On the recurrence set of planar Markov random walks (Q742108): Difference between revisions

Based on the authors' abstract and introduction: In this paper, the authors consider the following Markov random walk \((X_n, S_n)_{n \in \mathbb{N}}\) with space \(\mathbb{X}\times\mathbb{R}^2\). Let \(\mathbb{S}\) be a two-dimensional closed subgroup of \(\mathbb{R}^2\) satisfying the following condition: \[ \forall x \in \mathbb{X}, \forall n \in \mathbb{N} \;\mathbb{P}_{(x,0)}(S_n\in\mathbb{S})=1, \] where the transition kernel of \((X_n)_{n\in\mathbb{N}}\) admits an invariant probability measure on \(\mathbb{X}\), called \(\pi\), and \(S_1\) is \(\mathbb{P}_{(\pi,0)}\) centered, namely: \(S_1\) is \(\mathbb{P}_{(\pi,0)}\)-integrable and \(\mathbb{E}_{(\pi,0)}[S_1]=0\). The Haar measure on \(\mathbb{S}\) is called \(m_{\mathbb{S}}\). Denote by \(B(s,\varepsilon)\) the open ball in \(\mathbb{R}^2\) centered at \(s\) with radius \(\varepsilon\). Denote by \(\mathcal{B}(\mathbb{R}^2)\) the Borel \(\sigma\)-algebra of \(\mathbb{R}^2\). The authors investigate properties of recurrent planar Markov random walks. More precisely, they study the set of recurrence points with the use of local limit theorems. The recurrence property of \(d\)-dimensional random walks has been investigated in many papers. This study is well-known for i.i.d.\ increments. For the dependent case, early results were stated in [\textit{H. Berbee}, Z. Wahrscheinlichkeitstheor. Verw. Geb. 56, 531--536 (1981; Zbl 0443.60070); \textit{J. P. Conze}, Ergodic Theory Dyn. Syst. 19, No. 5, 1233--1245 (1999; Zbl 0973.37007); \textit{F. M. Dekking}, Z. Wahrscheinlichkeitstheor. Verw. Geb. 61, 459--465 (1982; Zbl 0479.60070); \textit{K. Schmidt}, Z. Wahrscheinlichkeitstheor. Verw. Geb. 68, 75--95 (1984; Zbl 0535.60003); C. R. Acad. Sci., Paris, Sér. I, Math. 327, No. 9, 837--842 (1998; Zbl 0923.60090); in: Dynamics and stochastics. Festschrift in honor of M. S. Keane. Selected papers based on the presentations at the conference `Dynamical systems, probability theory, and statistical mechanics', Eindhoven, The Netherlands, January 3--7, 2005, on the occasion of the 65th birthday of Mike S. Keane. Beachwood, OH: IMS, Institute of Mathematical Statistics. 78--84 (2006; Zbl 1137.37004)] for random walks with stationary increments, in [\textit{H. Hennion}, Ann. Inst. Henri Poincaré, Nouv. Sér., Sect. B 18, 277--291 (1982; Zbl 0487.60058)] for MRWs (case \( d \geq 3\)) associated with uniformly ergodic Markov chains, in [\textit{Y. Guivarc'h}, Lect. Notes Math. 1096, 301--332 (1984; Zbl 0562.60074)] for MRWs associated with strongly ergodic Markov chains, and in [\textit{G. Högnäs}, Math. Scand. 58, 35--45 (1986; Zbl 0578.60011)] for additive functionals of Harris-recurrent Markov chains. For general stationary \(\mathbb{R}^2\)-valued random walks, the link between CLT and recurrence of \((S_n)_n\) has been investigated by J. P. Conze in [Zbl 0973.37007] and by K. Schmidt [Zbl 0923.60090] in the situation when the CLT holds with the standard normalization in \(\sqrt{n}\). Transience/recurrence properties of MRWs (with \(\mathbb{R}^d\)-valued second component) have been investigated in [Zbl 0562.60074] on the basis of a local limit theorem (LLT) obtained via the standard Nagaev-Guivarc'h spectral method. The authors' work uses a similar approach to that of [Zbl 0562.60074] and [\textit{D. Szász} and \textit{T. Varjú}, Ergodic Theory Dyn. Syst. 24, No. 1, 257--278 (2004; Zbl 1115.37009); J. Stat. Phys. 129, No. 1, 59--80 (2007; Zbl 1128.82011)], but it goes beyond the question of recurrence. In fact, they want to investigate the set of recurrence points \(\mathcal{R}_{(\mu, 0)}\), also called recurrence set, defined by \[ \mathcal{R}_{(\mu,0)} := \{ s\in \mathbb{S} : \forall \varepsilon >0, \mathbb{P}_{(\mu, 0)}(|S_n -s | < \varepsilon \text{i.o.})=1 \}. \] The recurrence set is well known in the i.i.d.\ case, and it has been fully investigated in [\textit{G. Alsmeyer}, Probab. Math. Stat. 21, No. 1, 123--134 (2001; Zbl 0992.60068)] for one-dimensional MRW (i.e., \((S_n)_n\) is real-valued). However, the recurrence set has not been investigated for planar MRWs. Therefore, it is not surprising that local limit theorems will play here an important role in the study of the set \(\mathcal{R}_{(\mu,0)}\) for planar MRWs. The LLTs involved in this work are obtained by using the weak Nagaev-Guivarc'h spectral method developed in [\textit{L. Hervé} and \textit{F. Pène}, Bull. Soc. Math. Fr. 138, No. 3, 415--489 (2010; Zbl 1205.60133)]. The authors state the main results and give applications. First, they state two key theorems (Theorems I and II) giving \(\mathcal{R}_{(\mu,0)}=\mathbb{S}\) under conditions related to LLTs. Second, they state two theorems (Theorems III and IV) giving these local limit conditions, under general assumptions, as soon as \((S_n)_n\) satisfies a (standard or non-standard) central limit theorem with suitable normalization as well as a non-sublattice condition in \(\mathbb{S}\). The authors illustrate their general results with classes of models. In Section 4, the authors prove Theorems I and II using classical arguments derived from the i.i.d.\ case [\textit{R. Durrett}, Probability. Theory and examples. Pacific Grove, CA: Wadsworth \& Brooks/Cole Advanced Books \& Software (1991; Zbl 0709.60002)] and the Kochen-Stone adaptation of the Borel-Cantelli lemma [\textit{S. Kochen} and \textit{C. Stone}, Ill. J. Math. 8, 248--251 (1964; Zbl 0139.35401)].