Page 76 - 6099
P. 76

We may use the following subtraction device to reduce (7.29) to a simpler problem. Let
                   V (x, t) = v(x, t) − h(t). Then V (x, t) will satisfy

                                                          0
                                  V t − kV xx = f(x, t) − h (t) for 0 < x < ∞, 0 < t < ∞,
                                                        V (0, t) = 0,                              (7.30)
                                                   V (x, 0) = φ(x) − h(0).                         (7.31)

                   To verify (7.30), just subtract! This new problem has a homogeneous boundary condition
                   to which we can apply the method of reflection. Once we find V , we recover v by
                   v(x, t) = V (x, t) + h(t). This simple subtraction device is often used to reduce one linear
                   problem to another.The domain of independent variables (x, t) in this case is a quarter-
                   plane with specified conditions on both of its half-lines. If they do not agree at the corner
                   [i.e., if φ(0) 6= h(0)], then the solution is discontinuous there (but continuous everywhere
                   else). This is physically sensible . Think for instance, of suddenly at t = 0 sticking a hot
                   iron bar into a cold bath.
                       For the inhomogeneous Neumann problem on the half-line,

                                      w t − kw xx = f(x, t) for 0 < x < ∞, 0 < t < ∞,
                                                      w x (0, t) = h(t),                           (7.32)
                                                      w(x, 0) = φ(x),                              (7.33)

                   we would subtract off the function xh(t). That is, W(x, t) = w(x, t) − xh(t). Differen-
                   tiation implies that W x (0, t) = 0.


                   7.5      Diffusion revisited


                   In this section we make a careful mathematical analysis of the solution of the diffusion
                   equation that we found in previous sections. (On the other hand, the formula for the
                   solution of the wave equation is so much simpler that it does not require a special justi-
                   fication.)
                       The solution formula for the diffusion equation is an example of a convolution, the
                   convolution of φ with S (at a fixed t). It is
                                           Z  ∞                     Z  ∞
                                 u(x, t) =     S(x − y, t)φ(y)dy =       S(z, t)φ(x − z)dz,        (7.34)
                                            −∞                        −∞
                                      √         2                                        √
                   where S(z, t) = 1/ 4πkte   −z /4kt . If we introduce the variable p = z/ kt, it takes the
                   equivalent form
                                                        Z  ∞
                                                     1        −p /4        √
                                                                 2
                                         u(x, t) = √         e     φ(x − p kt)dp.                  (7.35)
                                                     4π   −∞
                   Now we are prepared to state a precise theorem.

                   Theorem 7.5.1 Let φ(x) be a bounded continuous function for−∞ <
                   x < ∞. Then the formula (7.35) defines an infinitely differentiable
                   function u(x, t) for −∞ < x < ∞, 0 < t < ∞, which satisfies the

                   equation u = ku      xx  and lim  t→+0  u(x, t) = φ(x) for each x.
                                t


                                                             69
   71   72   73   74   75   76   77   78   79   80   81