How is it going? Harald wrote this about rho eqn: We hebben het algoritme in sterke vorm gepresenteerd voor de helderheid. Vgl (54) kan direct: rho zit dan in dezelfde FEM space als phi. Vgl (57) doen we via L2 projectie. rho is in same space as phi. Eqn (57) via L2-projection. Check whether phi goes over 1 or below -1: that can cause troubles but should go better on finer grids. ________________________________________ From: firedrake-bounces@imperial.ac.uk <firedrake-bounces@imperial.ac.uk> on behalf of Fryderyk Wilczynski <scfw@leeds.ac.uk> Sent: Wednesday, April 6, 2016 5:43 PM To: firedrake@imperial.ac.uk Subject: Re: [firedrake] Integral is missing an integration domain. Yes, that should be phi0. Thank you. ________________________________________ From: firedrake-bounces@imperial.ac.uk <firedrake-bounces@imperial.ac.uk> on behalf of Lawrence Mitchell <lawrence.mitchell@imperial.ac.uk> Sent: 06 April 2016 17:30 To: firedrake@imperial.ac.uk Subject: Re: [firedrake] Integral is missing an integration domain. On 06/04/16 17:14, Fryderyk Wilczynski wrote:
Thank you, that conditional worked.
I have realized that I should in fact have trial functions, (equations are linear in (n+1) terms), I have been mistakenly following example scripts for nonlinear problems.
I have changed
v, p, phi, mu = split(u)
rho = Function(P1)
to
v, p, phi, mu = TrialFunctions(W)
rho = TrialFunction(P1)
You have a term in a3: inner(chi, phi*grad(mu))*dx but phi and mu are both trial functions. Did you mean phi0? Lawrence _______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake