1-stable fluctuations in branching Brownian motion at critical temperature. I: The derivative martingale (Q2280545)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | 1-stable fluctuations in branching Brownian motion at critical temperature. I: The derivative martingale |
scientific article |
Statements
1-stable fluctuations in branching Brownian motion at critical temperature. I: The derivative martingale (English)
0 references
18 December 2019
0 references
Let \((Z_t, \mathcal{F}_t)_{t\geq 0}\) be the derivative martingale in a branching Brownian motion, where \((\mathcal{F}_t)_{t\geq 0}\) is the natural filtration of the branching Brownian motion. It is assumed that the reproduction random variable \(L\) satisfies \(\mathbb{E}L>1\) (supercriticality) and \(\mathbb{E}L\log_+^3 L<\infty\), and that some additional standard conditions hold which particularly ensure that \(\lim_{t\to\infty} Z_t=Z_\infty\) almost surely (a.s.), where \(Z_\infty>0\) a.s. on the event of survival. The authors are interested in the rate of distributional convergence of \(Z_t\) to its limit \(Z_\infty\). Their main result states that the conditional finite-dimensional distributions of \((t^{1/2}(Z_\infty-Z_{at}+Z_\infty (2\pi at)^{-1/2}\log t))_{a\geq 1}\) given \(\mathcal{F}_t\) converge weakly in probability as \(t\to\infty\) to the conditional finite-dimensional distributions of \((S_{a^{-1/2}Z_\infty})_{a\geq 1}\) given \(\mathcal{F}_\infty\). Here, \((S_t)_{t\geq 0}\) is a spectrally positive \(1\)-stable Lévy process with explicitly given parameters which is assumed independent of \(Z_\infty\). It is known from Maillard's earlier work that \(\mathbb{E}Z_\infty 1_{\{Z_\infty\leq x\}}-\log x\) converges as \(x\to\infty\) to a finite constant provided that \(\mathbb{E}L\log_+^3 L<\infty\). This justifies the appearance (and actually the necessity as the authors argue) of the latter condition, for the aforementioned constant pops up as a parameter of \((S_t)_{t\geq 0}\). The authors derive the one-dimensional convergence in their limit theorem from the convergence of the corresponding characteristic functions. As far as the finite-dimensional convergence is concerned, this straightforward idea does not seem to apply. Instead, the authors develop a non-trivial approach which includes introducing a killing barrier and a modified derivative martingale, and a detailed investigation of the contributions to \(Z_\infty\) of particles that stay above the barrier and those that stay below the barrier. For a branching random walk a similar result concerning the one-dimensional convergence was proved in [\textit{D. Buraczewski} et al., ``On the derivative martingale in a branching random walk'', Preprint, \url{arXiv:2002.05215}].
0 references
branching Brownian motion
0 references
derivative martingale
0 references
fluctuations
0 references
functional convergence
0 references
1-stable Lévy process
0 references
F-KPP equation
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references