Formalism around models about modelingtoolkit.jl HOT 3 CLOSED

That is the engineering pragmatism that gets things done, which I like. It would also be nice to generate some outputs that can be appreciated by the CS community. I think that requires either a horserace paper that shows that ModelingToolkit makes scientific programmers (in the broad sense) more productive, or a theoretical contribution that grapples with "scientific modeling" in a formal sense and proves some theorems about it.

Mathematical functions are a good formalism, but modeling needs its own language of variables and composition rules. One would like to talk about how to compose and transform models in a valid way. For example, Unitful.jl does a great job of implementing the formalism of "Dimensional Analysis", which is a composition rule for valid scientific models. Type systems or category theory people have probably thought about this a good bit.

from modelingtoolkit.jl.

I was going to make a package that does what I need, and then let someone build a formalization around it. Sounds like you're the guy.

from modelingtoolkit.jl.

We've had quite a few discussions over the years, and I think we've settled on something good. Thanks @jpfairbanks . I think our current documentation is a lot more clear about the "what ModelingToolkit" than before.

from modelingtoolkit.jl.