Tail probability of maximal displacement in critical and subcritical branching stable processes
Haojie Hou, Yiyang Jiang, Yan-Xia Ren, and, Renming Song · 2025
Read Tail probability of maximal displacement in critical and subcritical branching stable processes by Haojie Hou, Yiyang Jiang, Yan-Xia Ren, and, Renming Song (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-05540.txt.
Introduction
Haojie Hou, Yiyang Jiang, Yan-Xia Ren, and Renming Song worked out in 2025 how far a branching process can reach. In the process, each particle splits into offspring that move by jumps with a heavy tail. The team studied forms that are just at, or below, the line where the family dies out in time. They asked how likely the farthest reach is to pass a high mark.
They give exact rates for that tail chance. For families below the line with upward jumps, the rate holds in full. For one-sided downward jump forms, it holds when the offspring law meets a mild growth rule. For the borderline case, it holds when offspring counts fall in the pull of a set spread law. The results pin down rare far runs.
Catalog record
| CREATOR | Haojie Hou, Yiyang Jiang, Yan-Xia Ren, and, Renming Song |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 78 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-05540
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Tail probability of maximal displacement in critical and subcritical branching stable processes Haojie Hou, Yiyang Jiang, Yan-Xia Ren, and, Renming Song Abstract In this paper, we study critical and subcritical branching α \alpha -stable processes, α ∈ ( 0 , 2 ) \alpha\in(0,2) . We obtain the exact asymptotic behaviors of the tails of the maximal positions of all subcritical branching α \alpha -stable processes with positive jumps. In the case of subcritical branching spectrally negative α \alpha -stable processes, we obtain the exact asymptotic behaviors of the tails of the maximal positions under the assumption that the offspring distributions satisfy the L log L L\log L condition. For critical branching α \alpha -stable processes, we obtain the exact asymptotic behaviors of the tails under the assumption that the offspring distributions belong to the domain of attraction of a γ \gamma -distribution, γ ∈ ( 1 , 2 ] \gamma\in(1,2] . Keywords— Branching stable process; rightmost position; integral equation 2020 MSC— 60J80; 60G52; 60G51; 60G70. 1 Introduction and main results 1.1 Introduction A branching Lévy process is a continuous-time Markov process defined as follows. At time 0, there is a particle at x\in\mathbb{R} and it moves according to a Lévy process (\xi_{t},\mathbf{P}_{x}) . After an exponential time with parameter 1 , independent of the motion, it dies and produces k offspring with probability p_{k} , k\geqslant 0 . The offspring move independently according to \xi from the place where they are born and obey the same branching mechanism as their parent. This procedure goes on. For t\geqslant 0 , let N_{t} be the collection of particles alive at time t . For u\in N_{t} , we use X_{u}(t) to denote the position of the particle u at time t . The point process (Z_{t})_{t\geqslant 0} defined by Z_{t}:=\sum_{u\in N_{t}}\delta_{X_{u}(t)},\hskip 18.49988ptt\geqslant 0, is called a branching Lévy process. We will denote the law of (Z_{t})_{t\geqslant 0} by \mathbb{P}_{x} and write \mathbb{P} for \mathbb{P}_{0} for simplicity. When \xi is a Brownian motion, (Z_{t})_{t\geqslant 0} is called a branching Brownian motion. When \xi is an \alpha -stable process, \alpha\in(0,2) , (Z_{t})_{t\geqslant 0} is called a branching \alpha -stable process. Let m:=\sum^{\infty}_{k=0}kp_{k} be