History for Minutes 19 June 2006

changed:
-
Present: Shane, Karl, Travis, Poul.

An "audio recording":ftp://ftp.bioeng.auckland.ac.nz:/cmiss/opencmiss/meeting-2006-06-19.ogg is available.

MPI Test problem proposal and implementation

* Karl updated the suggested test problem:

  * made the time step implicit as a linear solve is required for the
    non-diagonal mass-matrix anyway.

  * changed the solution to make boundary conditions homogeneous.

  * No feedback, so will go ahead with this problem.


* libmesh has example that is very close to what we want.

  * Easy to change solver, preconditioner, tolerance, and probably
    maxiterations.

  * One thing we would want to change is that matrix assembly is done at each
    time step.

  * Another thing is the penalty method as discussed on [OpenCMISS/libMesh Development Notes].

    * Travis is happy with the method.

    * Karl is not.

  * but can get something going fairly quickly and improve after we have a
    base to improve on.

* Need to set up cm examples as well.

  * cm has time-dependent heat equation.

  * Could also possibly use "grid-based fem" monodomain model with a cell
    model that does nothing.


Libmesh and visualization

* Shane investigated visualization for libmesh.

  * Where is solution information collated before writing out the
    solution?

  * cmgui currently knows when a (some component of) field has
    changed and can automatically update aspects that depend on
    this.

  * Integrating with libmesh might require an explicit function to
    update graphics from libmesh data.  e.g. create isosurface from
    mesh and solution.

* GMV is a gui (executable only) available to display libmesh output.


eclipse

* provides a debugger that has MPI support.


cmiss.org development

* job description for new position may be moving away from web
  development as Gareth may be looking after web stuff.

* Gareth (logged in as stevens) fixed the long-standing "Missing emails for
  this tracker" bug which meant no notification of new bugs.


Decisions:

* Shane: want to make a decision on a framework ASAP (last February).

  * Shane away next week but keen to set up some criteria to make decisions on
    framework.

* Criteria

  * execution speed

  * language

  * time to develop

  * user community for support

  * flexibility


SUR cluster

* meant to arrive in NZ on 30 June.

Microphone

* We should buy a microphone for the meeting room.