equation} F_b(r_0+s)-F_b(r_0)\ge\kappa J^2/h. \label{eq:tu-51e} \end{equation} There is no range \(Y\ge Y_*+\delta J\) because the upper barrier permits only an \(O(\log J)\) overshoot above \(Y_*\). Finally, if \(Y<\epsilon L\), take \(r=a_0/h\) directly. The integral formula \(\theta(r)=\int_0^r\theta'(t)\,dt\) gives \(\theta(a_0/h)\ge c_aa_0/h\). First choosing \(\epsilon\) and then \(a_0\) small yields \begin{equation} F_b(a_0/h)\ge\kappa_0J^2/h, \qquad F_b(r_0)=O(J)=o(J^2/h). \label{eq:tu-51f} \end{equation} Equations \eqref{eq:tu-51d}--\eqref{eq:tu-51f} supply the gain beyond the first tilt. Since \(p-(B_L+1-Y_*)=Y_*-Y\), this is the decay claimed in \eqref{eq:tu-51}. The expansion \eqref{eq:tu-51-2} is uniform in \(x\), so a common choice \[ p_*=B_L+1-2L+\frac c2\log J \] differs from each site's preferred gap by \(O(1)\). The inequality \((u-v)^2\ge u^2/2-v^2\) absorbs this bounded shift by changing \(C,\kappa\). Thus the union over sites retains the same decay factor, proving both endpoint alternatives in \eqref{eq:tu-51}. \smallskip\noindent \underline{\textit{Step 7.}}\quad Account for the initial and final portions. In the applications, all core occupation belongs to completed pieces. A lower source block begins and ends on interfaces outside \(D_y\). An upper block with \(k\le C_0\log J\) starts at the origin, which is outside \(D_y\), and ends at \(\sigma_y\), also outside \(D_y\). Before the first inward crossing of radius \(M\) the path cannot visit \(V_y\). After any visit to \(V_y\), reaching the block's outer endpoint forces a later crossing of radius \(eM\), so that inward crossing and all intervening core occupation form a completed piece. Thus the initial and final incomplete portions contribute exactly zero to every \(X(x)\) used here. Upper cells with \(k>C_0\log J\), including the cells near the walk's starting point, are treated separately by the one-site Green estimate in \eqref{eq:tu-62-2}. \end{proof} \subsection{A local upper bound for the maximum} \begin{proposition}\label{prop:localized-disk-upper} For every fixed \(\beta>0,c>2\) and \(A<\infty\), \begin{equation} \begin{split} P\left(\max_{\lVert x\rVert\le e^JJ^{-\beta}} L_{\tau_{e^J}}(x)\ge u_{J,c}\right) \le J^{-2\beta+c-2+o(1)}+O(J^{-A}). \end{split} \label{eq:tu-53} \end{equation} The \(o(1)\) is uniform when \(\beta,c\) range over compact subsets of \((0,\infty)\) and \((2,\infty)\), respectively. \end{proposition} \begin{proof}[Proof of Proposition~\ref{prop:localized-disk-upper}] We work with one terminal cell at a time and extend the walk to a disk centered at that cell. This puts the source law, traversal barrier, and terminal word in the same centered geometry. The traversal estimate and the terminal exponential moment then have opposite Gaussian factors; after their cancellation, the ballot prefactor supplies the required power \(J^{-2}\). \smallskip\noindent \underline{\textit{Step 1.}}\quad Choose cells and exclude barrier crossings. Use the upper buffer \[ h=(C_h\log J)^4,\qquad M=e^h,\qquad L=J-h, \] with \(C_h\) fixed large. Cover the target disk by original \(M\)-cells obtained from the dilated image of Rosen's predetermined net \(F_L\), rounding each image center to its nearest lattice point. The chart has uniformly bounded distortion on the relevant compact patch, so these centers are \(cM\)-separated and form a \(CM\)-cover. After altering the fixed core and terminal radius factors, every lattice point lies in a core, the overlap multiplicity is bounded, and Lemmas \ref{lem:buffered-transfer} and \ref{lem:terminal-feynman-kac} apply to the same cells. We continue to denote their radii by \(M/2\) and \(eM\). For a cell center with \(\|y\|>M\), use the exact index from \eqref{eq:ct-transfer-45a0}, \begin{equation} u_y=\log \frac{3e^J}{\|y\|},\qquad k=k(y)=\lceil u_y\rceil,\qquad m=k+1. \label{eq:tu-54a} \end{equation} Cells with \(\|y\|\le M\) are assigned to the final sitewise range \(k>C_0\log J\). Uniform distortion of the fixed chart makes \eqref{eq:tu-54a} equal to the spherical source index \eqref{eq:ct-external-24} up to a bounded additive error, which is absorbed by the barrier slack. Every target cell has \begin{equation} k\ge s:=\beta\log J+O(1). \label{eq:tu-54} \end{equation} For this fixed cell continue the walk from \(\tau_{e^J}\) to \(\sigma_y=\tau_{D(y,3e^J)}\). By \eqref{eq:ct-transfer-39l}, a high local time in the cell before \(\tau_{e^J}\) is also high before \(\sigma_y\). All counts, words, terminal bridges, and the source index below refer to this single \(y\)-centered extended path. This domination is applied separately to each cell before the final union bound. First exclude a crossing of the fixed positive-intercept coefficient-two barrier \eqref{eq:tu-52} at or before level \(L\). Since \(h>(4\log J)^4\) after choosing \(C_h\) large, all these levels lie in the range of \eqref{eq:ct-transfer-45d}--\eqref{eq:ct-transfer-45e}. Restricting the source shells to \(k\ge s\) and using \eqref{eq:ct-transfer-40} gives \begin{equation} \begin{split} P(\hbox{some bad prefix in the target}) &\le C\sum_{k\ge s}\sum_{l=k+1}^{L} e^{2(l-k)+\frac12(l\wedge(J-l))^{1/4}} e^{-2l-(l\wedge(J-l))^{1/4}}\\ &\le J^{-2\beta+o(1)}. \end{split} \label{eq:tu-55} \end{equation} The direct lattice estimates \eqref{eq:ct-transfer-45a}--\eqref{eq:ct-transfer-45e}, for both the representative count and the probability of a crossing at one center, apply through the source-shell range \(k\le C_0\log J\). \smallskip\noindent \underline{\textit{Step 2.