Catherine Lloyd wrote:
Hi Andrew and Randall
This has gone beyond a curation question... Can one of you please
explain daes to Lucian and also how OpenCell handles them?
In a CellML model, if you have two equations:
f(x, y, z, a, b, c) = g(x, y, z, a, b, c)
h(x, y, z, a, b, c) = i(x, y, z, a, b, c)
then this means that both equations in the model are true simultaneously
(CellML is declarative, not procedural).
The CellML Integration Service included with the CellML API currently
handles this case by using the Levenberg-Marquardt non-linear
optimisation algorithm to look for a solution to
(f(x, y, z, a, b, c) - g(x, y, z, a, b, c))^2 = 0
(h(x, y, z, a, b, c) = i(x, y, z, a, b, c))^2 = 0
It will satisfy as many of x, y, z, a, b, and c as possible using other
equations first.
Work is also in progress towards supporting IDA.
Best wishes,
Andrew
That would be much appreciated - Thank you :)
Best wishes
Catherine
Begin forwarded message:
*From: *Lucian Smith <lpsm...@spod-central.org
<mailto:lpsm...@spod-central.org>>
*Date: *11 March 2010 9:01:27 AM
*To: *Catherine Lloyd <c.ll...@auckland.ac.nz
<mailto:c.ll...@auckland.ac.nz>>
*Subject: **Re: Haugh 2004b*
Aha, thanks! So, I don't know anything about differential algebraic
equations, and I'm afraid the wikipedia page about it is opaque to me.
But the loop in question occurs with a couple 'Assignment rules' (as
they're known in SBML), with the first variable depending on the second
variable, and the second depending on the first. Is the claim that if
the
system is set up correctly, there is always a set of variables such that
those assignment rules will both be satisfiable? And that this
relates to
differential algebraic equations somehow? Or does OpenCell simply
execute
the first in each iteration based on the previous iteration's values, and
then execute the second based on those results?
(We're probably moving away from curation questions at this point...)
Thanks for the assistance! And I'm glad you appreciated the follow-up
post--I had been hoping to get at least a little bit of discussion out of
it, but I'll settle for a thanks ;-)
-Lucian
* Catherine Lloyd <c.ll...@auckland.ac.nz
<mailto:c.ll...@auckland.ac.nz>> [2010-03-10 01:15] writes:
Hi Lucian
Yes, it's a curation question :)
OK, so Haugh_2004 has this curation statement:
This CellML model runs in both COR and OpenCell. The units have been
checked and they are consistent. The CellML model may recreate the
results of the original published model but there is no simple
validation method as there are no "concentration against time" figures
in the paper. The CellML model is based on equations A1a, A1b, A1c,
A1d and A1e from the Appendix. Parameter values were taken from table
1 in the paper and were also supplied through correspondence with the
original model author.
I checked the model in COR and in OpenCell and it runs fine in both.
HOWEVER, Haugh_2004b has this statement:
This CellML model runs in OpenCell (but not COR due to the presence of
differential algebraic equations). The units have been checked and
they are consistent. The CellML model may recreate the results of the
original published model but there is no simple validation method as
there are no "concentration against time" figures in the paper. The
CellML model is based on equations A2-A5 from the Appendix (steady
state model). Parameter values were taken from table 1 in the paper
and were also supplied through correspondence with the original model
author.
COR reports an error similar to your tool - that the model has
"circular arguments". OpenCell can handle differential algebraic
equations though so the model runs in OpenCell - but whether or not it
makes any sense we don't know as there were no results figures in the
published paper to compare the CellML to.
Regarding the stars - this is an unfortunate problem in PMR2 that we
are aware about - we can only assign a star rating once in a model
exposure - regardless of how many different models there may be.
Tommy knows this and hopefully it will be fixed in the not too distant
future.
I hope this makes some sense - please let me know if it doesn't or
if you need more information.
Also thank you for your feedback post-meeting - I spotted it on cellml
discuss when I got back to work on Monday.
Best wishes
Catherine
On 10/03/2010, at 12:41 PM, Lucian Smith wrote:
So, I have a question that essentially amounts to a curation
question (I
think), so I figured I'd ask you directly instead of the list. But
I am
happy to ask the list if you think that would be better!
When translating the model haugh_2004b.cellml from
http://models.cellml.org/exposure/4ce2912573256c6a5483da117ed26d9e
my software found an error: The definition of 'C' in component 'C'
includes a variable 'R', and the the definition of 'R' in component
'R'
includes a variable 'C'. Then there is a connection between C.C and
R.C,
and between R.R and C.R. Which I *think* should end up with an
overdetermined model. In all other cases when I've found such a
loop, if
I look at the model's curation status, it will say something like
"this
model is overdetermined". But this has a two-star curation, and
claims
there are no problems (of this nature, at least). So is there not
actually a problem and my software thinks there is, or is there a
problem
that is not mentioned in the curation paragraph? Or was the model
changed
since its last curation? Or is the curation paragraph about
haugh2004.txt, and not haugh2004b.txt?
Thanks!
-Lucian
_______________________________________________
cellml-discussion mailing list
cellml-discussion@cellml.org
http://www.cellml.org/mailman/listinfo/cellml-discussion