Python utils.argmax函数代码示例

	def _add_to_cluster(self, cluster, _doc):
		super(WavgNetCONFIRM, self)._add_to_cluster(cluster, _doc)

		# competitive stage
		similarities = self._cluster_sim_scores(_doc)
		idx = utils.argmax(similarities)
		del similarities[idx]
		if similarities:
			idx2 = utils.argmax(similarities)
			if idx2 <= idx:
				idx2 += 1
			sim_vec2 = self.clusters[idx2].center.similarity_vector(_doc)
			self.clusters[idx2].network.learn(sim_vec2, 0.2)

def best_policy(mdp, U):
    """Given an MDP and a utility function U, determine the best policy,
    as a mapping from state to action. (Equation 17.4)"""
    pi = {}
    for s in mdp.states:
        pi[s] = argmax(mdp.actions(s), key=lambda a: expected_utility(a, s, U, mdp))
    return pi

 def actions(self, state):
     search_list = [c for c in self.decoder.chardomain if c not in state]
     target_list = [c for c in alphabet if c not in state.values()]
     # Find the best charater to replace
     plainchar = argmax(search_list, key=lambda c: self.decoder.P1[c])
     for cipherchar in target_list:
         yield (plainchar, cipherchar)

def determine_optimum_variants(unit1, unit2):
    """Determines the optimum variants between two units."""
    # TODO - improve performance by considering variants (1,1) (1, 2) and (2,1)
    # as equivalent.
    outcomes = defaultdict(dict)

    for v1 in MeleeRangedStrategy.VARIANTS:
        if not MeleeRangedStrategy.is_compatible(unit1, v1):
            continue
        unit1.strategy = MeleeRangedStrategy(unit1, v1)
        for v2 in MeleeRangedStrategy.VARIANTS:
            if not MeleeRangedStrategy.is_compatible(unit2, v2):
                continue
            unit2.strategy = MeleeRangedStrategy(unit2, v2)

            turn_order = (unit1, unit2)
            game_state = AveragingVersusGameState(turn_order, verbosity=0)
            game_state.run_combat()

            outcomes[v1][v2] = game_state.hp_delta

    # What's your best strategy?
    unit_1_strategies = { v1: min(outcomes[v1].values()) for v1 in outcomes }
    unit1_strategy = utils.argmax(unit_1_strategies)
    unit2_strategy = utils.argmin(outcomes[unit1_strategy])

    # for v1 in outcomes:
    #     for v2, hp_delta in sorted(outcomes[v1].items()):
    #         print '(%d, %d) => %+.2f' % (v1, v2, hp_delta)

    # print '%s\'s strategy: %s' % (unit1, unit1_strategy)
    # print '%s\'s strategy: %s' % (unit2, unit2_strategy)

    return (unit1_strategy, unit2_strategy)

def genetic_algorithm(problem, population, fitness_fn, ngen=1000, pmut=0.1):
  "[Fig. 4.8]"
  #MAX = 0
  for i in range(ngen):
    new_population = []
    '''
    print i, '------------'
    print '  ', MAX
    for p in population:
      print problem.value(p)
      if problem.value(p) > MAX:
        MAX = problem.value(p)
    '''
    for p in population:
      fitnesses = map(fitness_fn, population)
      s1, s2 = weighted_sample_with_replacement(population, fitnesses, 2)
      p1 = copy.copy(problem)
      p1.set_state(s1)
      p2 = copy.copy(problem)
      p2.set_state(s2)
      child = p1.mate(p2)
      child.mutate(pmut)
      new_population.append(child.initial)
    population = new_population
  return utils.argmax(population, fitness_fn)

def genetic_algorithm_stepwise(population):
	root.title('Genetic Algorithm')
	for generation in range(ngen):
		# generating new population after selecting, recombining and mutating the existing population
		population = [search.mutate(search.recombine(*search.select(2, population, fitness_fn)), gene_pool, mutation_rate) for i in range(len(population))]
		# genome with the highest fitness in the current generation
		current_best = ''.join(argmax(population, key=fitness_fn))
		# collecting first few examples from the current population
		members = [''.join(x) for x in population][:48]

		# clear the canvas
		canvas.delete('all')
		# displays current best on top of the screen
		canvas.create_text(canvas_width / 2, 40, fill=p_blue, font='Consolas 46 bold', text=current_best)

