From: KIRK, BENJAMIN (JSCEG) (NASA) <benjamin.kirk1@na...>  20050119 15:25:22

Excellent! Will this be the default behavior and then let specific finite elements (e.g. LAGRANGE) overload or reimplement the method? The L2 projection method is perfect since it addresses the problem for arbitrary element types. In fact, it can be extended to use the normal equations to handle p mismatch on an edge as well. In general (see ~benkirk/Documents/lab_presentations/amr/slides.pdf in the CFDLab) A u_c = B u_p u_c = (A^1 B) u_p u_c = C u_p There is no assumption about corner DOFs in libMesh, the implementation in fe_lagrange.C just recognizes that for Lagrangetype elements A=I, so u_c = C u_p directly. Line 622 in fe_lagrange.C simply says keep constraints that look like this: u_1 = 0.5*u_2 + 0.5*u_3 and ignore ones that look like this: u_1 = 0.9999 * u_1 + 1.e6 * u_2 + 1.e6 * u_3 Which we can recognize as u_1 = u_1, which is not a constraint at all... So, for the general case the constraint matrix C = (A^1 B) will have some rows that are like the identity matrix (within floatingpoint error, of course) which should be ignored and other rows that are valid DOF constraints. Clear? Ben Original Message From: libmeshdeveladmin@... [mailto:libmeshdeveladmin@...] On Behalf Of Roy Stogner Sent: Monday, January 17, 2005 9:30 PM To: libmeshdevel@... Subject: [Libmeshdevel] compute_constraints I'm writing a "fallback" implementation of compute_constraints; without knowing the details of every single degree of freedom it should be possible to use local L2 projections (of face values for C0 elements like the Hierarchic, face values and fluxes for C1 elements like the CloughTocher) to generate correct constraint matrices. One concern I have: is libMesh assuming that corner degrees of freedom and edge/face degrees of freedom should be equal when one element's corner node is a less refined element's edge or face node? If not, then I must be misunderstanding how the Lagrange compute_constraints is working, since it throws away identity values. If so, then that's going to be a problem for nonLagrange elements whose edge and corner degrees of freedom are essentially unrelated.  Roy Stogner  The SF.Net email is sponsored by: Beat the postholiday blues Get a FREE limited edition SourceForge.net tshirt from ThinkGeek. It's fun and FREE  well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ Libmeshdevel mailing list Libmeshdevel@... https://lists.sourceforge.net/lists/listinfo/libmeshdevel 