Numerical integration of fully nonlinear size-structured population models (Q1878384)

The paper is devoted to the numerical approximation and subsequent numerical analysis of a nonlinear partial differential equation (PDE) model describing the evolution of a size-structured population. The PDE model is as follows: \[ u_t + (g(x,I_g(t),t))u)_x = -\mu(x,I_\mu(t),t)u,\quad 0<x<1,\quad t>0, \] and it is supposed to be subject to the nonlocal boundary condition \[ g(0,I_g(t),t)u(0,t) = \int_0^1 \alpha(x,I_\alpha(t),t) u(x,t)\;dx,\quad t>0, \] and to the initial condition \(u(x,0)=u_0(x)\), \(0\leq x\leq 1\). The internal variable \(x\) is formally named ``size'', but it can also stand for length, weight, mass, maturity or some other characteristic of the population, depending on the physical system that one wants to model. The maximum size that individuals may reach has been normalized to one. The unknown \(u(x,t)\) is the population of size \(x\) at time \(t\). It is assumed that \[ dx/dt= g(x,I_g(t),t) \] and that the maximum size (\(x=1\)) cannot be reached by any individual: \(u(1,t)=0\). \(t>0\). The quantities \(I_\phi\), for \(\phi\in \{\mu,\alpha,g\}\), are given by \[ I_\phi(t)=\int_0^1 \gamma_\phi(x) u(x,t)\;dx,\quad t\geq 0. \] The vital functions \(g\), \(\alpha\) and \(\mu\) stand for the growth, the fertility and mortality rates. Those three functions, as well as \(u_0\), \(\gamma_\mu\), \(\gamma_\alpha\) and \(\gamma_g\), are data. The key starting point is to rewrite the continuous problem in a convenient way, thus transforming the original problem into the equations (2.2) and (2.5) of the paper. After doing that, the authors introduce two numerical methods based on integration along characteristic curves and carry out their convergence analysis. Under suitable hypotheses, those methods are proved to be of order two. Both schemes are two-step methods. In the final section, numerical results are shown.