Hello Gentlemen,
I have attached gatech.tar, which contains a 2D code based on my paper
specnse.f - 2D Navier-Stokes spectral solver (Fourier-Chebyshev)
myfftD.f - required VFFTPK routines
mylapack.f - required LAPACK routines
As before, the code is completely self-contained. Note that the variables
are advanced in time Fourier transformed in the x-direction. We only have
to transform back to physical space to compute the nonlinear convection
terms at each time step.
Please take a look and feel free to ask any questions. I have sprinkled
the code with some comments, but not extensively. Note that when one
takes a derivative spectrally in the periodic direction the 1st and nth
coefficients are set to 0! This is why these cases are distinguished.
The 1st coeff is for the constant, and the nth is the highest mode
which I simply set to 0.
The main change to come is computing the Chebyshev differentiation matrices,
and their required decompositions for the Poisson solvers, in quadruple
precision to avoid significant roundoff errors in the computation of the
evals and evectors for, say, N >= 100. This will be done in a stand alone
code, and the numbers stored.
Again, any questions, just ask!
Hans