}}\quad Use the source-inclusive endpoint estimate. Fix an original cell with \(s\le k\le C_0\log J\), where its source index is \(k=k(y)\) in \eqref{eq:tu-54a}. The extension to \(\sigma_y\) makes all its radial interfaces centered at \(y\), as required by Lemma~\ref{lem:buffered-transfer}; our choice of \(h\) satisfies \(h/\log J\to\infty\) and \(h=o(J)\). On the complement of the crossing event in Step 1, its prefix satisfies the barrier \eqref{eq:tu-52} and hence the no-crossing event in \eqref{eq:ct-transfer-41}. Put \(Y=\sqrt{2T^S_{y,L}}\), and use the same small cutoff \(\epsilon_0=\epsilon\) as in Lemma~\ref{lem:terminal-feynman-kac}. Take \(\epsilon_0L\le q\le B_L+1\). With \(p=B_L+1-q\), the bound \eqref{eq:ct-transfer-41} gives \begin{equation} P(\hbox{prefix good},\,Y\in I_q) \le C(k+1)^C(1+p)L^{-2}\exp\{-q^2/(2L)\}. \label{eq:tu-58} \end{equation} Here \(I_q=[q,q+1]\), and \(B_L=B_L^J\) in that lemma because \(L=J-h\). The estimate already includes the source population and its last death decision; no conditioning on a source height is needed. \smallskip\noindent \underline{\textit{Step 3.}}\quad Combine the traversal and terminal costs. The complementary endpoint range is controlled directly by the terminal tilt. If \(Y<\epsilon_0L\), then \(b\le\epsilon_0^2L^2/2+O(L)\), and the saturated line of \eqref{eq:tu-51}, uniformly in the radial word, gives \begin{equation} P(\hbox{terminal mark}\mid w) \le C e^{-2L}J^c \exp\{\epsilon_0^2L/2-\kappa J^2/h+O(1)\}. \label{eq:tu-58a-2} \end{equation} After summing this over at most \(Ce^{2L}\) original cells, the result is \(o(J^{-A})\) for every fixed \(A\), because \(h=o(J)\), after fixing \(\epsilon_0\) sufficiently small. This is why neither \eqref{eq:ct-external-29} nor \eqref{eq:ct-transfer-41} was claimed in the extinction range. For the number \(b=T_L+O(1)\) of completed outward pieces in the band \(I_q\), \[ b=q^2/2+O(L). \] Since $q\ge\epsilon_0L$, uniformly in this band the actual terminal gap satisfies $B_L+1-\sqrt{2b}=p+O(1)$. The bounded shift is absorbed by the same quadratic inequality used to choose a common $p_*$ above. Conditional on the radial word, Lemma~\ref{lem:terminal-feynman-kac} applies; the start/end observation at the end of that lemma shows that all visits to \(V_y\) belong to its completed pieces. Multiply \eqref{eq:tu-58} by \eqref{eq:tu-51}. Since \(b=q^2/2+O(L)\), the combined endpoint cost is \[ \exp\left\{-\frac{q^2}{2L}+\frac bL\right\}=e^{O(1)}. \] Thus the cancellation is uniform throughout the central range. The remaining ballot factor \(1+p\) is summable because of the additional terminal decay: \begin{equation} \sum_{0\le p\le B_L+1}(1+p) \exp\left\{-\kappa\min\left( \frac{(p-p_*)^2}{h},\frac{J^2}{h}\right)\right\} =J^{o(1)}. \label{eq:tu-59} \end{equation} Indeed the nonsaturated Gaussian sum is \[ O\bigl((1+|p_*|)\sqrt h+h\bigr)=J^{o(1)}, \] while its saturated complement contributes \(O(J^2)e^{-\kappa J^2/h}=o(1)\). The small-\(Y\) part has the saturated factor in \eqref{eq:tu-51}. Thus, uniformly over all endpoint bands and all radial words, \begin{equation} P(\hbox{this original cell is good and marked}) \le (k+1)^Ce^{-2L}J^{c-2+o(1)}. \label{eq:tu-60} \end{equation} The terminal maximum has already included all sites of the original cell, so the remaining union bound is over the original cells. There are \(O(e^{2(L-k)})\) original cells in shell \(k\). Summing \eqref{eq:tu-60} and using \eqref{eq:tu-54} yields \begin{equation} \sum_{k=s}^{C_0\log J}e^{2(L-k)} (k+1)^Ce^{-2L}J^{c-2+o(1)} \le J^{-2\beta+c-2+o(1)}. \label{eq:tu-61-2} \end{equation} \smallskip\noindent \underline{\textit{Step 4.}}\quad Treat cells near the starting point. It remains to treat \(k>C_0\log J\). For a lattice site in shell \(k\), Green-function identities and Lemma~\ref{lem:elementary-walk-estimates} give \[ P_0(H_x<\sigma_y)\le C(k+1)/J \] and, conditionally on the hit, \[ P_x(L_{\sigma_y}(x)\ge u_{J,c}) \le Ce^{-2J}J^c. \] There are \(O(e^{2J-2k})\) sites in that shell. Their total contribution is \begin{equation} J^{c-1-2C_0+o(1)}. \label{eq:tu-62-2} \end{equation} Choose \(C_0\) after \(A,c\) to make \eqref{eq:tu-62-2} \(O(J^{-A})\). The barrier-crossing bound \eqref{eq:tu-55} is absorbed by \eqref{eq:tu-61-2} because \(c>2\). These two estimates and the sitewise remainder therefore give \eqref{eq:tu-53}. The word-cap and digital row errors are superpolynomial by Lemma~\ref{lem:buffered-transfer} and are included in \(O(J^{-A})\). \end{proof} % ---- terminal_lower ---- \section{The terminal maximum and the lower bound} \label{sec:terminal-lower} The coarse construction stops at depth $L=J-h$ and produces a cell with $b=2L^2-2L\log L+O(L)$ completed terminal pieces. Their mean occupation, about $gb$, misses a contribution of order $Jh$ from the last $h$ logarithmic scales. Maximizing over the terminal disk recovers this contribution: after conditioning on the entrance and exit pairs, the pieces are independent, and their normalized centered occupation field has covariance close to $2g$ times the killed Green function. A Gaussian approximation and the discrete Gaussian free field maximum then give a gain of $2\sqrt2\,g\sqrt b\,h$, which restores the missing depth. We prove this estimate for one fixed cell before combining it with coarse-root existence. Uniform conditional failure bounds allow us to sum over the candidates and obtain a disk maximum with positive probability. Independent walk segments then yield the almost-sure lower bound at deterministic times. \subsection{Conditional occupation in one terminal disk} Recall that $D_y=D(y,eM)$ is the killing disk, $V_y=D(y,M/2)$ is its core, and $X_i(x)$ is the occupation of $x$ by the $i$th completed bridge. Write $G=G_{D_y}$ and retain the conditional moments from Section~\ref{sec:terminal-upper}: \[ m_x(\omega)=E^\omega X(x),\qquad q_{x,x'}(\omega)=E^\omega[X(x)X(x')],\qquad \omega=(a,e). \] Here $a$ is the entrance vertex, $e$ is the directed exit edge, and $\Lambda_M(a,e)=\lambda_A(a)H(a,e)$ is the reference endpoint law. As before, $\Lambda_M f=\sum_{a,e}\Lambda_M(a,e)f(a,e)$ denotes its average. We need the empirical means and second moments to be close to these reference averages for every site and pair of sites. The terminal scale below is chosen so that this simultaneous control and the final error both fit within the $J\log J$ margin. \begin{lemma}[The terminal maximum] \label{lem:terminal-maximum-lower} Fix $1<\xi<2$ and put \[ h=(\log J)^\xi,\qquad M=e^h,\qquad L=J-h. \] Suppose an admissible centered block word has $G_0\le m_0$ blocks and $b$ completed terminal pieces, where \begin{equation} b=2L^2-2L\log L+O(L) \label{eq:terminal-piece-count} \end{equation} with a uniform remainder constant. Put $t_J=\log J/(J\log\log J)$. Call the endpoint realization \emph{moment regular} if, for every $x,x'\in V_y$, \begin{equation} \begin{split} \left|\frac1b\sum_{i\le b}m_x(\omega_i)-\Lambda_M m_x\right| &\le 2t_J,\\ \left|\frac1b\sum_{i\le b}q_{x,x'}(\omega_i) -\Lambda_Mq_{x,x'}\right| &\le 2t_J\bigl(1+G(x,x')\bigr). \end{split} \label{eq:terminal-moment-regularity} \end{equation} Uniformly over these words, \begin{equation} P(\text{moment regularity fails}\mid w) \le CM^4e^{-cbt_J^2}=o(J^{-A}) \label{eq:terminal-moment-failure} \end{equation} for every fixed $A>0$. Condition on the intrinsic sigma-field $\mathcal R_y$ defined in Section~\ref{sec:continuum-transfer}. It records the exterior pieces, the word, and every entrance and directed-exit pair, while erasing the completed bridges and their durations. For every moment-regular realization, under the conditional law given $\mathcal R_y$, \begin{equation} \max_{x\in V_y}\sum_{i\le b}X_i(x) =gb+2\sqrt2\,g\sqrt b\,h +O_P\left(J\sqrt h+\frac{J\log J}{\log\log J}\right). \label{eq:terminal-maximum-expansion} \end{equation} The remainder is uniform over such realizations. In particular, for every fixed $\eta>0$, the conditional probability that this maximum is at least $u_{J,2+\eta}$ tends to one uniformly. \end{lemma} \begin{proof} All probabilities, moments, and covariances of bridge interiors in this proof are conditional on $\mathcal R_y$. We will show that the normalized centered field has maximum $2\sqrt2\,gh+O_P(\sqrt h)$. Multiplication by $\sqrt b$ and restoration of the means then give \eqref{eq:terminal-maximum-expansion}; the final calculation checks that its deterministic part reaches the required level. \smallskip\noindent \underline{\textit{Step 1.}}\quad Conditional moments and their common center. Fix the entrance and directed-exit pairs $(a_i,e_i)_{i\le b}$. The erased occupation vectors are independent killed bridges under the conditional law. With $H(a,e)$ denoting the exit-edge probability from $a$, the one-visit and two-visit decompositions give \begin{equation} E^{a,e}X(x)=\frac{G(a,x)H(x,e)}{H(a,e)} \label{eq:terminal-exact-mean} \end{equation} and \begin{equation} \begin{split} E_a[X(x)X(x');E=e] &=G(a,x)G(x,x')H(x',e)\\ &\quad+G(a,x')G(x',x)H(x,e) -\mathbf1_{\{x=x'\}}G(a,x)H(x,e). \end{split} \label{eq:terminal-exact-second-moment} \end{equation} The subtraction accounts for the visit counted in both orders when $x=x'$. Dividing \eqref{eq:terminal-exact-second-moment} by $H(a,e)$ gives $q_{x,x'}(a,e)$. To apply the endpoint concentration estimate, first normalize the two moment functions. The fixed-ratio Harnack estimates yield \begin{equation} \|m_x/C\|_\infty\le1,\qquad \left\|\frac{q_{x,x'}}{C(1+G(x,x'))}\right\|_\infty\le1 \label{eq:terminal-normalized-moments} \end{equation} uniformly in $x,x'\in V_y$. These are exactly the functions in the finite Green class \eqref{eq:tu-47i}, whose exit-edge averages are \eqref{eq:tu-43a}--\eqref{eq:tu-43b}. Thus the word concentration estimate \eqref{eq:tu-49} applies to at most $CM^4$ functions. Its deterministic bias and likelihood correction are small enough: admissibility gives $G_0\le m_0$ and $Q\le J^3$, while the choice $h=(\log J)^\xi$ with $\xi>1$ makes $M=e^h$ grow faster than every power of $J$. Hence \begin{equation} G_0/b+\varepsilon_M+Q/M=o(t_J),\qquad \eta_w^*=O(1). \label{eq:terminal-screen-errors} \end{equation} Apply \eqref{eq:tu-49} with $t=t_J/C_1$, where $C_1$ is a sufficiently large fixed multiple of the normalization constant in \eqref{eq:tu-47i}. After restoring the normalizations in \eqref{eq:terminal-normalized-moments}, the tolerances are those in \eqref{eq:terminal-moment-regularity}. Thus \begin{equation} \begin{split} P(\text{\eqref{eq:terminal-moment-regularity} fails}\mid w) &\le CM^4\exp\left\{-c\frac{\log^2J}{(\log\log J)^2}\right\}\\ &=o(J^{-A}) \end{split} \label{eq:terminal-screen-union-bound} \end{equation} for every fixed $A>0$. Indeed, $\log M^4=4h$, whereas $bt_J^2\asymp\log^2J/(\log\log J)^2$; the latter dominates both $h$ and $\log J$ because $\xi<2$. The estimate is uniform over the normalized reset kernels, which are integrated out in the conditional endpoint law. This proves \eqref{eq:terminal-moment-failure}. The smaller terminal scale is needed here because all pairs of sites are screened simultaneously; the upper bound used the one-point exponential estimate \eqref{eq:tu-50} before its spatial union bound. We next identify the reference averages, which determine the mean and covariance of the occupation field. Multiplying each conditional moment by $\Lambda_M(a,e)=\lambda_A(a)H(a,e)$ cancels its denominator $H(a,e)$. Summing over $e$ uses $\sum_eH(x,e)=1$; averaging the remaining entrance factor gives $d_x=\sum_a\lambda_A(a)G(a,x)$ from \eqref{eq:tu-44}. The Green comparison \eqref{eq:tu-45}--\eqref{eq:tu-46} therefore yields \begin{equation} \begin{split} \Lambda_M m_x&=d_x,\\ \Lambda_Mq_{x,x'} &=d_xG(x,x')+d_{x'}G(x',x)-\mathbf1_{\{x=x'\}}d_x,\\ \sup_{x\in V_y}|d_x-g|&\le C\varepsilon_M. \end{split} \label{eq:terminal-common-moments} \end{equation} In particular, a moment-regular realization satisfies \begin{equation} \max_{x\in