Limit theorems for bifurcating integer-valued autoregressive processes (Q2339215): Difference between revisions

This paper studies the asymptotic behavior of the weighted least squares (WLS) estimators of the parameters in first-order bifurcating integer-valued autoregressive processes. These processes are an adaptation of integer-valued autoregressive processes to binary tree structured data. The proofs utilize asymptotic results for vector-valued martingales, modifying techniques developed in [\textit{B. Bercu} et al., Electron. J. Probab. 14, 2492--2526 (2009; Zbl 1190.60019)] for least squares estimators in bifurcating autoregressive processes. Sufficient conditions on the immigration process are provided to ensure the almost sure convergence of the weighted least squares estimators. The almost sure rate of convergence, a quadratic strong law and asymptotic normality of the WLS estimators are established. Furthermore, the WLS variance and covariance estimators are shown to converge almost surely.