The UCB algorithm is a little different from the paper

[tobegit3hub]

According to the paper A Survey of Monte Carlo Tree Search Methods, the UCB function should look like this.

And here is the pseudo-code about the implementation.

The coefficient 2 should be outsize of the log. And this line of code is a little different from the paper https://github.com/haroldsultan/MCTS/blob/master/mcts.py#L115 .

[haroldsultan]

thanks. i corrected the code

[tobegit3hub]

Great and thanks for your sharing 🎉