Sam Isaacson

> We used partial propensity and composition rejection. We had plans to implement tau-leaping but we never got there. Our typical reaction networks had about 30 species and about 50...

I have no idea :) Never looked into it. But we can still make the basic stoichiometry matrices as dense matrices at least, it is just they may have floating...

But I don't see why the definition would change for the basic substrate, product and net stoichiometric matrices. It is more that I haven't thought about whether the various analyses...

Sounds good. Please do link any good references you find in this issue.

I think it is separate as this works: ```julia @add_reactions coupled_switch begin gD, ∅ → D end gD ``` but this fails ```julia @add_reactions coupled_switch begin gD*B, ∅ → D...

This also works ```julia @add_reactions coupled_switch begin gD*D, ∅ → D end gD ``` I think we need to update the macro to check for species that are in `coupled_switch`...

Hmm, does it make sense to make anything that doesn't appear in the parameter list or in a reaction a species by default? Maybe that could also useful in building...

The behavior should probably be consistent between `@add_reactions` and `@reaction_network`, so either we always allow undefined symbols to become species, we add a parallel set of macros that assumes this,...

Otherwise I guess we just modify the docs to tell people to add fictitious reactions? ```julia @reaction_network begin k1*X1, ∅ → X1 k2*X2, X1 → X2 k3*X3, X2 → X3...

Yeah, that is reasonable too. I guess we don't need to require the same behavior since it is an extension macro. If something is in the base network allow it...