RE: Event assignment semantics
04 Oct '05 17:27
Sauro> With respect to the event question I asked
Sauro> earlier, the problem appears to have a
Sauro> wider scope. Not only does this problem
Sauro> relate to event assignments but also rules
Sauro> and triggers. There are probably other
Sauro> areas too. What this means is that one
Sauro> cannot reference local parameters in any
Sauro> rule, event trigger or event
I know other people already chimed in, but just to be clear
on this issue, here are some points about SBML L1-L2v2:
* Events and rules don't have provisions for their own
lists of local parameters. Only kinetic laws can have
* Local parameters in kinetic laws are local and can't
be "seen" outside the kinetic law rate expressions.
(That's what makes them "local" :-)).
* It is indeed the case that one can't reference local
kinetic law parameters by rules, events or other
things outside the kinetic law.
Sauro> Obviously this is very restrictive and make
Sauro> the utility of local parameters somewhat
Well, yes, that may be so. But the alternative of allowing
outside things to reference *local* parameters would
basically do away with the concept of local parameters
altogether, wouldn't it? For instance, you would have to
assure uniqueness of the local parameter identifiers with
respect to global parameters.
Sauro> It's only since we've been experimenting with a new
Sauro> simulator in the last few months which we thought,
Sauro> in the interests of compatibility, make fully sbml
Sauro> friendly, have these issues become apparent.
This may be answered by your reply to my other message on
this topic, but I'm curious about what it is that your
software was doing before that (apparently) was at odds with
provisions in SBML. (And to be perfectly clear, I'm not
asking this in a left-handed attempt to imply that you were
doing something wrong; quite the contrary. It's a genuine
question about where the differences lie.)
Sauro> Unless I've have got this wrong, I would
Sauro> like to know what recommendations there are
Sauro> to deal with the issue. There are many
Sauro> simulators which use rules and a few which
Sauro> use events so the problem must have been
I think, as you wrote in a follow up posting, it's not
actually an issue about events. It's simply a manifestation
of what "local parameters" mean.
Sauro> Someone mentioned (Nicolas) making it a
Sauro> rule that all symbols in the global and
Sauro> local spaces be unique. This would
Sauro> certainly appear to solve the problem, are
Sauro> there any plans for this in the version 2
Sauro> revision? Will this be discussed at the
Sauro> sbml meeting in Boston?
We could certainly have a discussion about this in Boston.
It promises to be a lively debate :-).
Sauro> I am curious to know whether libSBML has any
Sauro> sematic check on this, i.e to flag expressions
Sauro> which reference local parameters?
No, not yet, but it is planned. There is a specific rule
listed under the MathML checks on
for the following: "The value of an id attribute of a local
parameter can only be used in ci elements within the MathML
contained in the same kineticLaw element that contains the
parameter element. [implicit in L2V1 specification section
3.5 & section 4.9.7]"
This isn't in libsbml yet, but is planned. The MathML
checks will require a little bit more work than other types
of checks; that's why they've been pushed off.
Powered by FUDforum. (Copyright Advanced Internet Designs Inc.)