Minutes 27 February 2006
Meeting held 11am 27 February 2006
Present: Carey, Travis, Poul, Greg, Peter, Karl
Apologies: Andre, Chris, Shane
An audio recording of the meeting can be "downloaded":ftp://ftp.bioeng.auckland.ac.nz/cmiss/opencmiss/meeting-27.2.2006.ogg
1 Data structures.
How to handle interaction of adjacent or contained subregions.
- Is equivalence of higher order subregions required or useful?
- Are the "TSTT":http://www.tstt-scidac.org/software/ Geometry, Mesh, Field, and Relation specifications helpful?
Poul has been looking at TSTT, Karl has had a good look through it too. Their specification handles adjacency and other element types. Their implementations only support tet meshes. Their implementations handle distributed processing but it isn't covered in the spec.
Examples from Poul.
2 Development framework.
- How are we going to develop code?
Karl is going to use Andre's code to start to get a feel for MPI. This is not necessarily the start of opencmiss - more an exercise in education for Karl and those who want to be involved.
We need to get out an advert for the position to fill - awaiting us to decide what the necessary skills would be. Peter to dig up the advert drafted ealier and send it around. We'll check it and get it posted.
Chris is developing an MPI test program in Fortran90 and is concerned that his and Karl's effort might be complimentary but he hasn't had any communication. Karl has not begun to do anything yet. Will get Chris to send an explation of what he is doing. Peter, Karl and Travis to get in touch with Chris and see how the task could possibly be split up between them. Karl and Travis to write up what they see as the division of tasks.
Regards the solvers, there a some good parallel solvers available, we don't need to do that part ourselves.
It might be useful for us to write a distributed matrix-vector multiply for MPI experiance, we are going to need one anyway.