[Miase-discuss] SED-ML Simulation class update

[Dagmar K. <da...@eb...>]

Dear all,

we had some discussion on extending the SED-ML simulation class during 
the CellML combined workshop in Auckland. Frank and I tried to come up 
with a good class structure to map bifurcation analyses and steady state 
analyses. The according UML diagram can be found on sourceforge (PDF):
http://miase.svn.sourceforge.net/viewvc/miase/sed-ml/documents/sed-om/sedom-tmp.pdf?revision=115

I marked the changed classes in red.
We do have three main simulation classes now, namely: 
BifurcationSearch1D (a bifurcation analysis over a parameter with a 
uniform range), TimeCourse (time courses with uniform, vector or 
functional range) and SteadyStateParameterScan1D (over 1 parameter with 
different ranges again).

Questions are:
1) Do you think this is a good way/structure of mapping steady state and 
bifurcation experiments to SED-ML?
2) Frank suggested that the AnySimulation class should be removed from 
the diagram as (1) the use of already defined classes should be 
encouraged,  (2) and self-defined simulation experiment classes cannot 
be reused anyways... he also mentioned that (3) it was still possible to 
describe an experiment that is not representable in SED-ML so far. It 
could always be described inside the notes/annotation element. I thought 
however that it might be useful to have at least a structure (even if 
very general) for self-defined experiment types. But maybe it is 
sufficient to state in the documentation that "not-representable 
experiments should be defined in the according notes/annotation". Are 
there any opinions on that?

Best,
Dagmar