Re: [moose-devel] Implementing a psuedo-reaction

[Subhasis R. <ray...@gm...>]

Hi,
  This is more of a user issue than a development issue, hence cc to
moose-generic mailing list.

On Tue, Mar 31, 2015 at 4:38 AM, Dilawar Singh <dil...@nc...> wrote:

>
> An equation in chemical model is following:
>
>      x(t) = a(t)*b(t)*c(t);   (A)
>
> Variable 'x' is a proxy for three independent moose.Pool with species a,
> b, and
> c having their own kinetics. In other reactions, 'x' is treated as input;
>

What do you mean by input to a reaction?


> therefor it should be implemented as moose.Pool or moose.BufPool.
>
>
In continuation of previous query, can you tell us more about the reaction
system? A pool is a pool of molecules and I cannot imagine the number of
molecules in one pool being the product of three others in a physical
process. In another way, your equation will be dimensionally invalid.


>
> AFAIK, currently moose.Pool does not have a Dest-Finfo which I can use to
> set
> the number of molecules in Pool.


For each ValueFinfo 'x', a moose class automatically gets a 'getX' and
'setX' DestFinfos. You could see this using the builtin doc function:

>>> moose.doc('Pool')
...
*Destination message field*
    setN: double
    getN: PSt6vectorIdSaIdEE
    setNInit: double
    getNInit: PSt6vectorIdSaIdEE
    setDiffConst: double
    getDiffConst: PSt6vectorIdSaIdEE
...



> I can only use 'increment' or 'decrement' which
> is fine if the equation was:
>
>      dx/dt = a*b*c (A1)
>
> So it boils down to having a moose.Pool where kinetics can be described
> using
> the equation (A); perhaps a combination of moose.Function and moose.Table.
> But
> in the end, I have to set the conc/n of moose.Pool using a message.
>

I am not sure if setting n externally is the right approach though. The
pools are supposed to participate in reactions and get their n updated
through that process.


>
> Another way of doing it to use the moose.Function and connect
> 'getDerivative' to
> the x (moose.Pool) 'increment'. I don't know if it would be mathematically
> wrong
> but moose.connect does not allow connecting these two Finfos. I can
> implement
> this if it is mathematically correct or does not cause numerical issues.
>

You cannot connect a DestFinfo to another DestFinfo. You can use
'derivativeOut' SrcFinfo from Function.


>
> I extended the moose.Pool class by adding a 'N' destinfo which set the n_
> to
> given value


As I mentioned above, this was unnecessary.


> but I don't know what values should A_ and B_ take. Merely changing
> n_ and setting A_ and B_ to 0 does give me correct shape of curves, but the
> values are way too high (seems like conc is replaced by number of
> molecules in
> Pool).
>

What are A_ and B_ ? conc is calculated from the number of molecules in
Pool and should be proportional.
It is not clear what you are trying to model. I would suggest you review
your model and see if chemical kinetics is really necessary.

Best,
  Subhasis