Page 65 - 6099
P. 65

integrating first over the horizontal line segments in Figure 5 and then

                   vertically.
                       A variant of the method of characteristic coordinates is to write
                   (6.10) as the system of two equations

                                             u + cu = v, v − cv = f,
                                                                 t
                                                      x
                                               t
                                                                        x
                   the first equation being the definition of v, as above. If we first solve
                   the second equation, then v is a line integral of f over a characteristic
                   line segment x + ct = constant. The first equation then gives u(x, t) by
                   sweeping out these line segments over the characteristic triangle ∆. To
                   carry out this variant is a little tricky, however, and we leave it as an

                   exercise.


                   6.6.1     Source on a half-line

                   The solution of the general inhomogeneous problem on a half-line
                                                      2
                                        DE : v tt − c v xx = f(x, t) in 0 < x < ∞,
                                            IC : v(x, 0) = φ(x) v t (x, 0) = φ(x)                  (6.17)
                                                    BC : v(0, t) = h(t)

                   is the sum of four terms, one for each data function φ, ψ, f, and h. For x > ct > 0, the
                   solution has precisely the same form as in (6.12), with the backward triangle ∆ as the
                   domain of dependence. For 0 < x < ct, however, it is given by
                                                                                ZZZZ
                                                                        x
                                   v(x, t) = φ term + ψ term + h(t − ) +      1         f          (6.18)
                                                                        c    2c
                                                                                      D
                   where t − x/c is the reflection point and D is the shaded region in Figure 6.5
                       The only caveat is that the given conditions had better coincide at the origin.That is,
                                                             0
                   we require that φ(0) = h(0) and ψ(0) = h (0). If this were not assumed, there would be
                   a singularity on the characteristic line emanating from the corner.
                       Let’s derive the boundary term h(t − x/c) for x < ct. To accomplish this, it is
                   convenient to assume that φ = ψ = f = 0. We shall derive the solution from scratch
                   using the fact that v(x, t) must take the form v(x, t) = j(x + ct) + g(x − ct). From
                   the initial conditions (φ = ψ = 0), we find that j(s) = g(s) = 0 for s > 0. From the
                   boundary condition we have h(t) = v(0, t) = g(−ct) for t > 0. Thus g(s) = h(−s/c) for
                   s < 0. Therefore, if x < ct, t > 0, we have v(x, t) = 0 + h(−(x − ct/c)) = h(t − x/c).


                   6.6.2     Finite interval

                   For a finite interval (0, l) with inhomogeneous boundary conditions v(0, t) = h(t), v(l, t) =
                   k(t), we get the whole series of terms

                                                    x
                                     v(x, t) = h(t − ) − h(t +  x−2l ) + h(t −  x+2l  ) + . . .
                                                    c            c             c

                                                             58
   60   61   62   63   64   65   66   67   68   69   70