		# displaying a part of the population on the screen
		for i in range(len(members) // 3):
			canvas.create_text((canvas_width * .175), (canvas_height * .25 + (25 * i)), fill=lp_blue, font='Consolas 16', text=members[3 * i])
			canvas.create_text((canvas_width * .500), (canvas_height * .25 + (25 * i)), fill=lp_blue, font='Consolas 16', text=members[3 * i + 1])
			canvas.create_text((canvas_width * .825), (canvas_height * .25 + (25 * i)), fill=lp_blue, font='Consolas 16', text=members[3 * i + 2])

		# displays current generation number
		canvas.create_text((canvas_width * .5), (canvas_height * 0.95), fill=p_blue, font='Consolas 18 bold', text=f'Generation {generation}')

		# displays blue bar that indicates current maximum fitness compared to maximum possible fitness
		scaling_factor = fitness_fn(current_best) / len(target)
		canvas.create_rectangle(canvas_width * 0.1, 90, canvas_width * 0.9, 100, outline=p_blue)
		canvas.create_rectangle(canvas_width * 0.1, 90, canvas_width * 0.1 + scaling_factor * canvas_width * 0.8, 100, fill=lp_blue)
		canvas.update()

		# checks for completion
		fittest_individual = search.fitness_threshold(fitness_fn, f_thres, population)
		if fittest_individual:
			break

def WalkSAT(clauses, p=0.5, max_flips=10000):
    """Checks for satisfiability of all clauses by randomly flipping values of variables
    """
    # Set of all symbols in all clauses
    symbols = set(sym for clause in clauses for sym in prop_symbols(clause))
    # model is a random assignment of true/false to the symbols in clauses
    model = {s: random.choice([True, False]) for s in symbols}
    for i in range(max_flips):
        satisfied, unsatisfied = [], []
        for clause in clauses:
            (satisfied if pl_true(clause, model) else unsatisfied).append(clause)
        if not unsatisfied:  # if model satisfies all the clauses
            return model
        clause = random.choice(unsatisfied)
        if probability(p):
            sym = random.choice(prop_symbols(clause))
        else:
            # Flip the symbol in clause that maximizes number of sat. clauses
            def sat_count(sym):
                # Return the the number of clauses satisfied after flipping the symbol.
                model[sym] = not model[sym]
                count = len([clause for clause in clauses if pl_true(clause, model)])
                model[sym] = not model[sym]
                return count
            sym = argmax(prop_symbols(clause), key=sat_count)
        model[sym] = not model[sym]
    # If no solution is found within the flip limit, we return failure
    return None

    def update(self,x,y):
        """
        updates the ORT

        - x : list of k covariates (k x 1)
        - y : response (scalar)

        usage: 

        ort.update(x,y)
        """
        k = self.__poisson(1)
        if k == 0:
            self.__updateOOBE(x,y)
        else:
            for u in xrange(k):
                self.__age += 1
                (j,depth) = self.__findLeaf(x,self.tree)
                j.elem.update(x,y)
                #if j.elem.numSamplesSeen > self.minSamples and depth < self.maxDepth: # FIXME: which is the correct approach?
                if j.elem.stats.n > self.minSamples and depth < self.maxDepth:
                    g = self.__gains(j.elem)
                    if any([ gg >= self.minGain for gg in g ]):
                        bestTest = j.elem.tests[argmax(g)]
                        j.elem.updateSplit(bestTest.dim,bestTest.loc)
                        j.updateChildren( Tree(Elem(self.param)), Tree(Elem(self.param)) )
                        j.left.elem.stats = bestTest.statsL
                        j.right.elem.stats = bestTest.statsR
                        j.elem.reset()

def minimax_decision(state, game):
    """Given a state in a game, calculate the best move by searching
    forward all the way to the terminal states. [Figure 5.3]"""

    player = game.to_move(state)

    def max_value(state):
        if game.terminal_test(state):
            return game.utility(state, player)
        v = -infinity
        for a in game.actions(state):
            v = max(v, min_value(game.result(state, a)))
        return v

    def min_value(state):
        if game.terminal_test(state):
            return game.utility(state, player)
        v = infinity
        for a in game.actions(state):
            v = min(v, max_value(game.result(state, a)))
        return v

    # Body of minimax_decision:
    return argmax(game.actions(state),
                  key=lambda a: min_value(game.result(state, a)))

def minimax_decision(state, game):
    """Given a state in a game, calculate the best move by searching
    forward all the way to the terminal states. [Fig. 6.4]"""

    player = game.to_move(state)

    def max_value(state):
        if game.terminal_test(state):
            return game.utility(state, player)
        v = -infinity
        for (_, s) in game.successors(state):
            v = max(v, min_value(s))
        return v

    def min_value(state):
        if game.terminal_test(state):
            return game.utility(state, player)
        v = infinity
        for (_, s) in game.successors(state):
            v = min(v, max_value(s))
        return v

