Use the Fourier transform . For a rapidly decaying smooth function, integration by parts gives , and differentiation under the integral gives . Both are justified here by Gaussian decay. Since ,
The Gaussian integral supplied in the hint gives . Solving this first-order equation yields
The calculation states the transform normalization and the two differentiation properties explicitly.
Taking the spatial Fourier transform of the diffusion equation gives and . Hence . Fourier inversion and the Gaussian transform from part (a), or direct evaluation of the inverse Gaussian integral, give the heat kernel
Substituting the integral for and interchanging absolutely integrable terms produces the convolution solution
For the stated nonnegative compactly supported data, understood to be nonzero on a set of positive measure, the kernel is strictly positive everywhere. Therefore for every real and every . Property P holds: diffusion has infinite propagation speed. The distant value can be extremely small, but is not zero. For continuous initial data, any nontrivial nonnegative automatically has the required positive-measure set.
For the wave equation, introduce characteristic coordinates , when . The equation becomes , so . The initial conditions imply and . Constants can be absorbed into , giving the D'Alembert formula
If vanishes outside , then whenever . Thus Property P fails: the wave equation has finite propagation speed . No arbitrarily short time can transmit the disturbance arbitrarily far. If , the zero initial velocity instead gives , and the conclusion still holds.

Articles by others on the same topic (0)

There are currently no matching articles.