Let . For an exact smooth solution of the diffusion equation, the second-order central difference gives
Since the Courant number is fixed, . Taylor expansion in time gives
Using and cancels the leading terms. Because , the residual, and hence the local truncation error in the convention of the question, is
For a discrete Fourier mode , put
The recurrence becomes
so its amplification polynomial of a multilevel finite difference scheme is
Its roots are real. Moreover,
The root-location criterion in the question therefore puts both roots in exactly when . Fourier stability analysis for every wavenumber is equivalent to , so
For the same Fourier mode, the leapfrog finite-difference scheme for the diffusion equation gives
The two amplification roots are
For every nonzero mode , one has
Thus every admits a growing Fourier mode, and

Articles by others on the same topic (0)

There are currently no matching articles.