Extending Optimization tool of sympy | GSoC2019

[Yograj Singh Mandloi]

Hello Everyone,

The current optimization in sympy is limited to root finding ref. I would like to work on adding more features to the optimization library. I am thinking of implementing various direct-indirect techniques for linear, non-linear dynamic optimization, for example, finding the optimal trajectory where the decision variables are functions of time and states of a dynamic system. Problems can be specified with dynamic or static constraints over parameters and the state variables while selecting the desired optimization techniques (like collocation, shooting or proposed in trajectory optimization research) as input. What do mentors think about this idea?  any suggestion would be appreciated. 

I previously implemented some optimization algorithms like genetic algorithm, gradient descent etc from scratch in python. As an aerospace engineering background, I have good of understanding the dynamics optimization. 

Thanks

Regards

Yograj      

[Aaron Meurer]

The link you referenced is part of mpmath, which is a dependency of SymPy.

SymPy submodules should generally be symbolic in nature. Purely
numeric applications are out of scope. See
https://github.com/sympy/sympy/wiki/GSoC-2019-Ideas#non-ideas. Unless
you can convince us that your idea requires a symbolic system.

I would suggest looking at the scipy project if you are interested in
numeric optimization. They are participating in GSoC through NumFOCUS.
https://github.com/scipy/scipy/wiki/GSoC-2019-project-ideas

Aaron Meurer

> --
> You received this message because you are subscribed to the Google Groups "sympy" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sympy+un...@googlegroups.com.
> To post to this group, send email to sy...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sympy.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/5c631fb5-6f85-4b65-88b2-e2b9b42bf8a5%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.