V_y} \left|\frac1b\sum_{i\le b}E^{\omega_i}X_i(x)-g\right| \le Ct_J+O(\varepsilon_M). \label{eq:terminal-conditional-mean} \end{equation} Center each bridge and normalize the sum: \[ \widetilde X_i(x)=X_i(x)-m_x(\omega_i),\qquad U_x=b^{-1/2}\sum_{i\le b}\widetilde X_i(x). \] Conditional independence gives the covariance explicitly as \[ \operatorname{Cov}(U_x,U_{x'}) =\frac1b\sum_{i\le b} \bigl(q_{x,x'}(\omega_i)-m_x(\omega_i)m_{x'}(\omega_i)\bigr). \] The moment screen replaces the average of $q_{x,x'}$ by its reference value with error $O(t_J(1+h))=o(1)$, since $\sup_{x,x'}G(x,x')\le Ch$. In \eqref{eq:terminal-common-moments}, replacing $d_x,d_{x'}$ by $g$ costs $O(\varepsilon_Mh)=o(1)$, and symmetry gives $gG(x,x')+gG(x',x)=2gG(x,x')$. The diagonal correction and the average of $m_x(\omega_i)m_{x'}(\omega_i)$ are bounded by \eqref{eq:terminal-normalized-moments}. Consequently \begin{equation} \max_{x,x'\in V_y} \left|\operatorname{Cov}(U_x,U_{x'})-2gG_{D_y}(x,x')\right| \le C. \label{eq:terminal-covariance-comparison} \end{equation} Thus the reference averages give a nearly constant mean $g$ and a covariance equal to $2gG$ up to a bounded error. These are the two inputs needed for the Gaussian approximation and comparison. \smallskip\noindent \underline{\textit{Step 2.}}\quad Approximation of the maximum by a Gaussian maximum. The number of coordinates is $P=|V_y|=e^{2h+O(1)}$, whereas there are only $b\asymp J^2$ independent vectors. We therefore need an approximation whose dimension dependence is logarithmic. Let $Y_i=(Y_i(x):x\in V_y)$ be independent centered Gaussian vectors with the respective conditional covariances of $\widetilde X_i$, and set \[ \Gamma_x=b^{-1/2}\sum_{i\le b}Y_i(x). \] Then $\Gamma$ has the conditional covariance of $U$. Chernozhukov--Chetverikov--Kato \cite[Theorem~3.1]{CCK2016}, applied under the conditional bridge law, gives, for every Borel set $A\subset\mathbb R$ and $\delta>0$, \begin{equation} \begin{split} P\bigl(\max_{x\in V_y}U_x\in A\bigr) &\le P\bigl(\max_{x\in V_y}\Gamma_x\in A^{C_7\delta}\bigr)\\ &\quad+\frac{C_8\log^2P}{\delta^3\sqrt b} \{L_b+M_{b,X}(\delta)+M_{b,Y}(\delta)\}. \end{split} \label{eq:terminal-cck} \end{equation} Here $A^r=\{s:\operatorname{dist}(s,A)\le r\}$, the constants are universal, and the error depends on the following third moments: \[ \begin{split} L_b&=\max_{x\in V_y}\frac1b\sum_{i\le b} E^{\omega_i}|\widetilde X_i(x)|^3,\\ M_{b,X}(\delta) &=\frac1b\sum_{i\le b}E^{\omega_i}\left[ \max_{x\in V_y}|\widetilde X_i(x)|^3; \max_{x\in V_y}|\widetilde X_i(x)|> \frac{\delta\sqrt b}{\log P}\right]. \end{split} \] The definition of $M_{b,Y}$ is the same with $Y_i$ in place of $\widetilde X_i$ and expectation under its Gaussian law. We will use $\delta=1$, so it remains to bound these moments and the maximum of $\Gamma$. A bridge reaches a fixed core site with probability of order at most $h^{-1}$; conditional on reaching it, its number of visits has an exponential tail on scale $h$. Retaining both factors makes the third-moment error small enough. The rank-one identity \eqref{eq:tu-42} gives this tail estimate directly. Indeed, at $\theta=c_0/h$, with $c_0>0$ fixed and sufficiently small, \eqref{eq:tu-39} gives $r_\theta\le C/h$, and hence \[ E^\omega(e^{\theta X(x)}-1) =r_\theta m_x(\omega)\le C/h. \] For $s\ge h$, exponential Markov's inequality gives $P^\omega(X(x)>s)\le Ch^{-1}e^{-cs/h}$. To see the same scale for smaller $s$, the generating function in \eqref{eq:tu-42} identifies $X(x)$ as zero with probability $1-m_x(\omega)/G(x,x)$ and, otherwise, as a geometric random variable of mean $G(x,x)$. Since $G(x,x)\asymp h$ on $V_y$, we obtain, uniformly in the endpoint pair, \begin{equation} P^\omega(X(x)>s)\le Ch^{-1}e^{-cs/h},\qquad E^\omega|X(x)-E^\omega X(x)|^3\le Ch^2. \label{eq:terminal-bridge-tails} \end{equation} Here and below $s\ge0$. Integrating the tail proves the moment bound, and therefore $L_b=O(h^2)$. For $\delta=1$, the truncation threshold is \[ T=\frac{\sqrt b}{\log P}\asymp\frac Jh. \] A union bound over $P$ coordinates, followed by integration of \eqref{eq:terminal-bridge-tails}, gives the truncated third moment. To make the truncation explicit, put $V_i=\max_{x\in V_y}|\widetilde X_i(x)|$. Since $P\le Ce^{2h}$ and $T\asymp J/h\to\infty$, \eqref{eq:terminal-bridge-tails} implies $P(V_i>t)\le CP h^{-1}e^{-ct/h}$ for $t\ge T$. Therefore \[ \begin{aligned} E[V_i^3;V_i>T] &=T^3P(V_i>T)+3\int_T^\infty t^2P(V_i>t)\,dt\\ &\le CP h^{-1}(T^3+hT^2+h^2T+h^3)e^{-cT/h}. \end{aligned} \] Here $T/h\asymp J/h^2$ and $\log P=O(h)$, which gives \begin{equation} \frac1b\sum_{i\le b}E^{\omega_i}\left[ \max_{x\in V_y}|\widetilde X_i(x)|^3; \max_{x\in V_y}|\widetilde X_i(x)|>T\right] \le \exp\{-cJ/h^2+Ch\}. \label{eq:terminal-truncated-third-moment} \end{equation} The polynomial factors are absorbed by the negative exponent, since $J/h^2$ dominates $h$ and $\log J$. The matching Gaussian vectors have coordinate variances at most $Ch$, by the second derivative of \eqref{eq:tu-42}. Their truncated third-moment contribution is bounded by $\exp\{-cJ^2/h^3+Ch\}$ after the same union bound. Finally, $\log^2P/\sqrt b=O(h^2/J)$ and $L_b=O(h^2)$, so the error in \eqref{eq:terminal-cck} is at most \begin{equation} O(h^4/J)+\exp\{-cJ/h^2+Ch\}=o(1). \label{eq:terminal-cck-error} \end{equation} All estimates depend on the endpoint data only through the uniform bounds already established. \smallskip\noindent \underline{\textit{Step 3.