SkipTree9/3Hedged

AuthorSean
Submission date2016-02-20 21:03:55.359170
Rating5741
Matches played353
Win rate57.79

Use rpsrunner.py to play unranked matches on your computer.

Source code:

if input == "":

    import collections
    import random
    import math

    log = math.log
    sqrt = math.sqrt

    class MarkovTree:
        def __init__(self, counts = None):
            self.counts = [0 for _ in xrange(3)]
            self.children = None
            self.total = 0

        def update_helper(self, h, i, p, d, skips):
            stop = False
            for j in xrange(p, len(h)):
                k = h[j]
                self.counts[i] += 2
                self.total += 2
                if stop or d >= 9 or skips >= 3:
                    return
                d += 1
                if self.children is None:
                    self.children = [None for _ in xrange(4)]
                    self.children[3] = MarkovTree()
                if self.children[k] is None:
                    self.children[k] = MarkovTree()
                    stop = True
                self.children[3].update_helper(h, i, j + 1, d, skips + 1)
                self = self.children[k]

        def update(self, h, i):
            self.update_helper(h, i, 0, 0, 0)

        def predict_helper(self, h, p, n0):
            for j in xrange(p, len(h)):
                k = h[j]
                for i, x in enumerate(self.counts):
                    n0[i] += x
                if self.children is None:
                    return
                self.children[3].predict_helper(h, j + 1, n0)
                child = self.children[k]
                if child is None:
                    return
                self = child

        def predict(self, h):
            n0 = [1, 1, 1]
            self.predict_helper(h, 0, n0)
            return n0

    R, P, S = 0, 1, 2
    index = {"R": R, "P": P, "S": S}
    beat = ("P", "S", "R")

    tree = MarkovTree()

    history = collections.deque([])
    output = random.choice(beat)
    random_plays = 1
    model_plays = 1
    score = 0
    random_play = 0
    model_play = 1
    play_type = random_play
else:
    m = 2.0 * (random_plays + model_plays)
    a = sqrt(m / random_plays)
    b = (score / float(model_plays)) + sqrt(m / model_plays)
    if a > b:
        play_type = random_play
        output = random.choice(beat)
        random_plays += 1
    else:
        model_plays += 1
        i = index[input]
        j = index[output]
        if play_type == model_play:
            if output == beat[i]:
                score += 1
            elif input == beat[j]:
                score -= 1
        play_type = model_play
        tree.update(history, i)
        history.appendleft(i)
        history.appendleft(j)

        counts = tree.predict(history)
        t = sum(counts)
        r = random.uniform(0, t)
        x = 0
        for i, p in enumerate(counts):
            x += p
            if r <= x:
                break
        output = beat[i]

random_plays *= 0.99
model_plays *= 0.99
score *= 0.99