def create_variable_to_message_binding(self, variable):
new_distribution = Gaussian.from_precision_mean(0.0, 0.0)
binding = Factor.create_variable_to_message_binding_with_message(
self, variable,
Message(new_distribution,
"message from %s to %s" % (self, variable)))
return binding
def update_message_variable(self, message, variable):
old_marginal = copy(variable.value)
old_message = copy(message.value)
message_from_var = old_marginal / old_message
c = message_from_var.precision
d = message_from_var.precision_mean
sqrt_c = sqrt(c)
d_on_sqrt_c = d / sqrt_c
epsilon_times_sqrt_c = self.epsilon * sqrt_c
d = message_from_var.precision_mean
denom = 1.0 - w_exceeds_margin(d_on_sqrt_c, epsilon_times_sqrt_c)
new_precision = c / denom
new_precision_mean = (d + sqrt_c * v_exceeds_margin(d_on_sqrt_c, epsilon_times_sqrt_c)) / denom
new_marginal = Gaussian.from_precision_mean(new_precision_mean, new_precision)
new_message = (old_message * new_marginal) / old_marginal
message.value = new_message
variable.value = new_marginal
return new_marginal - old_marginal
def update_message_variable(self, message, variable):
old_marginal = copy(variable.value)
old_message = copy(message.value)
message_from_var = old_marginal / old_message
c = message_from_var.precision
d = message_from_var.precision_mean
sqrt_c = sqrt(c)
d_on_sqrt_c = d / sqrt_c
epsilon_times_sqrt_c = self.epsilon * sqrt_c
d = message_from_var.precision_mean
denom = 1.0 - w_exceeds_margin(d_on_sqrt_c, epsilon_times_sqrt_c)
new_precision = c / denom
new_precision_mean = (d + sqrt_c * v_exceeds_margin(
d_on_sqrt_c, epsilon_times_sqrt_c)) / denom
new_marginal = Gaussian.from_precision_mean(new_precision_mean,
new_precision)
new_message = (old_message * new_marginal) / old_marginal
message.value = new_message
variable.value = new_marginal
return new_marginal - old_marginal
def __init__(self, mean, variance, variable):
GaussianFactor.__init__(self, "Prior value going to %s" % variable)
self.new_message = Gaussian(mean, sqrt(variance))
new_message = Message(Gaussian.from_precision_mean(0, 0),
"message from %s to %s" % (self, variable))
self.create_variable_to_message_binding_with_message(
variable, new_message)
def create_variable_to_message_binding(self, variable):
new_distribution = Gaussian.from_precision_mean(0.0, 0.0)
binding = Factor.create_variable_to_message_binding_with_message(
self,
variable,
Message(new_distribution, "message from {} to {}".format(self, variable))
)
return binding
def update_message_variable(self, message, variable):
old_marginal = copy(variable.value)
old_message = message
new_marginal = Gaussian.from_precision_mean(
old_marginal.precision_mean + self.new_message.precision_mean - old_message.value.precision_mean,
old_marginal.precision + self.new_message.precision - old_message.value.precision)
variable.value = new_marginal
new_message = self.new_message
message.value = new_message
return old_marginal - new_marginal
def update_message_variable(self, message, variable):
old_marginal = copy(variable.value)
old_message = message
new_marginal = Gaussian.from_precision_mean(
old_marginal.precision_mean + self.new_message.precision_mean -
old_message.value.precision_mean, old_marginal.precision +
self.new_message.precision - old_message.value.precision)
variable.value = new_marginal
new_message = self.new_message
message.value = new_message
return old_marginal - new_marginal
def partial_update(self, prior, full_posterior, update_percentage):
prior_gaussian = Gaussian(prior.mean, prior.stdev)
posterior_gaussian = Gaussian(full_posterior.mean, full_posterior.stdev)
partial_precision_diff = update_percentage * (posterior_gaussian.precision - prior_gaussian.precision)
partial_precision_mean_diff = update_percentage * (posterior_gaussian.precision_mean - prior_gaussian.precision_mean)
partial_posterior_gaussian = Gaussian.from_precision_mean(
prior_gaussian.precision_mean + partial_precision_mean_diff,
prior_gaussian.precision + partial_precision_diff)
return Rating(partial_posterior_gaussian.mean,
partial_posterior_gaussian.stdev)
def partial_update(self, prior, full_posterior, update_percentage):
prior_gaussian = Gaussian(prior.mean, prior.stdev)
posterior_gaussian = Gaussian(full_posterior.mean, full_posterior.stdev)
partial_precision_diff = update_percentage * (posterior_gaussian.precision - prior_gaussian.precision)
partial_precision_mean_diff = update_percentage * (posterior_gaussian.precision_mean - prior_gaussian.precision_mean)
partial_posterior_gaussian = Gaussian.from_precision_mean(
prior_gaussian.precision_mean + partial_precision_mean_diff,
prior_gaussian.precision + partial_precision_diff)
return Rating(partial_posterior_gaussian.mean,
partial_posterior_gaussian.stdev)
def __init__(self, teams, team_ranks, game_info):
game_info = TrueSkillGameInfo.ensure_game_info(game_info)
FactorGraph.__init__(self)
self.prior_layer = PlayerPriorValuesToSkillsLayer(self, teams)
self.game_info = game_info
new_factory = VariableFactory(lambda: Gaussian.from_precision_mean(0.0, 0.0))
self.variable_factory = new_factory
self.layers = [
self.prior_layer,
PlayerSkillsToPerformancesLayer(self),
PlayerPerformancesToTeamPerformancesLayer(self),
IteratedTeamDifferencesInnerLayer(self,
