On 12/09/11 20:30, Jonathan Cooper wrote:
Dear all,

I haven't gone through the detail of Andrew's draft, but a general point
occurred to me on reading the features list. Many tools would only
support a subset of these features (the most common being just ODEs with
no events). Does it make sense to require a secondary specification for
each combination of features, or should secondary specifications allow
for being combined to build up a desired feature set? In other words,
should the example instead be 3 specs (ODEs, DAEs, events) that can be
combined as a model author wishes?

Hi Jonathan,

I think the features potentially interact to some extent, so I don't think they are entirely directly composable. However, it would certainly be possible to build one secondary specification on top of one or more other specifications with same manual work to sort out some semantics.

I think we certainly could produce an ODEs spec, and a DAEs spec (so that software which suppports the DAEs spec would also support all models coded to ODEs), and a DAEs + events spec that would support models coded to the former two specs, as well as the infinitesimally delayed / boolean variables features needed for events.

Best wishes,

Also, I can't recall whether there will be any mechanism for a model
document to declare which secondary specification(s) it uses, if any?

Best wishes,

On 12/09/2011 03:27, Andrew Miller wrote:
Hi all,

The draft CellML 1.2 specification I sent out last week

is very generic and delegates the responsibility for narrowing down the
scope of CellML to something that can be implemented to secondary

To get an idea of what a secondary specification might look like, I've
created a draft for models with the following features:
* Differential algebraic equations (including ODEs) are supported.
* There can be multiple dependent variables, but only one independent
* Basic events can be handled using boolean variables and piecewise
based rules to make changes when events happen.

The draft is available at:


It is also up on GitHub:

Best wishes,
cellml-tools-developers mailing list
cellml-tools-developers mailing list

cellml-discussion mailing list

Reply via email to