lp:~douglas-brinkerhoff/um-feism/profile-obsolete
- Get this branch:
- bzr branch lp:~douglas-brinkerhoff/um-feism/profile-obsolete
Branch merges
Branch information
Recent revisions
- 6. By Douglas Brinkerhoff
-
Updated all of the memory structures and solvers to be more controllable. Eliminated the usage of the solve() command. As it turns out, an update to the development code in dolfin would have been sufficient, but this is a better code structure anyway. The code is now not experiencing slowdown. TODO: determine why the SubMesh command doesn't work. This would be really useful in terms of extracting the 1D mesh for surface calculations, and almost indispensible for extracting a surface mesh.
- 5. By Douglas Brinkerhoff
-
I realized recently that each time step was taking longer than the previous one, and that the amount of time for each step was linearly increasing. This is a problem in the solve() command. Switching all of the solvers to manually operating on PETSc vectors, so that we can have control. Need to manually allocate vectors for assembly as well, so that we're not reallocating each time. General memory management needs to be improved.
- 4. By Douglas Brinkerhoff
-
Added a basal water module and a crude dependence of precipitation on elevation
- 2. By Douglas Brinkerhoff
-
Added comments to most things. Also, fixed a couple things relate to memory addressing, and moved some stuff around to make this more efficient. Most notably, made the adjoint into a full adjoint by employing automatic differentiation on the forward model.
- 1. By Douglas Brinkerhoff
-
A major refactor of the profile model code. Still needs documentation, which is on my agenda for tomorrow. Everything seems to be working, including the adjoint.
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)
- Stacked on:
- lp:um-feism