the mean of the offspring distribution. When m>1(=1,<1) , we say that the branching Lévy process (Z_{t})_{t\geqslant 0} is supercritical (critical, subcritical). It is well known that in the critical and subcritical cases, (Z_{t})_{t\geqslant 0} will die out in finite time with probability 1. Thus in this case we can define the maximal position of (Z_{t})_{t\geqslant 0} by M:=\sup_{t>0}\sup_{u\in N_{t}}X_{u}(t). When m=1 and \xi is a standard Brownian motion, Sawyer and Fleischman [ 17 ] proved that, under the assumption \sum_{k=0}^{\infty}k^{3}p_{k}<\infty , \displaystyle\lim_{x\to+\infty}x^{2}\mathbb{P}\left(M\geqslant x\right)=\frac{6}{\sigma^{2}}, (1) where \sigma^{2} is the variance of the offspring distribution. Profeta [ 13 ] extended the result above to some critical branching spectrally negative Lévy processes under the same third moment condition on the offspring distribution. When m=1 , \xi is an \alpha -stable process with positive jumps and \sum_{k=0}^{\infty}k^{3}p_{k}<\infty , Lalley and Shao [ 10 ] , and Profeta [ 12 ] proved that \displaystyle\lim_{x\to+\infty}x^{\alpha/2}\mathbb{P}\left(M\geqslant x\right)=c_{1}(\alpha), (2) where c_{1}(\alpha) is an explicit positive constant. For critical branching spectrally negative \alpha -stable processes with \alpha\in(1,2) , Profeta [ 13 , Corollary 1.3] proved that, when \sum_{k=0}^{\infty}k^{3}p_{k}<\infty , \displaystyle 0<\liminf_{x\to\infty}x^{\alpha}\mathbb{P}\left(M\geqslant x\right)\leqslant\limsup_{x\to\infty}x^{\alpha}\mathbb{P}\left(M\geqslant x\right)<\infty. (3) When m<1 and \xi is a standard Brownian motion, Sawyer and Fleischman [ 17 ] proved that, under the assumption \sum_{k=0}^{\infty}k^{3}p_{k}<\infty , there exists a function c(\cdot) bounded between two positive constants such that \displaystyle\lim_{x\to+\infty}\frac{\mathbb{P}\left(M\geqslant x\right)}{(1-m)c(x)e^{-\sqrt{2(1-m)}x}}=1. (4) For subcritical branching \alpha -stable processes with positive jumps, Profeta [ 12 ] proved that, under the assumption \sum_{k=0}^{\infty}k^{3}p_{k}<\infty , \displaystyle\lim_{x\to+\infty}x^{\alpha}\mathbb{P}\left(M\geqslant x\right)=\frac{c_{2}(\alpha)}{1-m} (5) for some explicit constant c_{2}(\alpha) . For subcritical branching spectrally negative processes, Profeta [ 13 , Theorem 1] proved that, under the assumption \sum_{k=0}^{\infty}k^{3}p_{k}<\infty , \displaystyle\lim_{x\to\infty}e^{c_{4}x}\mathbb{P}\left(M\geqslant x\right)=c_{3} (6) for some explicit positive constants c_{3},c_{4} . For results on the asymptotic behaviors of the tails of maximal positions of (sub)critical branching random walks, see [ 9 , 11 ] . The purpose of this paper is to establish the exact asymptotic behaviors of the tails of the maximal positions of (sub)critical branching \alpha -stable processes under minimal conditions on the offspring distributions. More precisely, we will obtain (i) the exact asymptotic behaviors of the tails of the maximal positions of subcritical branching \alpha -stable processes with positive jumps without any extra assumption on the offspring distributions; (ii) the exact asymptotic behaviors of the tail of the maximal positions of subcritical branching spectrally negative \alpha -stable processes under the assumption that the offspring distributions satisfy the L\log L condition; (iii) the exact asymptotic behaviors of the tails of critical branching \alpha -stable processes under the assumption that the offspring distributions belong to the domain of attraction of a \gamma -distribution, \gamma\in(1,2] . 