Competing Species near a Degenerate Limit (Q4429894)

Consider the semilinear parabolic system \[ \begin{aligned} u_t&=\mu\Delta u+u(\alpha(x)-u-v),\\ v_t&=\mu\Delta v+u(\beta(x)-u-v), \tag{1}\\ \frac{\partial u}{\partial n}| _{\partial\Omega}&=\frac{\partial v}{\partial n}| _{\partial\Omega}=0\end{aligned} \] in \(\Omega\times (0,\infty)\), where \(\Omega\) is a bounded region in \({\mathbb R}^N\) with smooth boundary \(\partial\Omega\); \(\mu>0\) is the diffusion rate. Let \(\alpha(x):=\beta(x)+\tau g(x)\) and treating both \(\tau\) and \(\mu\) as parameter. Under a fairly weak hypothesis, the semitrivial solutions \((\widetilde u,0)\) and \((0,\widetilde v)\) of (1) exist. It is shown in \textit{V. Hutson} et al. [Dynamical systems and applications. Singapore: World Scientific. World Sci. Ser. Appl. Anal. 4, 343--358 (1995; Zbl 0848.35057)] that if \(\int_{\Omega}\alpha> \int_{\Omega} \beta>0\), then for large \(\mu\), \((\widetilde u,0)\) is a global attractor and \((0,\widetilde v)\) is unstable. On the other hand, they also demonstrate that if \(\alpha_+-\beta_+\) (\(\alpha_+\) denotes the positive part of \(\alpha\)) changes sign, then for small enough \(\mu\) both semitrivial solutions \((\widetilde u,0)\) and \((0,\widetilde v)\) are unstable, thus there is at least one stable coexistence state \((u^*,v^*)\) with \(u^*>0,v^*>0\). These results suggest that: as \(\mu\) decreases from a large value, a branch of coexistence states of (1) bifurcates from \((\widetilde u,0)\) at some value \(\mu_0\). Moreover, for reasonable choice of \(\alpha\) and \(\beta\), as \(\mu\) decreases continuously, more bifurcations will appear which cause \((0,\widetilde v)\) to be stable at an interval of \(\mu\) and then another branch of coexistence states of (1) will bifurcate at \(\mu_1\). This could happen many times as \(\mu\) varies. This article provides a delicate analysis which describes the stability of these solutions as a function of \(\mu\) and \(\tau\). In case \(\tau\) is small, the authors provide a fairly complete characterization of the stability in terms of the zeros of a single function \[ G(\mu)=\begin{cases} \int_{\Omega}g((x)\widetilde v^2(x,\mu)\,dx &\text{if }\mu>0,\\ \int_{\Omega}g((x)\beta_+^2(x)\,dx&\text{if }\mu=0. \end{cases} \] In particular, they show that for any fixed number \(n\), one can choose the difference function \(g\) from an open set of possibilities in such a way that the stability of the semitrivial solutions changes at least \(n\) times as \(\mu\) is varied over \((0,\infty)\). This result allows us to make conclusions concerning the existence and uniqueness of coexistence states. Furthermore, they show that there is no optimal form of resource utilization. In case \(\tau\) is large, the authors also provide a theorem about the range of \(\mu\) for which \((\widetilde u,0)\) changes stability.