Completed projects since last we talked

• correct other O(n) data structures
  • this project was abandoned. After talking with Gernot, we can easily move to 10M nodes without fixing these structures.
• reformulate the memfem equations
  • done, Memfem can now implement both CARP and a Phi_e/Vm equation system. All changes have been documented and gone over with Umar
• Getting the input/output files to match in CARP and memfem
  • I have a script to generate a memfem mesh from a carp mesh
• Get limpet in memfem
  • done, with a caveat -- Limpet may have some strange bug? I couldn't find the problem after 2 days debugging.
• backwards IM Solvers - abandoned, and after talking with Gernot it may not have been all that useful.
• Get EasyML? working in LIMPET - Gernot has no real interest in this.
• test memfem on teragrid clusters
  • Done, Umar has everything compiling there. Memfem runs even faster than on UIUC cluster
• We found a bug in aLRd relating to update of Ki and Nai. Ischemia experiments might need updating.

Loose ends

• Mechanics code in memfem
  • I believe Umar has tested something that will work. Slava has been very busy, so it's stagnated.
• input file read in.
  • Using the larger meshes, memfem can take 42 minutes to read in the matrices. This is unacceptable and easily changed, but I didn't have the time. Umar knows what needs to be done.
• speed differences between LIMPET and Memfem
  • An initial single_cell test of limpet_ucla_rabbit and ucal_rabbit_base with lookup tables suggests that Gernot's code is 2x as fast?
  • Gernot's loop over the models happens inside his ionic model code. Mine happens outside the ionic model code. The only difference is that my version needs to call a virtual function every iteration whereas Gernot's doesn't. I modified alrd.cc and put all the calculations in a for loop, and compared this to a for loop placed outside the single_cell binary. I got results that suggest that Gernot's way of doing things is 30% faster.
• Pukinje Fibers
  • Umar has been spending alot of time on this, it's coming along. but Ed's paper isn't forthcoming. I helped Umar on the math behind it, and I think he knows what needs to be done now.