Showing 1–3 of 3 results for author: Högnäs, G

Exit times for some nonlinear autoregressive processes

Authors: Göran Högnäs, Brita Jung

Abstract: By using the large deviation principle, we investigate the expected exit time from the interval [-1,1] of a process of autoregressive type. The case when the autoregression function f is linear and the innovations have a normal distribution has been treated before. In this paper, we extend the results to more general functions f, with the main focus on piecewise linear functions. By using the large deviation principle, we investigate the expected exit time from the interval [-1,1] of a process of autoregressive type. The case when the autoregression function f is linear and the innovations have a normal distribution has been treated before. In this paper, we extend the results to more general functions f, with the main focus on piecewise linear functions. △ Less

Submitted 18 December, 2019; originally announced December 2019.

The fine structure of the stationary distribution for a simple Markov process

Authors: Andreas Anckar, Göran Högnäs

Abstract: We study the fractal properties of the stationary distrubtion π for a simple Markov process on R. We will give bounds for the Hausdorff dimension of π, and lower bounds for the multifractal spectrum of π. Additionally, we will provide a method for numerically estimating these bounds. We study the fractal properties of the stationary distrubtion π for a simple Markov process on R. We will give bounds for the Hausdorff dimension of π, and lower bounds for the multifractal spectrum of π. Additionally, we will provide a method for numerically estimating these bounds. △ Less

Submitted 5 October, 2015; v1 submitted 2 December, 2014; originally announced December 2014.

On a stochastic Ricker competition model

Authors: Göran Högnäs

Abstract: We model the evolution of two competing populations $U_t, V_t $ by a two-dimensional size-dependent branching process. The population characteristics are assumed to be close to each other, as in a resident-mutant situation. Given that $U_t = m$ and $V_t = n$ the expected values of $U_{t+1}$ and $V_{t+1}$ are given by $me^{r - K(m + bn)} $ and $ne^{\tilde r - \tilde K (n + am)}$, respectively, wher… ▽ More We model the evolution of two competing populations $U_t, V_t $ by a two-dimensional size-dependent branching process. The population characteristics are assumed to be close to each other, as in a resident-mutant situation. Given that $U_t = m$ and $V_t = n$ the expected values of $U_{t+1}$ and $V_{t+1}$ are given by $me^{r - K(m + bn)} $ and $ne^{\tilde r - \tilde K (n + am)}$, respectively, where $r, \tilde r$ model the intrinsic population growth, $K, \tilde K $ model the force of inhibition on the population growth by the present population (such as scarcity of food), and $ a, b $ model the interaction between the two populations. For small $K, \tilde K $ the process typically follows the corresponding deterministic Ricker competition model closely, for a very long time. Under some conditions, notably a mutual invasibility condition, the deterministic model has a coexistence fixed point in the open first quadrant. The asymptotic behaviour is studied through the quasi-stationary distribution of the process. We initiate a study of those distributions as the inhibitive force $K, \tilde K$ approach 0. △ Less

Submitted 13 February, 2013; originally announced February 2013.

Comments: Presented at ICDEA 2012, Barcelona

MSC Class: 60J80; 60F05; 60F10; 60J10