racko.py

# Rack-O simulation

import random

class GreedyHeuristicPlayer():
	def getScore(self, rack):
		return 1

	def choose(self, rack, newCard):
		bestScore = self.getScore(rack)
		bestIndex = -1
		for index in range(10):
			candidate = rack[:]
			candidate[index] = newCard
			newScore = self.getScore(candidate)
			if newScore > bestScore:
				bestScore = newScore
				bestIndex = index
	
		return bestIndex

# Try to move cards to their natural positions (slot * 6)
class NaturalDistributionPlayer(GreedyHeuristicPlayer):
	def getScore(self, rack):
		score = 0
		for x in range(10):
			score += (rack[x] - x * 6) ** 2
	
		return 1.0 / (score + 1.0)

# Find the longest run
class LongestRunPlayer(GreedyHeuristicPlayer):
	def getScore(self, rack):
		longestRun = 0
		lastCard = -1
		for x in rack:
			if x > lastCard:
				longestRun += 1
			else:
				longestRun = 1
				
			lastCard = x
			
		return longestRun

# The passed in player object will be used to make selections
def runSim(player):
	# Create the initial configuration
	deck = [ x + 1 for x in range(60) ]
	random.shuffle(deck)
	rack = deck[:10]
	deck = deck[10:]
	
	# Draw some cards to simulate the other players holding them
	deck = deck[30:]
	
	# Run some number of turns for the player
	for iterations in range(100):		# Give up after 100 iterations
		iterations += 1
		drawn = deck.pop(0)

		slot = player.choose(rack, drawn)
		if slot != -1:
			# Replace the card in the rack with this one.
			deck.append(rack[slot])
			rack[slot] = drawn
	
			# In order now?
			if rack == sorted(rack):
				return iterations
		else:
			# Discard: put it on the bottom of the deck
			deck.append(drawn)

	return 100

player = NaturalDistributionPlayer()

NUMTRIALS = 50
iterations = 0
for trial in range(NUMTRIALS):
	iterations += runSim(player)

print 'average', iterations / NUMTRIALS, 'iterations'