}}\quad The Gaussian maximum. We now compare the Gaussian field $\Gamma$ from Step~2 with $\sqrt{2g}\,\phi$, where $\phi$ is the zero-boundary discrete Gaussian free field in $D_y$: \[ E[\phi_x\phi_{x'}]=G_{D_y}(x,x'). \] The extremal-process result of Biskup--Louidor \cite[(1.2), (1.6), and Theorem~2.1]{BiskupLouidor2018}, together with its stated consequence for maxima on open subsets following (1.8), gives tightness of the maximum on a fixed inner subdisk after centering by \[ m_M=2\sqrt g\log M-\frac34\sqrt g\log\log M. \] The digital disks here approximate a disk, and $V_y$ is a fixed nonempty subdisk strictly inside it. In particular, \begin{equation} \max_{x\in V_y}\phi_x=2\sqrt g\,h+O_P(\log h). \label{eq:terminal-dgff-maximum} \end{equation} Only this localization, rather than the full limiting point process, is used below. To pass to expectations, convergence in distribution alone is insufficient. The tightness just stated places every median of the maximum at $m_M+O(1)$. Its maximal coordinate variance is $O(h)$, so Gaussian concentration places the mean within $O(\sqrt h)$ of a median. Therefore \begin{equation} E\max_{x\in V_y}\phi_x=2\sqrt g\,h+O(\sqrt h). \label{eq:terminal-dgff-mean} \end{equation} The covariance error in \eqref{eq:terminal-covariance-comparison} has no sign, so we compare expected maxima by Gaussian interpolation. The following smooth approximation to the maximum makes a uniform entrywise covariance error sufficient. For $\lambda>0$, define \[ F_\lambda(z)=\lambda^{-1}\log\sum_{x\in V_y}e^{\lambda z_x}. \] If $p_x=e^{\lambda z_x}/\sum_{x'}e^{\lambda z_{x'}}$, then \[ \partial_{xx'}F_\lambda =\lambda(p_x\mathbf1_{\{x=x'\}}-p_xp_{x'}),\qquad \sum_{x,x'}|\partial_{xx'}F_\lambda|\le2\lambda. \] Gaussian interpolation between $\Gamma$ and $\sqrt{2g}\,\phi$ therefore gives \[ \begin{aligned} &\left|EF_\lambda(\Gamma)-EF_\lambda(\sqrt{2g}\,\phi)\right|\\ &\quad\le\frac12\max_{x,x'}\left| \operatorname{Cov}(\Gamma_x,\Gamma_{x'})-2gG(x,x')\right| \sup_z\sum_{x,x'}|\partial_{xx'}F_\lambda(z)|\\ &\quad\le C\lambda. \end{aligned} \] Since \[ 0\le F_\lambda(z)-\max_xz_x\le\frac{\log P}{\lambda}, \] the difference of the two expected maxima is bounded by $C\lambda+2\log P/\lambda$. Balancing these errors with $\lambda\asymp\sqrt{\log P}$ yields \begin{equation} \left|E\max_x\Gamma_x-E\max_x\sqrt{2g}\,\phi_x\right| \le C\sqrt{\log P}. \label{eq:terminal-gaussian-mean-comparison} \end{equation} Both fields have maximal coordinate variance $O(h)$. Combining Gaussian concentration with \eqref{eq:terminal-dgff-mean}--\eqref{eq:terminal-gaussian-mean-comparison} gives \begin{equation} \max_x\Gamma_x=2\sqrt2\,g h+O_P(\sqrt h+\log h) =2\sqrt2\,g h+O_P(\sqrt h). \label{eq:terminal-gaussian-maximum} \end{equation} Apply \eqref{eq:terminal-cck} to each of the two half-lines. For any $K>0$, the probability that $\max_xU_x$ is below $2\sqrt2\,gh-K\sqrt h$ is bounded by the corresponding Gaussian lower-tail probability with its threshold increased by $C_7$, plus the error in \eqref{eq:terminal-cck-error}. The upper tail is bounded by the Gaussian upper tail with its threshold decreased by $C_7$, plus the same error. Equation~\eqref{eq:terminal-gaussian-maximum} thus transfers to $\max_xU_x$, uniformly in the regular endpoint data. This argument compares distributions; it requires no pathwise coupling of the maxima. \smallskip\noindent \underline{\textit{Step 4.}}\quad Restoring the means and the terminal depth. The conditional means are not exactly constant in $x$, but \eqref{eq:terminal-conditional-mean} gives \[ \sup_{x\in V_y}\left|\sum_{i\le b}m_x(\omega_i)-gb\right| \le Cb(t_J+\varepsilon_M) =O\left(\frac{J\log J}{\log\log J}\right). \] Using $\sum_iX_i(x)=\sum_im_x(\omega_i)+\sqrt b\,U_x$, we have \[ \left|\max_{x\in V_y}\sum_{i\le b}X_i(x) -gb-\sqrt b\max_{x\in V_y}U_x\right| \le\sup_{x\in V_y}\left|\sum_{i\le b}m_x(\omega_i)-gb\right|. \] The centered maximum estimate from Step~3 and $\sqrt b=O(J)$ now prove \eqref{eq:terminal-maximum-expansion}. Finally, \eqref{eq:terminal-piece-count} gives \begin{equation} \begin{split} gb+2\sqrt2\,g\sqrt b\,h &=2gJ^2-2gJ\log J+O(h^2+h\log J+J). \end{split} \label{eq:terminal-restored-depth} \end{equation} Indeed, \eqref{eq:terminal-piece-count} implies $\sqrt b=\sqrt2L-(\log L)/\sqrt2+O(1+(\log L)^2/L)$. Thus the terminal gain is $4gLh+O(h\log L+h)$, while $gb=2gL^2-2gL\log L+O(L)$. Substituting $L=J-h$ uses \[ 2gL^2+4gLh=2gJ^2-2gh^2,\qquad L\log L=J\log J+O(h\log J), \] which gives \eqref{eq:terminal-restored-depth}. This is how the terminal maximum restores the contribution lost by stopping the coarse construction at $L$. Since $\xi<2$, \[ J\sqrt h+\frac{J\log J}{\log\log J} +h^2+h\log J+J=o(J\log J). \] The level in \eqref{eq:terminal-restored-depth} exceeds $u_{J,2+\eta}$ by $\eta gJ\log J+o(J\log J)$, which proves the final assertion. \end{proof} \subsection{A positive probability of a large disk maximum} \label{sec:disk-lower} The terminal estimate applies to one fixed cell under its own coarse conditioning. We cannot simply select a coarse root and then apply it: the selection may reveal bridge interiors of other cells. Instead, we sum the failure probabilities over all candidate roots. The bounded expected number of roots makes this sum tend to zero, even though their terminal disks may overlap. \begin{proposition}[The disk maximum] \label{prop:disk-maximum-lower} For every fixed $\eta>0$, there is $p_\eta>0$ such that, for all sufficiently large $J$, \begin{equation} P\left(M_{\tau_{e^J}}\ge u_{J,2+\eta}\right)\ge p_\eta. \label{eq:disk-maximum-positive-probability} \end{equation} \end{proposition} \begin{proof} Take $h=(\log J)^\xi$, $1<\xi<2$, and $L=J-h$. Apply the lower-root construction \eqref{eq:ct-external-32}--\eqref{eq:ct-external-38} at depth $L$, with terminal physical radius $M=e^h$. Restrict the centers to a fixed compact subdisk so that their enlarged terminal domains are contained in $D(0,e^J)$. This keeps a fixed positive fraction of the available area, and the estimates are uniform on the restricted set. The endpoint band in \eqref{eq:ct-external-37}, with $\rho_L=2-(\log