Sounds like a fascinating project!
I guess if I wanted to use OT in this kind of capacity, I'd return to something like my earlier thought... i.e., set up the algorithm outside of OT and generate the derivative/change in slope at specified control points. Then import that data as x/y coordinates of an object within OT using the file import function.
(Unfortunately, I cannot steer you clear of the "function editor" since that's where the file/import feature is offered ... or at least that's one place where it appears ... perhaps access to that internal OT program logic is also available through other aspects of the program that are called something else ;-).
Best wishes - sounds like an interesting effort.