Lotka-Volterra-type kinetic equations for interacting species (Q6951017)

The authors start from the Lotka-Volterra dynamics described by the pair of differential equations: \(\frac{dX}{dt}(t)=\alpha X(t)-\beta X(t)Y(t)\), \( \frac{dY}{dt}(t)=-\delta Y(t)+\gamma X(t)Y(t)\), where \(X(t)\geq 0\) and \( Y(t)\geq 0\) are the densities of preys and predators at time \(t\geq 0\), \( \alpha >0\) is the Malthusian growth rate and \(\beta >0\) the predation rate of the preys' population, \(\delta >0\) the death rate of the predators' population, and \(\gamma >0\) its growth due to predation. They introduce a system of Boltzmann-type kinetic equations to study the time evolution of the size distribution densities \(f_{1}(x,t)\) and \(f_{2}(y,t)\) of two interacting populations with respective sizes \(x,y\in \mathbb{R}_{+}\) and following the rules of a predator-prey system. They assume that \(\int_{\mathbb{R}_{+}}f_{1}(x,0)dx=\int_{\mathbb{R}_{+}}f_{2}(y,0)dy=1 \) and that the number of agents/particles is infinite. The Boltzmann-type equations are written as: \(\frac{\partial f_{1}}{\partial t}(x,t)=R_{\chi }^{\alpha }(f_{1})(x,t)+Q_{12}(f_{1},f_{2})(x,t)\), \(\frac{\partial f_{2}}{ \partial t}(y,t)=R_{\theta }^{\nu }(f_{2})(y,t)+Q_{21}(f_{2},f_{1})(y,t)\), where \(R_{\chi }^{\alpha }(f_{1})\) and \(R_{\theta }^{\nu }(f_{2})\) are redistribution-type linear operators describing the balance of microscopic interactions, and \(Q_{12}(f_{1},f_{2})\) and \(Q_{21}(f_{2},f_{1})\) are Boltzmann-type bilinear operators accounting for microscopic interactions. They derive the associated kinetic model. After scaling, they end up with the system of Fokker-Planck-type equations: \(\frac{\partial f_{1}}{\partial t }(x,t)=\frac{\sigma _{1}m_{2}(t)}{2}\frac{\partial ^{2}}{\partial x^{2}} (x^{2p}f_{1}(x,t))+\frac{\partial }{\partial x}(((\beta m_{2}(t)+\alpha \chi )x-\alpha (\chi +1)m_{1}(t))f_{1}(x,t))\), \(\frac{\partial f_{2}}{\partial t} (y,t)=\frac{\sigma _{2}m_{1}(t)}{2}\frac{\partial ^{2}}{\partial y^{2}} (y^{2p}f_{2}(x,t))+\frac{\partial }{\partial y}(((\gamma (\mu -m_{1}(t))y+\nu \theta y-\nu (\theta +1)m_{2}(t))f_{2}(y,t))\), for any \( 0<p\leq 1\). They add the no-flux boundary conditions: \(\frac{\sigma _{1}m_{2}(t)}{2}\frac{\partial }{\partial x}(x^{2p}f_{1}(x,t))+((\beta m_{2}(t)+\alpha \chi )x-\alpha (\chi +1)m_{1}(t))f_{1}(x,t)\mid _{x=0,+\infty }=0\), \(\frac{\sigma _{2}m_{1}(t)}{2}\frac{\partial }{\partial y }(y^{2p}f_{2}(x,t))+((\gamma (\mu -m_{1}(t))y+\nu \theta y-\nu (\theta +1)m_{2}(t))f_{2}(y,t)\mid _{y=0,+\infty }=0\). They recall that the well-posedness of this system can be assessed by applying the results from the work by \textit{C.L. Le Bris} and \textit{P.L. Lions} in [Commun. Partial Differ. Equations 33, No. 7, 1272--1317 (2008; Zbl 1157.35301)]. They determine the local equilibrium states of the Fokker-Planck system and they describe the asymptotic behavior of the associated moments through numerical computations. Observing that the underlying macroscopic dynamics exhibit an oscillatory behavior around the equilibrium state, they finally introduce a correction to the preceding kinetic modeling, switching from a Malthusian-type growth for the preys to a Verhulst-type one, to obtain a Lotka-Volterra system with a unique equilibrium state that serves as an attractor for the macroscopic dynamics. They again draw numerical computations.