L)/L$, is \[ \sqrt{2T_L}\in[2L-\log L+z_0,\,2L-\log L+z_0+C_*]. \] Squaring gives the completed-piece count $b=2L^2-2L\log L+O(L)$ in \eqref{eq:terminal-piece-count}. Lemma~\ref{lem:buffered-transfer} transfers the probability of existence of a coarse root directly from Brownian motion; the pair estimate is used on the Brownian side. At each root, the terminal pieces are ordered separately within the $m_0$ source intervals. The inter-source connectors are exposed and omitted from the occupation sums. Their occupation is nonnegative, so omitting it preserves a lower bound for the full-walk maximum. For the lattice coarse-root count $Z_L^S$, the count assertion of that lemma gives \begin{equation} P(Z_L^S>0)\ge p_0,\qquad EZ_L^S\le C. \label{eq:disk-root-count} \end{equation} We first check that imposing moment regularity loses only $o(1)$ in the probability of root existence. Recall that the lattice event $I_y^{S,\circ}$ retains only the radial barrier and terminal band; the Brownian angular screens were dropped in its definition. Thus it is determined by the complete radial words. The scalar radial-word comparison \eqref{eq:ct-smooth-39d} and Rosen's Appendix Theorem~9.1(a), as used in \eqref{eq:ct-transfer-47cprime}, give \begin{equation} P(I_y^{S,\circ})\le Ce^{-2L},\qquad \sum_yP(I_y^{S,\circ})=EZ_L^S\le C, \label{eq:disk-unscreened-count} \end{equation} where the sum is over the deterministic net of $O(e^{2L})$ centers. For every complete radial word realizing $I_y^{S,\circ}$, \eqref{eq:terminal-moment-failure} bounds failure of moment regularity by the same $\epsilon_J=o(J^{-A})$. Consequently, \begin{equation} \begin{split} &E\sum_y\mathbf1_{I_y^{S,\circ}} \mathbf1_{\{\text{moment regularity fails at }y\}}\\ &\quad=\sum_y\sum_{w:\,w\text{ realizes }I_y^{S,\circ}} P(w)P(\text{failure at }y\mid w)\\ &\quad\le\epsilon_J\sum_yP(I_y^{S,\circ}) =\epsilon_JEZ_L^S=o(1). \end{split} \label{eq:disk-irregular-roots} \end{equation} The radial nature of $I_y^{S,\circ}$ is needed for this averaging: the endpoint failure estimate is conditional on $w$, so its bound must be weighted by $P(w)$. Equations \eqref{eq:disk-root-count} and \eqref{eq:disk-irregular-roots} therefore show that a moment-regular lattice root exists with probability at least $p_0-o(1)$. For a fixed root $y$, use its intrinsic erasure sigma-field $\mathcal R_y$. The event $\mathcal G_y$ that $y$ is a moment-regular coarse root is $\mathcal R_y$-measurable. Given this sigma-field, the erased pieces have the independent killed bridge laws used in Lemma~\ref{lem:terminal-maximum-lower}. Let \[ \mathcal A_y=\left\{\max_{x\in V_y}\sum_{i\le b}X_i(x) \ge u_{J,2+\eta}\right\}. \] The uniform conclusion of that lemma gives a deterministic $\widetilde\epsilon_J\downarrow0$ such that \begin{equation} P(\mathcal G_y\cap \mathcal A_y^c) =E[\mathbf1_{\mathcal G_y}P(\mathcal A_y^c\mid\mathcal R_y)] \le\widetilde\epsilon_JP(\mathcal G_y). \label{eq:disk-fixed-root-failure} \end{equation} Consequently \begin{equation} \begin{split} P(\exists y:\mathcal G_y\cap \mathcal A_y) &\ge P(\exists y:\mathcal G_y)-\sum_yP(\mathcal G_y\cap \mathcal A_y^c)\\ &\ge p_0-o(1)-\widetilde\epsilon_J EZ_L^S \ge p_0/2 \end{split} \label{eq:disk-successful-root} \end{equation} for all sufficiently large $J$. Each summand uses the intrinsic bridge disintegration for its own cell; the union bound does not require disjoint terminal disks. On the event in \eqref{eq:disk-successful-root}, the full local-time maximum is at least the terminal maximum of a successful cell. This proves \eqref{eq:disk-maximum-positive-probability}. \end{proof} \subsection{An almost-sure lower bound at deterministic times} \label{sec:deterministic-lower} The disk estimate gives one attempt with probability bounded away from zero. We make $(\log N)^a$ independent attempts within the first $N$ increments. Each attempt has a smaller time budget, so its logarithmic radius decreases by about $(a/2)\log\log N$. Choosing $a$ small keeps the resulting loss below the available $J\log J$ margin, while the probability that all attempts fail is summable. \begin{proposition}[The deterministic-time maximum] \label{prop:deterministic-maximum-lower} Let $N_j=\lfloor e^{2j}\rfloor$. For every $\eta>0$, almost surely, \[ M_{N_j}\ge u_{j,2+\eta} \] for all sufficiently large $j$. \end{proposition} \begin{proof} Reserve part of the target margin for shortening the independent blocks: first fix $0<\delta<\eta$, and then choose $a>0$ with $\delta+2a<\eta$. Proposition~\ref{prop:disk-maximum-lower} gives a disk-success probability $p_\delta>0$. By Lemma~\ref{lem:elementary-walk-estimates}, choose a fixed $C$ large enough that \[ \sup_R P_0(\tau_R>CR^2)\le p_\delta/2. \] For the walk started at the center of a radius-$R$ disk, \[ P\left(M_{\tau_R}\ge u_{\log R,2+\delta},\ \tau_R\le CR^2\right) \ge p_\delta-P(\tau_R>CR^2)\ge p_\delta/2. \] Translation invariance therefore gives the same success probability $p=p_\delta/2$ for a segment of length $CR^2$ from any starting site, uniformly for sufficiently large $R$. Partition the first $N_j$ increments into \[ q_j=\lfloor(\log N_j)^a\rfloor \] disjoint blocks of length $m_j=\lfloor N_j/q_j\rfloor$, ignoring the remainder. Use the sites immediately after the increments in each block to define its within-block local times. The translated block paths depend on disjoint increment sets and are independent. Their local times count disjoint sets of observation times, so each block maximum is bounded by $M_{N_j}$. Changing the endpoint convention changes a local time by at most a bounded amount, which the strict margin $\delta+2a<\eta$ absorbs. Choose $R_j$ by \[ \log R_j=\frac12\log(m_j/C) =j-\frac a2\log j+O(1). \] The loss in logarithmic radius is therefore $(a/2)\log j+O(1)$. Substituting this into the