Global performance curves
If we plot convergence time T versus social/confidence coefficient j for our two examples, we can immediatly make some remarks.
For Alpine function, the global result is better than the theoretical one. It means the best particle is not always the same during the process, and the change is made on a good way.
For Banana function, the global result is a bit worst than the theoretical one. It means either the best particle is not always the same and change is not very "intelligent", if I dare say, or re-hope process is used quite often.