ABI CellML Meeting Minutes, 27th of April, 2011

Present: Dougal Cowan, Andrew Miller, Poul Nielsen

Apologies: Randall Britten, Tommy Yu, David "Andre" Nickerson

Last week's action items:

Action item 1: Dougal - look at coding up the Collier et al. 1996 paper into CellML and comment on the practicality of doing this.

  • Dougal is not confident of being able to code this model up in time for Poul's trip next week. The problem will be referred to Hanne or Andre.

This week's agenda:

1) CellML 1.2: How to deal with locally bound variables (Andrew)

Discussion:

  • Discussion of tracker item 2902, showing two possibilities of how to implement this.
  • Poul: wouldn't be able to use the same bound variable for more than one integral.
  • There was a discussion about how to implement bound variables.
  • Andrew: possibly a rule in the primary spec that a bound variable cannot have an initial value, and cannot be connected to anything.
  • Poul: more important than making life easier is to avoid creation of more concepts.

2) PMR2 development - update

Progress:

  • Poul: Tommy has staged 0.4. This could do with being tested.
  • Dougal: writing documentation for creation of Zinc viewer "FieldML" exposures.

Discussion:

  • Poul: it may pay to make a list of tests to perform as a matter of routine for PMR2 stagings.

Planning:

  • Dougal: will run some simple tests on the staged PMR2 0.4. Perhaps Hanne could do some tests on Mac.

3)  Repository contributions - update

Progress:

  • Dougal: fixing a few broken exposures containing file:// URIs in the session files. It would be good to get Tommy to perform a global search for session files with this problem.
  • Dougal: looking at the pattern formation paper (Collier et al.). Will show the paper to Hanne and possibly others to see if they might be able to code it up for Poul.

4) CellML core specs - update

Progress:

  • Andrew: going through the secondary spec, specifying which MathML operators can be used to build expressions. Looked at how to represent uncertainty in models, investigating how it is done in BUGS.

Discussion:

  • Poul: it would be useful to have pre-defined and user-defined distributions.
  • Andrew: uers defined distributions are potentially extremely complex to use, depending on how they are handled. It may be possible to recognise (via pattern matching) distributions.
  • There was a discussion about BUGS models and probability density functions.

5) CellML API - update

Progress:

  • Andrew: working on making a release - the Windows build system is very slow and keeps crashing. Very close to getting a release of the binaries out. Once the binaries have been released, try to get some user feedback. No user testing has been performed.