quadratic leading term gives \[ 2g(\log R_j)^2 =2gj^2-2agj\log j+O(j+(\log j)^2), \] while $g(\log R_j)\log\log R_j=gj\log j+O(j+(\log j)^2)$. Hence \begin{equation} u_{\log R_j,2+\delta} =u_{j,\,2+\delta+2a+o(1)} \ge u_{j,2+\eta} \label{eq:deterministic-block-level} \end{equation} for all sufficiently large $j$. The coefficient $2a$ is the cost of splitting the time budget into $q_j$ blocks. Each block therefore succeeds at the target level with probability at least $p$, and independence implies \begin{equation} P(M_{N_j}\tfrac12\), almost surely, \begin{equation} \liminf_{n\to\infty} \frac{R_n^-}{\sqrt n/(\log n)^\gamma}=\infty. \label{eq:escape-80} \end{equation} \end{proposition} \begin{proof} Fix \(\gamma>1/2\) and a positive integer spatial multiplier \(A_0\). Choose \[ 0<\delta<\frac{2\gamma-1}{4}. \] Set \(N_j=\lfloor e^{2j}\rfloor\). Proposition~\ref{prop:deterministic-maximum-lower} gives, for all sufficiently large \(j\), \begin{equation} M_{N_j}\ge u_{j,2+\delta}. \label{eq:escape-81} \end{equation} Consider any \(n\in[N_j,N_{j+1}]\cap\mathbb N\). If \(x\in\mathcal F_n\) lies in \[ B\left(0,A_0\frac{\sqrt n}{(\log n)^\gamma}\right), \] then monotonicity of local times and \eqref{eq:escape-81} give \[ L_{N_{j+1}}(x)\ge L_n(x)=M_n\ge M_{N_j} \ge u_{j,2+\delta}. \] Since \(n\le N_{j+1}\) and \(\log n\ge\log N_j\), the same site lies in \begin{equation} B\left(0,C_{A_0}e^{j+1}j^{-\gamma}\right) \label{eq:escape-82} \end{equation} throughout the block. Take \[ K_j=C_*e^{j+1}\sqrt{\log j},\qquad J'_j=\log K_j. \] By the maximal inequality in Lemma~\ref{lem:elementary-walk-estimates}, \(C_*\) can be chosen so that \begin{equation} \sum_jP(\tau_{K_j}\le N_{j+1})<\infty. \label{eq:escape-83} \end{equation} For large \(j\), the ball \eqref{eq:escape-82} is contained in \[ B(0,K_j(J'_j)^{-\gamma}). \] Moreover \(J'_j=j+O(\log\log j)\), and the fixed extra coefficient \(\delta\) gives \begin{equation} u_{j,2+\delta}\ge u_{J'_j,2+2\delta} \label{eq:escape-84} \end{equation} for all large \(j\), because \[ u_{j,2+\delta}-u_{J'_j,2+2\delta} =\delta g j\log j+O(j\log\log j)>0. \] Thus the extra coefficient $\delta$ absorbs the enlargement of the stopping disk. Choose independently an upper-error exponent \(A_{\rm err}>2\). If \eqref{eq:escape-81} holds and the early exit in \eqref{eq:escape-83} does not occur, any such favorite forces \[ \max_{x\in B(0,K_j(J'_j)^{-\gamma})} L_{\tau_{K_j}}(x)\ge u_{J'_j,2+2\delta}. \] Proposition~\ref{prop:localized-disk-upper}, with \(\beta=\gamma\) and \(c=2+2\delta\), bounds the probability of this last event by \begin{equation} (J'_j)^{-2\gamma+2\delta+o(1)}+O(j^{-A_{\rm err}}). \label{eq:escape-85} \end{equation} Our choice of \(\delta\) makes the power in \eqref{eq:escape-85} strictly smaller than \(-1\). Hence \eqref{eq:escape-85} is summable. The lower bound \eqref{eq:escape-81} holds eventually almost surely, and Borel--Cantelli excludes both upper events and early exits eventually. The block implication therefore proves that eventually no favorite in the \(j\)-th block lies in the ball with multiplier \(A_0\). Intersecting over \(A_0\in\mathbb N\) proves \eqref{eq:escape-80}. \end{proof} \begin{proof}[Proof of Theorem~\ref{main:precise-escape-rate}] Propositions~\ref{prop:critical-recurrence} and \ref{prop:supercritical-escape} give the two assertions for each fixed exponent. Intersect the probability-one events for rational $\gamma>1/2$ and for the single critical exponent $1/2$. If $\gamma>1/2$ is real, choose a rational $q\in(1/2,\gamma)$; then \[ \frac{R_n^-}{a_\gamma(n)} =\frac{R_n^-}{a_q(n)}(\log n)^{\gamma-q}\longrightarrow\infty. \] For $\gamma\le1/2$, the inequality $R_n^+/a_\gamma(n)\le R_n^+/a_{1/2}(n)$ gives the other assertion on the same event. This proves the simultaneous formulation. \end{proof} % ---- notation ---- \clearpage \appendix \section{Notation guide}\label{sec:notation} We use $n$ for physical time and $L=J-h$ for the retained traversal depth; $J$ and $h$ are the outer and terminal logarithmic radii. The table collects notation used across subsections. \footnotesize \setlength{\tabcolsep}{4pt} \begin{longtable}{@{}>{\raggedright\arraybackslash}p{.18\textwidth} >{\raggedright\arraybackslash}p{.55\textwidth} >{\raggedright\arraybackslash}p{.23\textwidth}@{}} \toprule Symbol & Meaning & Where used \\ \midrule \endhead $L_n(x),M_n$ & Site local time and its spatial maximum. & Section~\ref{sec:introduction} \\ $\mathcal F_n,R_n^-,R_n^+$ & Favorite set and its nearest and farthest distances from the origin. & Section~\ref{sec:introduction} \\ $a_\gamma(n)$ & $\sqrt n/(\log n)^\gamma$. & Theorem~\ref{main:precise-escape-rate} \\ $g,u_{J,c}$ & $g=2/\pi$ and $u_{J,c}=2gJ^2-cgJ\log J$. & Both disk estimates \\ $D(x,r),\tau_r$ & Lattice disk and exit time from $D(0,r)$. & Section~\ref{sec:elementary} \\ $G_D$ & Green function for the walk killed on leaving $D$. & Sections~\ref{sec:elementary} and \ref{sec:terminal-upper} \\ $z,N,K$ & Geometric weight, $N=(1-z)^{-1}$ and clock with $P(K=n)=(1-z)z^n$. & Section~\ref{sec:recurrence} \\ $X_K^*,U_x$ & Priority-selected favorite and independent site priorities. & Lemma~\ref{lem:plateau-bijection} \\ $\sigma,\ell$ & First and last visits to the selected favorite. & Lemma~\ref{lem:plateau-bijection} \\ $\mathcal B,\tau_c$ & Exposed excursion/suffix tuple and terminal-arm killing time. & Lemma~\ref{lem:plateau-bijection} \\ $J,h,M,L$ & Outer logarithmic radius, terminal logarithmic radius, $M=e^h$, and retained traversal depth $L=J-h$. & Sections~\ref{sec:continuum-transfer}--\ref{sec:terminal-lower} \\ $r_j,\vartheta_j$ & Conformal radius $r_0e^{-j}$ and geodesic radius $2\arctan(r_j/2)$. & Section~\ref{sec:continuum-transfer} \\ $\rho_L,\alpha_{z,\pm}$ & Barrier slope $2-(\log L)/L$ and its upper/lower curved