Asymptotic properties of quadratic stochastic operators acting on the \(L^1\) space (Q2255019): Difference between revisions

The authors consider the \(L^1\)-space over a separable \(\sigma\)-finite measure space and study quadratic stochastic operators on this space, i.e., bilinear operators \(\boldsymbol Q: L^1 \times L^1 \to L^1\) which fulfil \(\boldsymbol Q(f,g) = \boldsymbol Q(g,f) \geq 0\) and \(\|\boldsymbol Q(f,g)\|_1 = \|f\|_1 \|g\|_1\) for all \(0 \leq f,g \in L^1\). For each normalized function \(0 \leq f \in L^1\), they define \(Q(f) := \boldsymbol Q(f,f)\). The asymptotic behaviour of the iterates \(Q^n(f)\) as \(n \to \infty\) is important in many applications which stem for example from mathematical biology. \textit{N. N. Ganikhodzhaev} observed in [J. Theor. Probab. 4, No. 4, 639--653 (1991; Zbl 0744.60075)] that there is a connection between quadratic stochastic operators and linear nonhomogeneous Markov chains. In the paper under review, the authors use these associated Markov chains to characterize the asymptotic behaviour of the iterates \(Q^n(f)\) in several topologies. On \(\ell^1\)-spaces, similar results were obtained before in [\textit{W. Bartoszek} and the second author, Nonlinear Anal., Theory Methods Appl., Ser. A, Theory Methods 86, 95--113 (2013; Zbl 1298.47069)]. However, the situation on \(L^1\) is more involved since not every operator is a kernel operator. This is why the authors restrict their attention in most (not all) of their theorems to kernel quadratic stochastic operators.