Weak Extinction Versus Global Exponential Growth of Total Mass for Superdiffusions
Janos Englaender,Yan-Xia Ren,Renming Song
DOI: https://doi.org/10.1214/14-aihp645
2016-01-01
Abstract:Consider a superdiffusion X on R-d corresponding to the semi-linear operator A(u) = Lu + beta u - ku(2), where L is a second order elliptic operator, beta(center dot) is in the Kato class, and k(center dot) >= 0 is bounded on compact subsets of R-d and is positive on a set of positive Lebesgue measure.The main purpose of this paper is to complement the results obtained in (Ann. Probab. 32 (2004) 78-99), in the following sense. Let lambda(infinity) be the L-infinity-growth bound of the semigroup corresponding to the Schrodinger-type operator L + beta. If lambda(infinity) not equal 0, then we prove that, in some sense, the exponential growth/decay rate of parallel to X-t parallel to, the total mass of X-t, is lambda(infinity). We also describe the limiting behavior of exp(-lambda(infinity)t)parallel to X-t parallel to, as t -> infinity. This should be compared to the result in (Ann. Probab. 32 (2004) 78-99), which says that the generalized principal eigenvalue lambda(2) of the operator gives the rate of local growth when it is positive, and implies local extinction otherwise. It is easy to show that lambda(infinity) >= lambda(2), and we discuss cases when lambda(infinity) > lambda(2) and when lambda(infinity) = lambda(2).When lambda(infinity) = 0, and under some conditions on beta, we give a necessary and sufficient condition for the superdiffusion X to exhibit weak extinction. We show that the branching intensity k affects weak extinction; this should be compared to the known result that k does not affect weak local extinction. (The latter depends on the sign of lambda(2) only, and it turns out to be equivalent to local extinction.)