barriers. & Section~\ref{sec:continuum-transfer} \\ $F_L,k_y$ & Deterministic net of centers and a center's source shell. & Section~\ref{sec:continuum-transfer} \\ $T_{y,j}^{1,m_0}$ & Traversal count within the first $m_0$ complete source excursions. & Section~\ref{sec:continuum-transfer} \\ $I_y^\circ,S_y$ & Truncated Brownian root event and connector safety event. & Section~\ref{sec:continuum-transfer} \\ $\chi_y,\Psi_J$ & Centered conformal coordinate and dilated stereographic chart. & Section~\ref{sec:continuum-transfer} \\ $w,Q,G_0$ & Complete radial word, number of shell decisions, and number of nonempty source blocks. & Section~\ref{sec:terminal-upper} \\ $D_y,A_y,V_y,E_y$ & Terminal killing disk, entrance interface, core, and directed exit edges. & Section~\ref{sec:terminal-upper} \\ $\omega_i,X_i(x)$ & Entrance/exit pair and occupation at $x$ of its killed bridge. & Section~\ref{sec:terminal-upper} \\ $m_x,q_{x,x'}$ & First and second occupation moments under the bridge with prescribed endpoints. & Sections~\ref{sec:terminal-upper}--\ref{sec:terminal-lower} \\ $P_w,\widetilde P_w$ & True complete-word endpoint law and its sequentially normalized comparison law. & Section~\ref{sec:terminal-upper} \\ $\Lambda_M$ & Reference entrance/exit measure obtained from harmonic measure at the center. & Section~\ref{sec:terminal-upper} \\ $\mathcal R_y$ & Intrinsic information exposing exterior paths and bridge endpoints while erasing bridge interiors. & Sections~\ref{sec:continuum-transfer} and \ref{sec:terminal-lower} \\ $Z_L^S$ & Number of lattice centers satisfying the radial lower barrier and terminal endpoint band. & Sections~\ref{sec:continuum-transfer} and \ref{sec:terminal-lower} \\ \bottomrule \end{longtable} \normalsize % ---- references ---- \clearpage \begin{thebibliography}{99} \raggedright \bibitem{AthreyaEtAl2022} S. Athreya, N. Gadhiwala and R. R. Radhakrishnan, \emph{Elliptic Harnack inequality for $\mathbb Z^d$}, Involve \textbf{15} (2022), 687--708. \href{https://doi.org/10.2140/involve.2022.15.687}{doi:10.2140/involve.2022.15.687}. \href{https://arxiv.org/abs/2201.11947}{arXiv version}. \bibitem{Bass2023} R. F. Bass, \emph{The rate of escape of the most visited site of Brownian motion}, Electron. J. Probab. \textbf{28} (2023), 1--12. \href{https://doi.org/10.1214/23-EJP916}{doi:10.1214/23-EJP916}. See also \href{https://arxiv.org/abs/1303.2040v5}{arXiv:1303.2040v5}, Theorem~2.2. \bibitem{BiskupLouidor2018} M. Biskup and O. Louidor, \emph{Full extremal process, cluster law and freezing for the two-dimensional discrete Gaussian free field}, Adv. Math. \textbf{330} (2018), 589--687. \href{https://doi.org/10.1016/j.aim.2018.02.018}{doi:10.1016/j.aim.2018.02.018}. \href{https://arxiv.org/abs/1606.00510}{arXiv:1606.00510}. \bibitem{CCK2016} V. Chernozhukov, D. Chetverikov and K. Kato, \emph{Empirical and multiplier bootstraps for suprema of empirical processes of increasing complexity, and related Gaussian couplings}, Stochastic Process. Appl. \textbf{126} (2016), 3632--3651. \href{https://doi.org/10.1016/j.spa.2016.04.009}{doi:10.1016/j.spa.2016.04.009}. \href{https://arxiv.org/abs/1502.00352}{arXiv:1502.00352}. \bibitem{Dembo2005} A. Dembo, \emph{Favorite points, cover times and fractals}, in \emph{Lectures on Probability Theory and Statistics}, Lecture Notes in Math. \textbf{1869}, Springer (2005), 1--101. \href{https://doi.org/10.1007/11429579_1}{doi:10.1007/11429579\_1}. \bibitem{Dembo2006} A. Dembo, \emph{Simple random covering, disconnection, late and favorite points}, in \emph{Proceedings of the International Congress of Mathematicians, Madrid 2006}, Vol.~III, EMS (2007), 535--558. \href{https://doi.org/10.4171/022-3/28}{doi:10.4171/022-3/28}. \bibitem{DPRZ2001} A. Dembo, Y. Peres, J. Rosen and O. Zeitouni, \emph{Thick points for planar Brownian motion and the Erd\H{o}s--Taylor conjecture on random walk}, Acta Math. \textbf{186} (2001), 239--270. \href{https://doi.org/10.1007/BF02401841}{doi:10.1007/BF02401841}. \bibitem{ErdosRevesz1987} P. Erd\H{o}s and P. R\'ev\'esz, \emph{Problems and results on random walks}, in \emph{Mathematical Statistics and Probability Theory}, Vol.~B, D. Reidel (1987), 59--65. \href{https://renyi.hu/~p_erdos/1987-26.pdf}{Full text}. \bibitem{HLOZ2025} C. Hao, X. Li, I. Okada and Y. Zheng, \emph{Favorite sites for simple random walk in two and more dimensions}, Probab. Theory Related Fields \textbf{195} (2026), 1765--1822. \href{https://doi.org/10.1007/s00440-025-01441-1}{doi:10.1007/s00440-025-01441-1}. \href{https://arxiv.org/abs/2409.00995}{arXiv:2409.00995}. The planar escape-rate question is discussed in Remark~1.4(1); the elementary walk estimates cited here are in Lemma~2.1, (2.5). \bibitem{KozdronLawler2005} M. J. Kozdron and G. F. Lawler, \emph{Estimates of random walk exit probabilities and application to loop-erased random walk}, Electron. J. Probab. \textbf{10} (2005), 1442--1467. \href{https://arxiv.org/abs/math/0501189}{arXiv version}. \bibitem{LifshitsShi2004} M. A. Lifshits and Z. Shi, \emph{The escape rate of favorite sites of simple random walk and Brownian motion}, Ann. Probab. \textbf{32} (2004), 129--152. \href{https://doi.org/10.1214/aop/1078415831}{doi:10.1214/aop/1078415831}. \bibitem{Revesz1990} P. R\'ev\'esz, \emph{Random Walk in Random and Non-Random Environments}, World Scientific, Singapore, 1990. \href{https://doi.org/10.1142/1107}{doi:10.1142/1107}. \bibitem{Rosen2022} J. Rosen, \emph{Tightness for thick points in two dimensions}, arXiv:2203.14394 (2022). \href{https://arxiv.org/abs/2203.14394}{arXiv version}. \bibitem{Zaitsev2003} A. Yu. Zaitsev, \emph{Estimates for the strong approximation in multidimensional central limit theorem}, arXiv:math/0304373 (2003). \href{https://arxiv.org/abs/math/0304373}{arXiv version}. \end{thebibliography} \end{document}