~pefarrell/dolfin-adjoint/fix-ode : changes

~pefarrell/dolfin-adjoint/fix-ode » Changes from revision 755

From Revision 657 to 638

Rev	Summary	Authors	Date
657	More annotate=True\|False\|None refactorisation More annotate=True\|False\|None refactorisation	Patrick Farrell	11 years ago
656	New annotation scheme for variational_solver. New annotation scheme for variational_solver.	Patrick Farrell	11 years ago
655	Fix svd_burgers_perturb. Fix svd_burgers_perturb.	Patrick Farrell	11 years ago
654	A start at the Great Refactorisation of annotate=True\|False\|... A start at the Great Refactorisation of annotate=True\|False\|None.	Patrick Farrell	11 years ago
653	Rename functions produced by dolfin-adjoint, as requested by... Rename functions produced by dolfin-adjoint, as requested by Martin Alnaes.	Patrick Farrell	11 years ago
652	Need libadjoint for the exception Need libadjoint for the exception	Patrick Farrell	11 years ago
651	Use the coarse version of the viscoelastic mesh Use the coarse version of the viscoelastic mesh	Patrick Farrell	11 years ago
650	Enable the user to force projections to be annotated. Enable the user to force projections to be annotated.	Patrick Farrell	11 years ago
649	Have the nullspace example working (once dolfin projects the... Have the nullspace example working (once dolfin projects the nullspace components out of the RHS).	Patrick Farrell	11 years ago
648	If you compute the derivative with the tangent linear model,... If you compute the derivative with the tangent linear model, it all works! Extremely interesting.	Patrick Farrell	11 years ago
647	Support nullspaces of KrylovSolvers in replays. Support nullspaces of KrylovSolvers in replays.	Patrick Farrell	11 years ago
646	Yet another way KrylovSolver can be used. Yet another way KrylovSolver can be used.	Patrick Farrell	11 years ago
645	Handle the case where the rhs-form of an equation is empty. ... Handle the case where the rhs-form of an equation is empty. Fixes lp:1177808.	Patrick Farrell	11 years ago
644	Fix for the projection option introduced yesterday evening. Fix for the projection option introduced yesterday evening.	Patrick Farrell	11 years ago
643	Make a project option to compute_gradient, just like James' Make a project option to compute_gradient, just like James' timestepping library.	Patrick Farrell	11 years ago
642	Fix a corner case with projection annotation. Fix a corner case with projection annotation.	Patrick Farrell	11 years ago
641	Implement the replay callback, as requested by Martin. Implement the replay callback, as requested by Martin.	Patrick Farrell	11 years ago
640	Handle Constant.assign. Handle Constant.assign. There are two options how to handle this. (Well, three, really). The first option is to just consider Constants like Functions in FunctionSpace(mesh, "R", 0), and then all the machinery just works: they are annotated as initial conditions, their updates get assignment equations, etc. The disadvantage of that is the unacceptable performance penalty: most likely Constants will be used in many equations, and so their adjoint equation will have lots and lots of entries, and be a nightmare to assemble (c.f. Marie's experiences of the integral-constraint Lagrange multiplier in the cardiac example). It isn't necessary at all to compute it, but it would be computed if they were treated like Functions. The second option (adopted) is to consider the value of Constants as a mutable part of the state to be bundled with the context of the callbacks registered with libadjoint, just as the values of parameters to Expressions are. This slightly increases the memory required, but hopefully people aren't declaring millions of Constants. It has a nasty side-effect which I'll discuss below. The third option (preferred) is to not assign to Constants. They're either Constant or they're NotConstant! A Constant that changes value is a contradiction in my book, but anyway. I don't get to define dolfin semantics. The nasty side effect: so, when a callback is called, it re-sets the values of all Constants to the values they had when it was registered -- as described above, the values of Constants are now part of the state to be re-set. Except when they're not! And when are they not? When you're perturbing the value of the damn Constant (such as for a Taylor test, or an optimisation). Then you don't want to re-set the Constant value! Agh! You only want to re-set the values of the "other" constants. Hideous, no? Anyway, the best solution I could come up with was to say that "Constants that are part of a ScalarParameter aren't to be re-set", which fixes the Taylor test/optimisation problem at the cost of a nauseating hack. An alternative would be to introduce a NotConstant class, which behaves like a Constant except it can be assigned to, but I doubt that would fly.	Patrick Farrell	11 years ago
639	Some fixes for symmetric BC application and dolfin == 1.2.0. Some fixes for symmetric BC application and dolfin == 1.2.0.	Patrick Farrell	11 years ago
638	Expose some utility functions useful for annotating nonlinea... Expose some utility functions useful for annotating nonlinear problems with Picard: annotate, and record.	Patrick Farrell	11 years ago