Sam Isaacson

I like the idea of keeping `ReactionSystem`s as minimial and barebones as needed. This will make it easier in defining future transforms since there are less things to worry about...

If I understand correctly from the SBML document, is what you are looking for the ability to pass a species-dependent scaling factor that simply multiplies the corresponding ODE/SDE rhs (what...

Do SBML files ever encode a model that is both for ODEs and jumps, or are they supposed to use separate files? Looking at the document you linked this wasn't...

Yes, I think the problem you are recognizing is basically what I was trying (perhaps poorly) to explain! I guess check back here if you do end up needing anything....

As you noticed; species are now represented as `ModelingToolkit.Term{Real}` since we store their functional representation, i.e. `S(t)` and not `S`. You can see how to construct such objects in the...

And of course, you can use the ModelingToolkit macros if you aren't trying to do this programmatically: ```julia @parameters t @variables u[1:100](t) ```

That’s 10^9 reactions? Multithreading should work; but I’m not sure about GPU currently. @ChrisRackauckas can you comment on that?

Oh wow, that's pretty big. ModelingToolkit recently received some updates to improve performance, and we've just started working on optimizing for larger systems, though we are looking at systems much...

In the BCR testing I also found `ODEProblem` on an `ODESystem` was very expensive. In that case as expensive as the first ODE solve call with `KenCarp4-1`. The default is...

Since creating variables with a given name is now much easier, just ```julia a = Symbol("varname") a5 = (@variables $(a)[5])[1] ``` which gives ``` varname[5] ``` for the value of...