How did we manage so long without such a page? I added some more references.

This seems a better place to reply to your comment about Petri nets and Bayesian nets in `Blog - Contribution to the MPE blog'.

>Groovy! This is an important ‘inverse problem’ I haven’t started thinking about: not starting with a reaction network and deriving predictions, but starting with empirical data and trying to fit it to a reaction network.

>In some way that I haven’t begun to ponder, this should connect reaction networks to machine learning and Bayesian networks.

I don't get your intuition here. They are both examples of statistical inference, but so many things are examples of statistical inference (especially to me!) that this is not much of a connection.

> There’s something big going on here… but it’s not a simple case of “I see a network here and a network there, so they must be somehow the same”. In Bayesian networks the edges indicate causality, some random variable affecting another. That’s a seemingly different use of edges from reaction networks, where they mean something turning into something else! So we’ll need to invent a sufficiently rich framework to fit these two together.

A Bayesian network is acyclic. That seems a very restricted kind of reaction network.

This seems a better place to reply to your comment about Petri nets and Bayesian nets in `Blog - Contribution to the MPE blog'.

>Groovy! This is an important ‘inverse problem’ I haven’t started thinking about: not starting with a reaction network and deriving predictions, but starting with empirical data and trying to fit it to a reaction network.

>In some way that I haven’t begun to ponder, this should connect reaction networks to machine learning and Bayesian networks.

I don't get your intuition here. They are both examples of statistical inference, but so many things are examples of statistical inference (especially to me!) that this is not much of a connection.

> There’s something big going on here… but it’s not a simple case of “I see a network here and a network there, so they must be somehow the same”. In Bayesian networks the edges indicate causality, some random variable affecting another. That’s a seemingly different use of edges from reaction networks, where they mean something turning into something else! So we’ll need to invent a sufficiently rich framework to fit these two together.

A Bayesian network is acyclic. That seems a very restricted kind of reaction network.