Comment on "Calculation of Turbulent Boundary-Layer Shock-Wave Interaction"

simply means that odd ordered variations in the energy, of higher order than two, vanish at P = Pcr. How does this affect convergence? To answer this question, consider a (potential energy) functional K(u) expanded about some function v in terms of Gateaux differentials at u0 K(v) = K(u0) + 6K(u0,n)+l/2(u0,n,n)+... (1) Here n = v— u0 and u,v may be vectorvalued. Let u0 = exact variational solution of the problem ; U = finite-element (Galerkin) approximation of the solution; and U = an arbitrary element in the same subspace of approximants containing U. Suppose that the postbuckling path is stable. Then