In Attendance: Matt Halstead, Carey Stevens, Andrew Millar, Peter Villiger
Mike Cooling, David Nickerson, Poul Nielsen
Mike: Problem with CellMl model variables that are either never set or initialized but are used in
equations - orphan variables.
Andrew: mozcellml needs single variable on one side - can not solve problem where right variable gets value
from another equation. Include tracking in mozcellml to recognize when an initial value is missing and produce
error - ie. not enough information. Include tool to find orphan variables.
Poul: should this be in API?
Matt: some models are intended as conceptual and not to be simulated at all. This suggests that model is not
broken at all even if it contains orphaned or uninitialized variables, as long as it is a direct representation
of the paper it comes from. In this case a new cellml model could be created for simulation purposes referencing the
original model so that the original representation of the paper is retained. If the original model is not
a direct representation of the paper it needs to be updated in the repository.
Finding values of variables at a previous timestep?
Poul: Find a point in a function of time where a max has been obtained and trigger something 2 seconds later.
Convolution in time to relate functions and represent delay.
Mike: Issue complicated because maximums are local.
Poul volunteered to check maths - makes no promise of solution!
matt: cellml repository impl.? and graphical representation of cellml biology - check on development site
Bill is working on cellml editor 8 hours a week - andrew can be involved on specific parts if he wants.
Binary releases on sourceforge. Bill will start tagging models and adding release notes. Mathml source will be
on sourceforge.
Problems - visualizing, editing mathml, performance problems. Documentation?
Andrew: still working on integration of cellml editor to mozcellml.
can't check all constraints in XSD. Changed repository models to 1.1 namespace. Issues - naming conventions,
misspelt names. List emailed to teamcellml.
Distribution/access from other languages. Korver link? Suggests API should be Korver compliant to allow access
from different languages. Set up editor to expose API across Korver.
Matt: kind of understands korver and the people that use it...
