jct/trueskill
1
0
Fork 0
mirror of https://github.com/furyfire/trueskill.git synced 2025-04-19 20:34:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

General cleanup and removal of all unnecessary references

Browse Source
This commit is contained in:
Alexander Liljengård
2016-05-24 16:31:21 +02:00
parent 519ad85fad
commit a65f2aa9f3
43 changed files with 366 additions and 386 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/TrueSkill/Factors/GaussianFactor.php
View File

@ -18,19 +18,19 @@ abstract class GaussianFactor extends Factor
* @param Variable $variable
* @return float|int
*/
protected function sendMessageVariable(Message &$message, Variable &$variable)
protected function sendMessageVariable(Message $message, Variable $variable)
{
$marginal = &$variable->getValue();
$messageValue = &$message->getValue();
$marginal = $variable->getValue();
$messageValue = $message->getValue();
$logZ = GaussianDistribution::logProductNormalization($marginal, $messageValue);
$variable->setValue(GaussianDistribution::multiply($marginal, $messageValue));
return $logZ;
}
public function &createVariableToMessageBinding(Variable &$variable)
public function createVariableToMessageBinding(Variable $variable)
{
$newDistribution = GaussianDistribution::fromPrecisionMean(0, 0);
$binding = &parent::createVariableToMessageBindingWithMessage($variable,
$binding = parent::createVariableToMessageBindingWithMessage($variable,
new Message(
$newDistribution,
sprintf("message from %s to %s", $this, $variable)));

32
src/TrueSkill/Factors/GaussianGreaterThanFactor.php
View File

@ -14,7 +14,7 @@ class GaussianGreaterThanFactor extends GaussianFactor
{
private $_epsilon;
public function __construct($epsilon, Variable &$variable)
public function __construct($epsilon, Variable $variable)
{
parent::__construct(\sprintf("%s > %.2f", $variable, $epsilon));
$this->_epsilon = $epsilon;
@ -23,20 +23,26 @@ class GaussianGreaterThanFactor extends GaussianFactor
public function getLogNormalization()
{
$vars = &$this->getVariables();
$marginal = &$vars[0]->getValue();
$messages = &$this->getMessages();
$message = &$messages[0]->getValue();
/** @var Variable[] $vars */
$vars = $this->getVariables();
$marginal = $vars[0]->getValue();
/** @var Message[] $messages */
$messages = $this->getMessages();
$message = $messages[0]->getValue();
$messageFromVariable = GaussianDistribution::divide($marginal, $message);
return -GaussianDistribution::logProductNormalization($messageFromVariable, $message)
+
log(
GaussianDistribution::cumulativeTo(($messageFromVariable->getMean() - $this->_epsilon) /
$messageFromVariable->getStandardDeviation()));
GaussianDistribution::cumulativeTo(
($messageFromVariable->getMean() - $this->_epsilon) /
$messageFromVariable->getStandardDeviation()
)
);
}
protected function updateMessageVariable(Message &$message, Variable &$variable)
protected function updateMessageVariable(Message $message, Variable $variable)
{
$oldMarginal = clone $variable->getValue();
$oldMessage = clone $message->getValue();
@ -55,16 +61,18 @@ class GaussianGreaterThanFactor extends GaussianFactor
$denom = 1.0 - TruncatedGaussianCorrectionFunctions::wExceedsMargin($dOnSqrtC, $epsilsonTimesSqrtC);
$newPrecision = $c / $denom;
$newPrecisionMean = ($d +
$newPrecisionMean = (
$d +
$sqrtC *
TruncatedGaussianCorrectionFunctions::vExceedsMargin($dOnSqrtC, $epsilsonTimesSqrtC)) /
$denom;
TruncatedGaussianCorrectionFunctions::vExceedsMargin($dOnSqrtC, $epsilsonTimesSqrtC)
) / $denom;
$newMarginal = GaussianDistribution::fromPrecisionMean($newPrecisionMean, $newPrecision);
$newMessage = GaussianDistribution::divide(
GaussianDistribution::multiply($oldMessage, $newMarginal),
$oldMarginal);
$oldMarginal
);
// Update the message and marginal
$message->setValue($newMessage);