1.2 Main results Before we state our main results, we recall some useful facts about stable processes. In this paper, we always assume that the spatial motion \xi is a (strictly) \alpha -stable process, \alpha\in(0,2) . For basic information on \alpha -stable process, see, for instance, [ 1 , Chap. VIII] . We assume that the Lévy measure of \xi is given by v_{\alpha}(\mathrm{d}x):=c_{+}x^{-(1+\alpha)}{\bf 1}_{(0,\infty)}(x){\rm d}x+c_{-}|x|^{-(1+\alpha)}{\bf 1}_{(-\infty,0)}(x){\rm d}x, (7) where c_{+} and c_{-} are non-negative numbers with at least one of them being positive. Let \Psi(\theta):=-\ln\mathbf{E}(e^{\mathrm{i}\theta\xi_{1}}) be the characteristic exponent of \xi . It is known (see, for instance [ 8 , Chapter 1.2.6] ) that, \Psi(\theta)=\begin{cases}c_{*}|\theta|^{\alpha}\left(1-\mathrm{i}\beta\tan\frac{\pi\alpha}{2}\operatorname{sgn}\theta\right),&\text{ for }\alpha\in(0,1)\cup(1,2)\text{ and }\beta\in[-1,1],\\ c_{*}|\theta|+\mathrm{i}\theta\eta,&\text{ for }\alpha=1\end{cases} where \eta\in\mathbb{R} , c_{*}:=-\left(c_{+}+c_{-}\right)\Gamma(-\alpha)\cos(\pi\alpha/2) , and \beta=\left(c_{+}-c_{-}\right)/\left(c_{+}+c_{-}\right) if \alpha\in(0,1)\cup(1,2) . Here \operatorname{sgn}\theta=1_{(\theta>0)}-1_{(\theta<0)} . Note that, for (strictly) 1-stable process, c_{+}=c_{-} (see, for example, [ 16 , Theorem 14.7 (v)] ). It follows from [ 15 , Example 1.1, Lemma 2.1 and the last sentence in Subsection 1.1] (see also [ 1 , Proposition VIII.4] ) that, when c_{+}>0 , \displaystyle\lim_{x\to\infty}x^{\alpha}\mathbf{P}_{0}(\xi_{1}\geqslant x)=\nu_{\alpha}((1,\infty))=\frac{c_{+}}{\alpha}. (8) When c_{+}=0 and \alpha\in(0,1) , -\xi is a subordinator and M=0 almost surely. A spectrally negative 1-stable process reduces to the drift -\eta t . When \eta\geqslant 0 , M=0 . Thus in the case of branching spectrally negative \alpha -stable processes, exclude the above cases and assume either \alpha\in(1,2) , or \alpha=1 and \eta<0 . In this case, it is easy to see that for any \lambda>0 , \displaystyle\mathbf{E}_{0}(e^{\lambda\xi_{1}})=\exp\left\{C_{1}(\alpha)\lambda^{\alpha}\right\},\hskip 9.24994pt\mbox{where}\hskip 9.24994ptC_{1}(\alpha)=-\Psi(-\mathrm{i}). (9) Define \tau_{y}=\inf\{t>0,\xi_{t}\geqslant y\} . For any x<y , combining ( 9 ) with the fact that \mathbf{P}_{x}\left(\xi_{\tau_{y}}=y\right)=1 , we have that for any \lambda>0 , \displaystyle\mathbf{E}_{x}\left(e^{-\lambda\tau_{y}}\right)=\exp\left\{-C_{1}(\alpha)^{-1/\alpha}\lambda^{1/\alpha}(y-x)\right\}. (10) In the critical case, we will need the following assumption on the offspring distribution: (H) The offspring distribution \{p_{k}:k\geqslant 0\} belongs to the domain of attraction of a \gamma -stable, \gamma\in(1,2] , distribution. More precisely, either there exist \gamma\in(1,2) and \kappa_{\gamma}\in(0,\infty) such that \lim_{n\to\infty}n^{\gamma}\sum_{k=n}^{\infty}p_{k}=\kappa_{\gamma}, or that (corresponding \gamma=2 ) \sum_{k=0}^{\infty}k^{2}p_{k}<\infty. Recall that \sigma^{2} is the variance of the offspring distribution. We define \displaystyle C_{2}(\gamma):=\begin{cases}\frac{\Gamma(2-\gamma)}{\gamma-1}\kappa_{\gamma}&\text{ for }\gamma\in(1,2);\\ \frac{1}{2}\sigma^{2}&\text{ for }\gamma=2.\end{cases} (11) Our first main result is on (sub)critical branching stable processes with positive jumps. Theorem 1 . Suppose that c_{+}>0 . i) If m<1 , then \displaystyle\lim_{x\to+\infty}x^{\alpha}\mathbb{P}\left(M\geqslant x\right)=\frac{c_{+}}{(1-m)\alpha}. (12) ii) If m=1 and (H) holds, then \lim_{x\rightarrow+\infty}x^{\alpha/\gamma}\mathbb{P}\left(M\geqslant x\right)=\left(\frac{c_{+}}{\alpha C_{2}(\gamma)}\right)^{\frac{1}{\gamma}}, where C_{2}(\gamma) is defined in ( 11 ). Remark 1 . Theorem 1 (ii) is also contained in Theorem