Potential-Aware-Abstraction-In-Texas-Hold-em-Poker
Potential-Aware-Abstraction-In-Texas-Hold-em-Poker copied to clipboard
ChengTsang
order of execution
In which order do I have to execute the code?
The readme says that I first have to run "python generate_data_5public.py" but I already need "centroids_2.csv" and "centroids.csv" for that.
Thank you for your fast response.
"the two csv files are in the ./data" How have you created these files?
first part of the main algorithm 1: for n = r to r^ do compute abstraction An with cn clusters using abstraction algorithm Sn with distance function dn end for
As far as I understand the first part of the "main algorithm 1" you have to do a preclustering for flop, turn and river (for example like in this paper http://johanson.ca/publications/poker/2013-aamas-kmeans-abstraction/2013-aamas-kmeans-abstraction.pdf), right?
What is your relation to poker because your other github repositories are non poker related and the "Potential-Aware-Abstraction-In-Texas-Hold-em-Poker" paper is very poker specific? What is your relation to coding (professional, student, just for fun, ...)?
Maybe you are interested to dive deeper into poker related coding. I have a lot of work and maybe you are interested to work as a freelancer?
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Tuesday, 17 September 2019 09:48, ChengTsang [email protected] wrote:
the two csv files are in the ./data
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.
As far as I understand the first part of the "main algorithm 1" you have to do a preclustering for flop, turn and river (for example like in this paper http://johanson.ca/publications/poker/2013-aamas-kmeans-abstraction/2013-aamas-kmeans-abstraction.pdf), right?
Strictly speaking you only need to pre-cluster the river and you should be able to build the other layers of abstraction from that. Although keep in mind you end up with quite a few infosets at the turn. In the original paper the authors test their implementation just on the flop to avoid having to cluster the turn.
Hi,Cheng,I also wonder how do you create the two file("centroids_2.csv" and "centroids.csv) first?you generate them in clustering_5public.py and clustering_4public.py later,but use them in generate_data_5public.py firstly.
