Asymptotic Solutions of Initial Value Problems for Nonlinear Partial Differential Equations
Asymptotic Solutions of Initial Value Problems for Nonlinear Partial Differential Equations
复制标题
非线性偏微分方程初值问题的渐近解
DOI:
10.1137/0118067
复制
发表时间:
1970
期刊:
影响因子:
--
通讯作者:
S. Kogelman
中科院分区:
文献类型:
--
作者:
J. Keller;S. Kogelman
1. Introduction. Let us consider the solution u (x, t, e) of an initial value problem for a nonlinear partial differential equation in which the nonlinearity is proportional to a parameter e. We wish to determine the asymptotic behavior of the solution when e is small and is large. The usual perturbation method, in which u is sought as a power series in e, is not applicable because it leads to secular terms and is therefore useless for large t. Instead we shall use the twotime method, which has been applied to ordinary differential equations by Cole and Kevorkian and others (see Cole [1]) and to statistical mechanics by Frieman, Sandri and others (see Frieman [2]). An alternative closely related procedure, the averaging method of Bogolyubov and Mitropol’skii, might have been used instead.In the analysis of periodic solutions of ordinary differential equations, Lindstedt and Poincar6 showed that secular terms could be eliminated by ex-panding the period of the solution, as well as the solutionitself, in powers of e. Their method has been adopted to partial differential equations by Keller and Ting [3], Millman and Keller [4] and others. The problem we consider is an initial boundary value problem for a nonlinear wave equation. Our detailed results are obtained for a particular equation which may be viewed as a generalized van der Pol equation. The boundary value problem for this equation possesses infinitely many periodic solutions or limit cycles, asymptotic expansions of which were found by Millman and Keller [4]. We shall see that in general the solution of the initial value problem tends to one of these periodic solutions, and for given initial data we shall show which one. However, for certain special initial conditions, the solution tends toa combinationof a finite number ofperiodic solutions. Thesespecial solutions are unstable, and only the individual periodic solutionsare stable. In a final sectionwe show how the two-time method can beapplied to a nonlinear initial value problem for a vector in a Hilbert space.