- Problem 6.3b: GB2=zero(Uy-Vx) seems to work fine for some triangulations,
but fails for others. (It always seems to work for 1D Galerkin problems).
The PDE2D documentation does not suggest including
*derivatives*of the solution inside the zero function (e.g., zero(Uy-Vx)= big_number*(Uy-Vx)), though it often works: here is one program for this problem that works. The free-slip boundary conditions given in Problem 1, Chapter 5, seem to be more reliable for 2D Galerkin problems.