22
src/TrueSkill/Factors/GaussianLikelihoodFactor.php
View File

@ -1,6 +1,7 @@
<?php namespace Moserware\Skills\TrueSkill\Factors;
use Exception;
use Moserware\Skills\FactorGraphs\KeyedVariable;
use Moserware\Skills\FactorGraphs\Message;
use Moserware\Skills\FactorGraphs\Variable;
use Moserware\Skills\Numerics\GaussianDistribution;
@ -14,7 +15,7 @@ class GaussianLikelihoodFactor extends GaussianFactor
{
private $_precision;
public function __construct($betaSquared, Variable &$variable1, Variable &$variable2)
public function __construct($betaSquared, Variable $variable1, Variable $variable2)
{
parent::__construct(sprintf("Likelihood of %s going to %s", $variable2, $variable1));
$this->_precision = 1.0 / $betaSquared;
@ -24,16 +25,18 @@ class GaussianLikelihoodFactor extends GaussianFactor
public function getLogNormalization()
{
$vars = &$this->getVariables();
$messages = &$this->getMessages();
/** @var KeyedVariable[]|mixed $vars */
$vars = $this->getVariables();
/** @var Message[] $messages */
$messages = $this->getMessages();
return GaussianDistribution::logRatioNormalization(
$vars[0]->getValue(),
$messages[0]->getValue());
$messages[0]->getValue()
);
}
private function updateHelper(Message &$message1, Message &$message2,
Variable &$variable1, Variable &$variable2)
private function updateHelper(Message $message1, Message $message2, Variable $variable1, Variable $variable2)
{
$message1Value = clone $message1->getValue();
$message2Value = clone $message2->getValue();
@ -45,7 +48,8 @@ class GaussianLikelihoodFactor extends GaussianFactor
$newMessage = GaussianDistribution::fromPrecisionMean(
$a * ($marginal2->getPrecisionMean() - $message2Value->getPrecisionMean()),
$a * ($marginal2->getPrecision() - $message2Value->getPrecision()));
$a * ($marginal2->getPrecision() - $message2Value->getPrecision())
);
$oldMarginalWithoutMessage = GaussianDistribution::divide($marginal1, $message1Value);
@ -62,8 +66,8 @@ class GaussianLikelihoodFactor extends GaussianFactor
public function updateMessageIndex($messageIndex)
{
$messages = &$this->getMessages();
$vars = &$this->getVariables();
$messages = $this->getMessages();
$vars = $this->getVariables();
switch ($messageIndex) {
case 0:

15
src/TrueSkill/Factors/GaussianPriorFactor.php
View File

@ -13,7 +13,7 @@ class GaussianPriorFactor extends GaussianFactor
{
private $_newMessage;
public function __construct($mean, $variance, Variable &$variable)
public function __construct($mean, $variance, Variable $variable)
{
parent::__construct(sprintf("Prior value going to %s", $variable));
$this->_newMessage = new GaussianDistribution($mean, sqrt($variance));
@ -23,18 +23,17 @@ class GaussianPriorFactor extends GaussianFactor
$this->createVariableToMessageBindingWithMessage($variable, $newMessage);
}
protected function updateMessageVariable(Message &$message, Variable &$variable)
protected function updateMessageVariable(Message $message, Variable $variable)
{
$oldMarginal = clone $variable->getValue();
$oldMessage = $message;
$newMarginal =
GaussianDistribution::fromPrecisionMean(
$oldMarginal->getPrecisionMean() + $this->_newMessage->getPrecisionMean() - $oldMessage->getValue()->getPrecisionMean(),
$oldMarginal->getPrecision() + $this->_newMessage->getPrecision() - $oldMessage->getValue()->getPrecision());
$newMarginal = GaussianDistribution::fromPrecisionMean(
$oldMarginal->getPrecisionMean() + $this->_newMessage->getPrecisionMean() - $oldMessage->getValue()->getPrecisionMean(),
$oldMarginal->getPrecision() + $this->_newMessage->getPrecision() - $oldMessage->getValue()->getPrecision()
);
$variable->setValue($newMarginal);
$newMessage = &$this->_newMessage;
$message->setValue($newMessage);
$message->setValue($this->_newMessage);
return GaussianDistribution::subtract($oldMarginal, $newMarginal);
}
}

