So we don't have good facilities for modifying the system in between Newton steps. If the only thing that changes in the form are the values of the coefficients, you can just write things in the natural
way.
However, you can't take a newton step, then (say) update an out-of-band equation of state and take another newton step easily. If you just need to update the coefficients in between solves, the that's easy.
The former I can realize with Constant() and assign, I guess. But for the latter, do I have to reassemble the form?
You need to do somewhat more than that, since you're solving for a new unknown, you'll need to use a different form entirely, no? Or do I misunderstand?
[Buesing, Henrik] I think, I need a new form yes. The unknowns stay the same. I'm solving for pressure and enthalpy, whether in the two-phase or single-phase region. But in the single-phase region one of the coefficients turns into an unknown.
[Buesing, Henrik] So, can I have a coefficient c(p,h), which is a nonlinear function of my primary variables and then for some h<=hmin or h>=hmax it turns into c(p,h)=h, where h is an unknown? Henrik
Lawrence -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux)
iQEcBAEBAgAGBQJWL0mVAAoJECOc1kQ8PEYvMEMIAOt1HRpkc+1HmUGhw+xpnVIC BT0B7tgrSRGj3i8/hR1sPKPMJwNFWiiNFITK2KSe2N6uqrvWmH17cOKvru0FAs8L OfWAdOFkeWHCl2eBuuVMyEnCy9LOO5SLwttTl6jePLFM+ROgh+/4VOkPhPtJ6Bjg 3JwOl15ACIWQSK2zxbjJvyZIYSVjAyKFpMPJmLEXkKdDkkCeouIREnfD/tWMrynz hQabk0RHJ6hg3JNPz3vq/aC7Gv7P0KDkVwSTtWWsuYYjfGkkR7WQ3wxGfFtG0C0M OVHE12HP0S1ibPS709nCMdgeWSqjHFoVSKgHCOBeUfG9JyFc6V1zZ75UJzwu3S4= =2xqt -----END PGP SIGNATURE-----
_______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake
_______________________________________________ firedrake mailing list firedrake@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/firedrake