Update binary.py

[CPapadim]

Hi, I'm the reporter of issue 33

I think the issue is that rather than moving toward and returning the historical social / global best, you are moving toward and returning the current social / global best. I think for PSO you always need to be moving to historical and not current positions whether its the personal or the social / global best.

I've made some updates to fix the issue.

I tested on my code, and with my updates every iteration I either get the same 'best' position I had in the previous iteration, or a better one.

I'm not super familiar with this package's codebase so this request likely needs review and testing.

[ljvmiranda921]

Hey, thanks for the fix! I can accept this already, but would you like to compare your fix to mine? It would be nice to have more contributors as possible. 👍

Unfortunately, the local_best.py suffers a similar problem (see my PR). If you wish to fix the same, it is also welcome!

Thanks a lot for your effort!

[CPapadim]

As we discussed, I think my fix keeps neighbors working properly. So that we're only using nearby particles for the social best rather than all of them. I can update local_best.py similarly

[ljvmiranda921]

Okay then, I'll just wait for the commit to be pushed :)

[ljvmiranda921]

Nice, this is good to go 👍

Just one last logistical request. Normally I ask our contributors to rebase commits into 1 or 2. For this, probably one commit for the binary.py and another for local_best.py. And also, we follow this commit format:

Short and sweet imperative title (#33)

Short description that describes why the commit was done. The title
usually should be in present tense. Maybe "Fix error in <file> (#XX)"
then this paragraph will describe a little about what was fixed or the
general idea of the fix.

Author: <your-github-username>
Email: <your-contact-email>

That's all, after this we can merge already. Thanks a lot!