    # Body of minimax_decision starts here:
    action, state = argmax(game.successors(state), lambda ((a, s)): min_value(s))
    return action

def alphabeta_full_search(state, game):
    """Search game to determine best action; use alpha-beta pruning.
    As in [Fig. 6.7], this version searches all the way to the leaves."""

    player = game.to_move(state)

    def max_value(state, alpha, beta):
        if game.terminal_test(state):
            return game.utility(state, player)
        v = -infinity
        for (a, s) in game.successors(state):
            v = max(v, min_value(s, alpha, beta))
            if v >= beta:
                return v
            alpha = max(alpha, v)
        return v

    def min_value(state, alpha, beta):
        if game.terminal_test(state):
            return game.utility(state, player)
        v = infinity
        for (a, s) in game.successors(state):
            v = min(v, max_value(s, alpha, beta))
            if v <= alpha:
                return v
            beta = min(beta, v)
        return v

    # Body of alphabeta_search starts here:
    action, state = argmax(game.successors(state),
                           lambda ((a, s)): min_value(s, -infinity, infinity))
    return action

 def predict(example):
     """Predict the target value for example. Consider each possible value,
     and pick the most likely by looking at each attribute independently."""
     def class_probability(targetval):
         return (target_dist[targetval] *
                 product(attr_dists[targetval, attr][example[attr]]
                         for attr in dataset.inputs))
     return argmax(targetvals, key=class_probability)

    def predict(example):
        """Predict the target value for example. Calculate probabilities for each
        class and pick the max."""
        def class_probability(targetval):
            attr_dist = attr_dists[targetval]
            return target_dist[targetval] * product(attr_dist[a] for a in example)

        return argmax(target_dist.keys(), key=class_probability)

    def genetic_algorithm(self, problem, map_canvas):
        """ Genetic Algorithm modified for the given problem """

        def init_population(pop_number, gene_pool, state_length):
            """ initialize population """

            population = []
            for i in range(pop_number):
                population.append(utils.shuffled(gene_pool))
            return population

        def recombine(state_a, state_b):
            """ recombine two problem states """

            start = random.randint(0, len(state_a) - 1)
            end = random.randint(start + 1, len(state_a))
            new_state = state_a[start:end]
            for city in state_b:
                if city not in new_state:
                    new_state.append(city)
            return new_state

        def mutate(state, mutation_rate):
            """ mutate problem states """

            if random.uniform(0, 1) < mutation_rate:
                sample = random.sample(range(len(state)), 2)
                state[sample[0]], state[sample[1]] = state[sample[1]], state[sample[0]]
            return state

        def fitness_fn(state):
            """ calculate fitness of a particular state """
            
            fitness = problem.value(state)
            return int((5600 + fitness) ** 2)

        current = Node(problem.initial)
        population = init_population(100, current.state, len(current.state))
        all_time_best = current.state
        while(1):
            population = [mutate(recombine(*select(2, population, fitness_fn)), self.mutation_rate.get()) for i in range(len(population))]
            current_best = utils.argmax(population, key=fitness_fn)
            if fitness_fn(current_best) > fitness_fn(all_time_best):
                all_time_best = current_best
                self.cost.set("Cost = " + str('%0.3f' % (-1 * problem.value(all_time_best))))
            map_canvas.delete('poly')
            points = []
            for city in current_best:
                points.append(self.frame_locations[city][0])
                points.append(self.frame_locations[city][1])
            map_canvas.create_polygon(points, outline='red', width=1, fill='', tag='poly')
            best_points = []
            for city in all_time_best:
                best_points.append(self.frame_locations[city][0])
                best_points.append(self.frame_locations[city][1])
            map_canvas.create_polygon(best_points, outline='red', width=3, fill='', tag='poly')
            map_canvas.update()
            map_canvas.after(self.speed.get())

	def _update(self, action, reward):
		"""Update Q according to reward received."""
		
		Q = self.Q

		maxaction = argmax(range(self.env.action_space.n), lambda a : Q[self.current_state][a] - Q[self.previous_state][action])
		maxactiondiff = Q[self.current_state][maxaction] - Q[self.previous_state][action]
				
		Q[self.previous_state][action] += self.alpha*(self.A*reward + self.B*self.gamma*maxactiondiff)