TeamPerformancesToTeamPerformanceDifferencesLayer(self),
TeamDifferencesComparisonLayer(self, team_ranks))
]
def __init__(self, teams, team_ranks, game_info):
game_info = TrueSkillGameInfo.ensure_game_info(game_info)
FactorGraph.__init__(self)
self.prior_layer = PlayerPriorValuesToSkillsLayer(self, teams)
self.game_info = game_info
new_factory = VariableFactory(
lambda: Gaussian.from_precision_mean(0.0, 0.0))
self.variable_factory = new_factory
self.layers = [
self.prior_layer,
PlayerSkillsToPerformancesLayer(self),
PlayerPerformancesToTeamPerformancesLayer(self),
IteratedTeamDifferencesInnerLayer(
self, TeamPerformancesToTeamPerformanceDifferencesLayer(self),
TeamDifferencesComparisonLayer(self, team_ranks))
]
def update_helper(self, weights, weights_squared, messages, variables):
message0 = copy(messages[0].value)
marginal0 = copy(variables[0].value)
inverse_of_new_precision_sum = 0.0
another_inverse_of_new_precision_sum = 0.0
weighted_mean_sum = 0.0
another_weighted_mean_sum = 0.0
for i in range(len(weights_squared)):
inverse_of_new_precision_sum += weights_squared[i] / (
variables[i + 1].value.precision -
messages[i + 1].value.precision)
diff = variables[i + 1].value / messages[i + 1].value
another_inverse_of_new_precision_sum += weights_squared[
i] / diff.precision
weighted_mean_sum += (weights[i] *
(variables[i + 1].value.precision_mean -
messages[i + 1].value.precision_mean) /
(variables[i + 1].value.precision -
messages[i + 1].value.precision))
another_weighted_mean_sum += weights[
i] * diff.precision_mean / diff.precision
new_precision = 1.0 / inverse_of_new_precision_sum
#another_new_precision = 1.0 / another_inverse_of_new_precision_sum
new_precision_mean = new_precision * weighted_mean_sum
#another_new_precision_mean = another_new_precision * another_weighted_mean_sum
new_message = Gaussian.from_precision_mean(new_precision_mean,
new_precision)
old_marginal_without_message = marginal0 / message0
new_marginal = old_marginal_without_message * new_message
messages[0].value = new_message
variables[0].value = new_marginal
final_diff = new_marginal - marginal0
return final_diff
def update_helper(self, message1, message2, variable1, variable2):
message1_value = copy(message1.value)
message2_value = copy(message2.value)
marginal1 = copy(variable1.value)
marginal2 = copy(variable2.value)
a = self.precision / (self.precision + marginal2.precision - message2_value.precision)
new_message = Gaussian.from_precision_mean(
a * (marginal2.precision_mean - message2_value.precision_mean),
a * (marginal2.precision - message2_value.precision))
old_marginal_without_message = marginal1 / message1_value
new_marginal = old_marginal_without_message * new_message
message1.value = new_message
variable1.value = new_marginal
return new_marginal - marginal1
def update_helper(self, message1, message2, variable1, variable2):
message1_value = copy(message1.value)
message2_value = copy(message2.value)
marginal1 = copy(variable1.value)
marginal2 = copy(variable2.value)
a = self.precision / (self.precision + marginal2.precision -
message2_value.precision)
new_message = Gaussian.from_precision_mean(
a * (marginal2.precision_mean - message2_value.precision_mean),
a * (marginal2.precision - message2_value.precision))
old_marginal_without_message = marginal1 / message1_value
new_marginal = old_marginal_without_message * new_message
message1.value = new_message
variable1.value = new_marginal
return new_marginal - marginal1
def update_helper(self, weights, weights_squared, messages, variables):
message0 = copy(messages[0].value)
marginal0 = copy(variables[0].value)
inverse_of_new_precision_sum = 0.0
another_inverse_of_new_precision_sum = 0.0
weighted_mean_sum = 0.0
another_weighted_mean_sum = 0.0
for i in range(len(weights_squared)):
inverse_of_new_precision_sum += weights_squared[i] / (variables[i + 1].value.precision - messages[i + 1].value.precision)
diff = variables[i + 1].value / messages[i + 1].value
another_inverse_of_new_precision_sum += weights_squared[i] / diff.precision
weighted_mean_sum += (weights[i] *
(variables[i + 1].value.precision_mean - messages[i + 1].value.precision_mean) /
(variables[i + 1].value.precision - messages[i + 1].value.precision))
another_weighted_mean_sum += weights[i] * diff.precision_mean / diff.precision
new_precision = 1.0 / inverse_of_new_precision_sum
# another_new_precision = 1.0 / another_inverse_of_new_precision_sum
new_precision_mean = new_precision * weighted_mean_sum
# another_new_precision_mean = another_new_precision * another_weighted_mean_sum
new_message = Gaussian.from_precision_mean(new_precision_mean, new_precision)
old_marginal_without_message = marginal0 / message0
new_marginal = old_marginal_without_message * new_message
messages[0].value = new_message
variables[0].value = new_marginal
final_diff = new_marginal - marginal0
return final_diff
def __init__(self, mean, variance, variable):
GaussianFactor.__init__(self, "Prior value going to %s" % variable)
self.new_message = Gaussian(mean, sqrt(variance))
new_message = Message(Gaussian.from_precision_mean(0, 0),
"message from %s to %s" % (self, variable))
self.create_variable_to_message_binding_with_message(variable, new_message)
