I have a large-scale MINLP optimization problem, implemented in GAMS. To do so, I decomposed my problem into two sequential stages of MILP and NLP. For solving the MILP stage, I am using Cplex and for the NLP stage, which is a nonconvex optimization problem, I am using BARON. The MILP stage has about 8000 variables and the NLP stage has 2000 variables.

Everything works perfectly so far, but now I need to add a neural network on top of my optimization problems to determine some of the parameters. For this aim, I can use either MATLAB or Python.

Now my question is, does Pyomo have enough capability and solvers to solve both large-scale MILP and nonconvex NLP problems, so that I can deal with the only Python for both optimization and neural netwotk, or It would be better to use GAMS for my optimization problem and implement my neural network in Matlab or Python and link it to GAMS?

I appreciate your responses.