Named variables/items in decision vector

[montanaviking]

Thanks so much for your hard work developing Pagmo!
I use Pagmo extensively for my transistor modeling work and plan to extend it for microwave circuit design and optimization.
I've written a node linear circuit solver for use with Pagmo.
However, I also extended Pagmo to use named variable i.e. a vector of C++ structure in which the individual items of the decision vector are named and also have the capability of being optimized or not depending on the needs. In a lot of optimization problems, the elements of the decision vector (dv) might need to be dynamically chosen, i.e. which items are to be optimized and some optimizations can be multi-level as well.
I feel that increasing the complexity of the decision vector could be useful to allow optimization flexibility as well as to clarify the coding.

[montanaviking]

HI,
By "named" variables in the decision vector I mean that the decision vector could use maps. I realize that named variable, e.g. as in Python, are not part of C++ and would be too complex to implement in the C++ Pagmo version. However, one could implement them as a map? One could also implement batch decision vectors as a two level map perhaps? I already translate the vector dc (decision vector) to a map internally in my fitness function, but this adds programming, reduces efficiency, and adds complexity. I do this because it makes my code much more readable and easier to modify and maintain. When I get more time next year, I'd like to look at the Pagmo source code and build these:

1. map dv
2. MPI or equivalent island for C++
   I'm also going to look at incorporating C++ Pagmo2 into QUCS - an open-source circuit simulator, for its optimizer.
   Thanks,
   Phil