> 2) tabulationStepSize does not belong in the specification, so it should 
> be removed. Software such as PCEnv should preferably not interpolate 
> onto a grid, but if it does, this should be defined elsewhere, and not 
> in the simulation metadata specification.

I'm not too sure whether this belongs in the simulation metadata or not, 
but I think it highlights that maybe people have differing views on what 
we are trying to achieve with this "simulation metadata".

Personally, I think that Andrew's abstract on the simulation metadata 
specification is pretty good. We basically want to provide the 
information required for a certain set of results to be generated from a 
specified model - i.e. model + simulation metadata = results.

For that process to be completely defined, don't we need to be able to 
describe the bound variable values at which model results will be 
generated? How else can you be sure that running the same simulation in 
two different pieces of software (or even multiple runs with the same 
software) will give you comparable results? Unless you are going to be 
happy matching curves or other types of interpolation/analysis 
introducing further error, don't you need to have the model results at 
exactly the same values of the bound variable(s)?

Maybe the use of the tabulationStepSize predicate (?) is not the best 
way to describe the set of values at which you want to have access to 
the model results, but its certainly sufficient for everything I have 
been doing. Perhaps we need some better way to specify the range of 
bound variable(s) values that we need results for, but the specification 
of a start value, an end value, and a tabulation step size seems a 
reasonable start to me.

Maybe its just the name that people don't like? I can't think of 
anything more suitable at the moment, but maybe someone has an idea?

I guess the more I think about it, the more I think this information 
should definitely be in the simulation metadata, assuming that my view 
of the simulation metadata providing all the required information to 
reproduce a specific set of results from a specified model is valid.

I would be interested to hear alternatives on where this sort of 
information should be provided?


Andre.
_______________________________________________
cellml-discussion mailing list
[email protected]
http://www.cellml.org/mailman/listinfo/cellml-discussion

Reply via email to