Glad you found the problem :) Redefining the form can be a bad idea, since low-level C code will be generated for each new form (this is disastrous if you're in the "small mesh, lots of timesteps" regime!) A better way is to use Constant. For example, Q = Constant(0.0) a = Q*....*dx t = 0.0 while t < T - 0.5*dt: Q.assign(A*sin(2*pi*F*t)) ... t += dt On 1 October 2015 at 14:17, Anna Kalogirou <a.kalogirou@leeds.ac.uk> wrote:
I was just ready to send you a code when I realised what the problem was.
When I define Q as a scalar expression, updating it after every time step does not update the expression in the weak form (I define a and L outside the time loop and then just call the solver at every time step). Defining the expression by
Q_expr = Expression("A*sin(2*pi*F*t)", t=t, A=A, F=F) Q.interpolate(Q_expr)
and updating it by Q_expr.t = t works though.
Therefore I had to re-define the linear and bilinear parts of the weak form inside the time loop for the first approach to work.
On 30/09/15 15:58, Lawrence Mitchell wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 30/09/15 15:51, Anna Kalogirou wrote:
Yes, I double - and triple - checked and I believe I am doing the same thing in both approaches. I don't understand then why the results are different. Hmph, can you show us some breaking code?
Lawrence
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1
iQEcBAEBAgAGBQJWC/imAAoJECOc1kQ8PEYvalEH/2z+lJMa3od/bKUbcbW73imx i3In1Wg9ZKVGeZG8VmVlcnirkr38EgzIi5ZoRZVtGQ6tmWsXTCwojfqsaLUv5k+b fWma1V9FWp8Vk6XSo8eqR+5oV0CBH0KRZtN8sZdGGqBm2QY5PBrQSyYcy5XHunIm h/mc7tT8oLRQKaPwYxuDt8WBtzc+BdNZPexLbGEa921qE1ms/SemDk85GbOQ5oqr oqvqfxpyrzwMBlPjz7GLN697u7r15e+qwGFni0Z6x3YsaBUF1OKXq7k0FvT5ObeI t30lYnhlxxKbbZQh1NI7goAZ7RUEd5RgAge2wUXN3V8EDyKVqzooRso1GGKGAwg= =jk88 -----END PGP SIGNATURE-----
_______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake
--
Dr Anna Kalogirou Research Fellow School of Mathematics University of Leeds
http://www1.maths.leeds.ac.uk/~matak/
_______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake
participants (1)
-
Andrew McRae