Some results on the Lotka--Volterra model and its small random perturbations (Q1415878)

Consider the Lotka-Volterra system \[ \dot{x}(t) = a x(t) - b x(t) y(t) ,\quad \dot{y}(t) = -c y(t) + d x(t) y(t), \] for competing species with prey \(x(t)\) and predator sizes \(y(t)\) at time \(t \geq 0\), where \(x(0)=x_0 >0\), \(y(0)=y_0>0\) and \(a,b,c,d\) are positive constants. The authors study effects of random perturbations of parameters \(a\) and \(c\) in form of Stratonovich-type ordinary stochastic differential equations \[ dX^\varepsilon(t) = X^\varepsilon(t) (a-bY^\varepsilon(t)) dt + \varepsilon \sigma_1 X^\varepsilon(t) \circ dW_1(t), \] \[ dY^\varepsilon(t) = Y^\varepsilon(t) (-c+dX^\varepsilon(t)) dt+ \varepsilon \sigma_2 Y^\varepsilon(t) \circ dW_2(t), \] driven by standard, independent Wiener processes \(W_1\) and \(W_2\). First, they show that, when the deterministic system is far from its equilibrium, then one of the populations is exponentially small most of the time. Second, they apply averaging principles such as large deviation techniques. In case of extinction of one of the populations, the technique of large deviations is used to find the likely path of extinction. One may note that the first integral of the Lotka-Volterra system is known. However, as the authors also claim, the first integral process explodes at the time of extinction, hence asymptotic techniques of \textit{M. Freidlin} and \textit{M. Weber} [Ann. Probab. 26, 925-967 (1998; Zbl 0935.60038)] do not seem to work here. The paper is a continuation of that of the second and the third author [Ann. Appl. Probab. 11, 1263--1291 (2001; Zbl 1015.92044)].