alasdairham

from __future__ import division
from random import random
import numpy as np
import pandas as pd

alasdairham

from __future__ import division
from random import random
import numpy as np
import pandas as pd

alasdairham

class RPS:
    actions = ['ROCK', 'PAPER', 'SCISSORS']
    n_actions = 3
    utilities = pd.DataFrame([
        # ROCK  PAPER  SCISSORS
        [ 0,    -1,    1], # ROCK
        [ 1,     0,   -1], # PAPER
        [-1,     1,    0]  # SCISSORS
    ], columns=actions, index=actions)

alasdairham

class Player:
    def __init__(self, name):
        self.strategy, self.avg_strategy,\
        self.strategy_sum, self.regret_sum = np.zeros((4, RPS.n_actions))
        self.name = name

    def __repr__(self):
        return self.name

    def update_strategy(self):

alasdairham

class Game:
    def __init__(self, max_game=10000):
        self.p1 = Player('Alasdair')
        self.p2 = Player('Calum')
        self.max_game = max_game

    def winner(self, a1, a2):
        result = RPS.utilities.loc[a1, a2]
        if result == 1:     return self.p1
        elif result == -1:  return self.p2

alasdairham

class Player:
    # ....
    def regret(self, my_action, opp_action):
        """
        We here define the regret of not having chosen an action as the difference between the utility of that action and the utility of the action we actually chose, with respect to the fixed choices of the other player. Compute the regret and add it to regret sum.
        """
        result = RPS.utilities.loc[my_action, opp_action]
        facts = RPS.utilities.loc[:, opp_action].values
        regret = facts - result
        self.regret_sum += regret

alasdairham

class Player:
    # ....
    def update_strategy(self):
        """
        Set the preference (strategy) of choosing an action to be proportional to positive regrets. e.g, a strategy that prefers PAPER can be [0.2, 0.6, 0.2]
        """
        self.strategy = np.copy(self.regret_sum)
        self.strategy[self.strategy < 0] = 0  # reset negative regrets to zero

        summation = sum(self.strategy)

alasdairham

if __name__ == '__main__':
    game = Game()
    print '==== Use simple regret-matching strategy === '
    game.play()
    print '==== Use averaged regret-matching strategy === '
    game.conclude()
    game.play(avg_regret_matching=True)

alasdairham

 https://gist.github.com/namoshizun/7a3b820b013f8e367e84c70b45af7c34

alasdairham

/*
 *
 * Eg. SEND_INTERVAL = 3000; CAPTURE_INTERVAL = 50
 *
 * - Capture mousemove & mousedown event after every 50 ms
 * - For each captured mouse event, save it into a local cache object
 * - After every 3 seconds, sends the cached mouse event history to server,
 *   then clear cache. Exceptions are:
 *   1. A click event triggers data submission to server (as described above) immediately
 *   2. Pause sending data to server when there is no new movement,