Page 145 -
P. 145

128  3. Further applications



          the boundary-  (or transition-) layer problem. We now turn to an examination of this
          classical singular perturbation problem in the context of partial differential equations.

          3.3  SINGULAR PROBLEMS  II
          There are two partial differential equation-types that are often encountered with small
          parameters multiplying the highest derivative(s): the elliptic equation (e.g. Laplace’s
          equation) and the parabolic equation (e.g. the heat conduction, or diffusion, equation).
          These two, together with the wave equation (i.e. of hyperbolic type) discussed in §3.2,
          complete the set of the three that constitutes the classification of second-order partial
          differential equations.  The two  new equations are exemplified by




          respectively. Of course, the use of singular perturbation methods to solve these partic-
          ular examples is somewhat redundant, because we are able to solve them exactly (for
               using standard techniques. Thus we will discuss two simple—but not completely
          trivial—extensions of these basic equations.

          E3.3  Laplace’s equation with nonlinearity
          We are going to find an asymptotic solution, for   of the  equation





          where u is prescribed on the boundary of the region. First, we will make a few general
          observations about this problem and then  obtain some  of the details in a particular
          case.
            The appearance  of  in equation (3.44) suggests that we may seek a solution






          which turns  out to be  consistent with  the  matching  requirements to  the  boundary
          layers; thus we obtain





          and so on. Immediately we see that we can find   which satisfies the given data
          on x = 0  and  x = a, but  only in very special  circumstances will this  also  satisfy the
          data on y= 0 and y = b: the solution will (in general) require boundary layers near
          y = 0  and near  y = b. No  such layers exist  near x = 0  and x = a. All this follows
          from the term   the term    simply contributes a (small) nonlinear adjustment
   140   141   142   143   144   145   146   147   148   149   150