> The exact solution is > > (u,v) = (-y,x) > > p(x,y) = 0 This means it is completely in the ansatz space and the errors should all be more or less zero (and should not depend on the refinement). -- Timo Heister http://www.math.clemson.edu/~heister/