Dynamic number of players in multi-players simulations (for static number of arms and static problem) ?

[pratayaa]

Thank you for your help and support, just one more thing i want to ask- How we can change number of USERS dynamically for satic number of arms ?

[Naereen]

OK, first of all, this is ridiculous, it has to stop. Please, use a SHORT title, and edit the text box when you raise a new issue.

[Naereen]

Then to answer, I didn't implement this either.

It would be very similar to the task lists I gave on issue #123, but even easier: the number M(t) of players is (usually) always bounded between 1 and K (nb of arms).
It shouldn't be too hard for the Evaluator/MAB/Result classes, but regarding algorithms, you have to think about what you want to model: sleeping users or arrivals/departures.

• "Sleeping users" refers to where there is M_max (e.g., K) users playing the game, but at each time, only a subset gets to play (and receive rewards). This should be quite easy to do, as it does not need any change from the algorithms side (some users just don't receive any feedback nor take decision at some step, but they are not even aware of it!)
• "Arrivals/departures" is harder, as an arrival means a new algorithm has to be initialized.

I am also interested in this extension, but like for #123, I cannot take some time right now to code it myself.
So either you try to implement all this yourself, try to do some simulations, and when it works fine, submit a pull request (then I'll try and probably merge); or you forget about it until at least a few weeks... Sorry to be too busy!