Aggregation in age and space structured population models: an asymptotic analysis approach (Q657016)

This paper deals with the equation \[ \mathbf n_t=-\mathbf n_a-\mathcal M\mathbf n+\frac{1}{\epsilon}\mathcal C\mathbf n,\tag{1} \] where \(\epsilon\) is a small parameter, \(\mathbf n(t,a)=(n_1(t,a),\dots,n_N(t,a))\), and \(n_i(t,a)\) is the population density, at time \(t\), of individuals residing in patch \(i\) and being of age \(a\). The matrix \(\mathcal M(a)=\{\mu_i(a)\}_{1\leq i\leq N}\) is the mortality matrix, and the \(N\times N\)-matrix \(\mathcal C (a)\) describes the transfer of individuals between patches. The McKendrick boundary condition \[ \mathbf n(t,0)=\int_0^{\infty}B(a)\mathbf n(t,a)da, \] where the \(N\times N\) matrix \(B(a)\) is the fertility matrix, and the initial condition \(\mathbf n(0,a)=\mathbf n_0(a)\) are associated to (1). It is assumed that \(\lambda=0\) is a dominant simple eigenvalue of \(\mathcal C(a)\) for each \(a\in \mathbb R_+\) with a corresponding normalized positive right eigenvector denoted by \(\mathbf k(a).\) Set \[ n=\sum _{i=1}^N n_i \] The \` limit\'-problem, called the aggregated model, is \[ n_t=-n_a- \mu^* n \] where \(\mu^*=\mathbf 1\cdot\mathcal M\mathbf k\) is the aggregated mortality, associated with the boundary condition \[ n(t,0)=\int_0^{\infty}(\mathbf 1\cdot B\mathbf k)(a)n(t,a)da. \] The following decomposition is used: \(\mathbf n=n\mathbf k+\mathbf w\), \(\mathbf n_0(a)=n_0(a)\mathbf k(a)+\mathbf w_0(a)\), where \(\mathbf w_0(a)=\mathbf w(0,a).\) Let \(\bar n\) be the solution of the aggregated problem. The main result states the existence of a constant \(C=C(T,\mathcal M, B, \mathcal C)\) such that, for any time interval \([0,T]\), as \(\epsilon \to 0\), \[ \|n(t,\cdot)-\bar n(t,\cdot)\|_{L_1({\mathbb R}_+)}\leq\epsilon\,C\,\|\mathbf n_0\|_{W_1^1(\mathbb R_+,\mathbb R^N)} \] and \[ \left\|\mathbf w(t,\cdot)-e^{\frac{1}{\epsilon}\mathcal C(\cdot)}\mathbf w_0(\cdot)\right\|_{L_1(\mathbb R_+,\mathbb R^N)}\leq \epsilon\, C\|\mathbf n_0 \|_{W_1^1(\mathbb R_+,\mathbb R^N)} \] uniformly in \(t\in[0,T].\) The proof is based on a careful analysis of the boundary, corner, and initial layers.