Singularly perturbed problems in case of exchange of stabilities (Q1780283)

This survey paper is devoted to a class of singularly perturbed problems whose solutions have angular limits. Such a behavior of solutions, which is due to exchange of stabilities in boundary layer equations, was already noticed by \textit{S. Haber} and \textit{N. Levinson} [Proc. Am. Math. Soc. 6, 866--872 (1955; Zbl 0066.06503)] half a century ago. Also, it turned out that the use of lower and upper solutions, with or without angles, was well-adapted to this problem. This method was used extensively by \textit{F. Howes} [Trans. Am. Math. Soc. 241, 155--182 (1978; Zbl 0389.34039)]. The authors of this survey and their co-workers continued in this direction. The present paper presents an overview of their work. They worked out the lower and upper solutions method for a large variety of singular perturbation problems with angular limit solution. These concern both 1st- and 2nd-order problems. Such results complete our knowledge of singular perturbations. It is also of interest to the lower and upper solutions method since it shows how such functions can be builded in case of singular perturbations. The first section considers an initial value problem for the ODE \[ \varepsilon u'=g(u,v,t,\varepsilon), \quad u(0)=u^0, \quad v'=f(u,v,t,\varepsilon), \quad v(0)=v^0, \] with \(u\) and \(v\in \mathbb{R}\). The second section considers the problem \[ \varepsilon u'=g(u,t,\varepsilon),\quad u(0)=u^0, \] in case the jump from one branch of the angle to the other is delayed. This creates a boundary layer and is typical of the so-called ``Canards''. The third and fourth sections concern boundary value problems such as \[ \varepsilon^2u''-g(u,x,\varepsilon)=0,\quad u'(0)=u'(1)=0, \] \[ \varepsilon^2 u''=g(u,v,x,\varepsilon),\quad u'(0)=u'(1)=0, \quad v'=f(u,v,x,\varepsilon),\quad v(0)=v^0, \] or \[ \varepsilon^2 u''=g(u,v,t,\varepsilon),\quad u'(0)=u'(1)=0, \quad v''=f(u,v,t,\varepsilon),\quad v(0)=v^0, v(1)=v^1, \] with \(u\) and \(v\in \mathbb{R}\). Section 5 deals with the scalar parabolic problem \[ \varepsilon^2(u_t-u_{xx})=g(u,x,t,\varepsilon), \quad u_x(0,t)=u_x(1,t)=0,\quad u(x,0)=u_0(x). \] Delayed jump from one branch of the angle to the other is considered in Section 6. Reaction-diffusion systems which are partly dissipative such as \[ \varepsilon^2 (u_t-u_{xx})=g(u,v,x,t,\varepsilon),\quad u_x(0,t)=u_x(1,t)=0,\quad u(x,0)=u_0(x), \] \[ v_t=f(u,v,x,t,\varepsilon),\quad v(x,0)=v_0(x), \] are studied in Section 7. At last, Section 8 considers elliptic problems \[ \textstyle \varepsilon^2 \nabla u=g(u,x,\varepsilon)\quad \text{in }\Omega, \quad \frac{\partial u}{\partial n}(x)-q(x)u(x)=0 \quad \text{on }\partial\Omega. \] Proofs are carried out in detail and each problem is illustrated by appropriate examples.