I think it would help if we knew more about the nature of your project
and what you're trying to accomplish.  There have been a number of
papers on combining GP and PSOs (Riccardo and Bill being authors of
several), so it certainly can be done.  Riccardo's TinyGP code is
fairly specialized for symbolic regression problems, however, and
would need non-trivial modification to work in a different domain.

Given my limited understanding of what you're trying to do, it seems
that you have a couple of options.  You could try to modify TinyGP to
suit your needs.  If you go this route I'd recommend carefully reading
through the description of the code in the appendix to the Field Guide
as a starting point, and try to be more specific about what your goals
are when asking questions here.

Alternatively, you could look at other, more sophisticated and general
purpose GP systems (several are listed in the resources section of the
Field Guide), and see if they are closer to what you need.

Hope this helps,

Nic