Meeting Minutes 21st August 2006

Meeting Minutes 21th August 2006
Present: Poul Nielsen, Peter Hunter, Tommy Yu, Shane Blackett, Andrew Miller

Peter: Dan Beard asked if we could adapt CellML to cope with a wider class of equations. However, it appears that CellML can already cope with every he is using. Perhaps he is asking for more elements in line with reaction? Peter will follow up to clarify that reaction is deprecated, and that the models can be created without providing this information.

The was also some discussion about what editor we should recommend to Dan Beard. PCEnv is not ready yet, and everyone agreed that the highest priority should be a robust simulator. However, there was agreement to add MathML to PCEnv in a later release.

Peter is concerned that website features which were present in the old site are no longer available in the old site. In particular, it is currently not possible to download the mathematics for a model as a PDF. Tommy will investigate this.

Tommy noted that there are currently problems uploading CellML 1.1 models into the repository. This appears to be a consequence of the fact that the stylesheet used has hardcoded references to the CellML 1.0 namespace. There are also issues of uploading multiple models which refer to each other. Tommy will correspond with Carey to see if he knows of any other problems, and look into fixing it.

Peter was keen to see all existing models get put into the new repository. Some of these models require hand-editing to fix RDF or similar problems.

Andrew has revised the earlier simulation metadata specification. It no longer describes important variables. There was no disagreement with this, but differing reasons for why it needed to be removed.

There was some discussion of whether a new specification is needed, or if this should extend the existing cmeta specification. However, no reason not to make a new specification was found.

Peter asked whether we need to reference the bound variable, instead of just assuming it was time (and argued that everything in the model has time as the only bound variable, and that FieldML can solve the issues if we have, e.g., models defined over space). However, there was no consensus that this was a good idea, for several reasons. Firstly, the word 'time' isn't special in CellML, and so we would be unnecessarily restricting the specification. Even if the bound variable refers to time, we don't know whether the variable will be called Time or something else. It is also possible for models to be defined in a single dimension other than time. Also, not much effort is required to refer to the variable cmeta:id, while having it saves a lot of potential ambiguity.

There was also some discussion about the way we describe the solver. The discussion was cut short due to lack of time, with the agreement to schedule another meeting on this.