docplex optimizers

[gcattan]

Follow up for #28

This PR introduces one classical and one naive quantum optimizer, as per comment #56 (comment)
It also bump the version of docplex and add a new dependency to qiskit_optimization.

[toncho11]

I am looking into the test program. There are actually no comments in it.

1. It looks like a docplex problem is specified with "test_get_square_cont_var", but then the optimizer is created without providing this docplex model? I do not understand what the "test_get_square_cont_var" is doing.

2. Can we also call "solve()" on the two optimizers (quantum and classical) and validate that a result has been returned? Or it is too early for that?

For me the workflow (with comments) should be:

• define a docplex model
• select optimizer
• provide the docplex model to the optimizer (or several optimizes)
• call "solve" on the optimizer (or on several optimizes)

[gcattan]

Hi,

I have added comments to "test_get_square_cont_var", I hope it is more clear.

As for the functional tests of the solve method of the optimizers themselves, I would prefer to do it along with #26 to reuse the docplex model that would be defined in #26. What do you think?

[toncho11]

Maybe you can add a general description of "docplex.py " such as:
"This module contains both classic and quantum optimizers, and some helper functions. The quantum optimizer allows an optimization problem with constraints (in the form of docplex model) to be run on a quantum computer. It can be used for ... "

I think some more general description will be quite helpful.

[gcattan]

Thx for the review @toncho11 :)