26
src/TrueSkill/Factors/GaussianWeightedSumFactor.php
View File

@ -20,7 +20,7 @@ class GaussianWeightedSumFactor extends GaussianFactor
private $_weights;
private $_weightsSquared;
public function __construct(Variable &$sumVariable, array &$variablesToSum, array &$variableWeights = null)
public function __construct(Variable $sumVariable, array $variablesToSum, array $variableWeights = null)
{
parent::__construct(self::createName($sumVariable, $variablesToSum, $variableWeights));
$this->_weights = array();
@ -29,7 +29,7 @@ class GaussianWeightedSumFactor extends GaussianFactor
// The first weights are a straightforward copy
// v_0 = a_1*v_1 + a_2*v_2 + ... + a_n * v_n
$variableWeightsLength = count($variableWeights);
$this->_weights[0] = \array_fill(0, count($variableWeights), 0);
$this->_weights[0] = array_fill(0, count($variableWeights), 0);
for ($i = 0; $i < $variableWeightsLength; $i++) {
$weight = &$variableWeights[$i];
@ -104,16 +104,16 @@ class GaussianWeightedSumFactor extends GaussianFactor
$this->createVariableToMessageBinding($sumVariable);
foreach ($variablesToSum as &$currentVariable) {
$localCurrentVariable = &$currentVariable;
foreach ($variablesToSum as $currentVariable) {
$localCurrentVariable = $currentVariable;
$this->createVariableToMessageBinding($localCurrentVariable);
}
}
public function getLogNormalization()
{
$vars = &$this->getVariables();
$messages = &$this->getMessages();
$vars = $this->getVariables();
$messages = $this->getMessages();
$result = 0.0;
@ -126,9 +126,7 @@ class GaussianWeightedSumFactor extends GaussianFactor
return $result;
}
private function updateHelper(array &$weights, array &$weightsSquared,
array &$messages,
array &$variables)
private function updateHelper(array $weights, array $weightsSquared, array $messages, array $variables)
{
// Potentially look at http://mathworld.wolfram.com/NormalSumDistribution.html for clues as
// to what it's doing
@ -185,23 +183,23 @@ class GaussianWeightedSumFactor extends GaussianFactor
public function updateMessageIndex($messageIndex)
{
$allMessages = &$this->getMessages();
$allVariables = &$this->getVariables();
$allMessages = $this->getMessages();
$allVariables = $this->getVariables();
Guard::argumentIsValidIndex($messageIndex, count($allMessages), "messageIndex");
$updatedMessages = array();
$updatedVariables = array();
$indicesToUse = &$this->_variableIndexOrdersForWeights[$messageIndex];
$indicesToUse = $this->_variableIndexOrdersForWeights[$messageIndex];
// The tricky part here is that we have to put the messages and variables in the same
// order as the weights. Thankfully, the weights and messages share the same index numbers,
// so we just need to make sure they're consistent
$allMessagesCount = count($allMessages);
for ($i = 0; $i < $allMessagesCount; $i++) {
$updatedMessages[] = &$allMessages[$indicesToUse[$i]];
$updatedVariables[] = &$allVariables[$indicesToUse[$i]];
$updatedMessages[] = $allMessages[$indicesToUse[$i]];
$updatedVariables[] = $allVariables[$indicesToUse[$i]];
}
return $this->updateHelper($this->_weights[$messageIndex],

25
src/TrueSkill/Factors/GaussianWithinFactor.php
View File

@ -14,7 +14,7 @@ class GaussianWithinFactor extends GaussianFactor
{
private $_epsilon;
public function __construct($epsilon, Variable &$variable)
public function __construct($epsilon, Variable $variable)
{
parent::__construct(sprintf("%s <= %.2f", $variable, $epsilon));
$this->_epsilon = $epsilon;
@ -23,11 +23,13 @@ class GaussianWithinFactor extends GaussianFactor
public function getLogNormalization()
{
$variables = &$this->getVariables();
$marginal = &$variables[0]->getValue();
/** @var Variable[] $variables */
$variables = $this->getVariables();
$marginal = $variables[0]->getValue();
$messages = &$this->getMessages();
$message = &$messages[0]->getValue();
/** @var Message[] $messages */
$messages = $this->getMessages();
$message = $messages[0]->getValue();
$messageFromVariable = GaussianDistribution::divide($marginal, $message);
$mean = $messageFromVariable->getMean();
$std = $messageFromVariable->getStandardDeviation();
@ -38,7 +40,7 @@ class GaussianWithinFactor extends GaussianFactor
return -GaussianDistribution::logProductNormalization($messageFromVariable, $message) + log($z);
}
protected function updateMessageVariable(Message &$message, Variable &$variable)
protected function updateMessageVariable(Message $message, Variable $variable)
{
$oldMarginal = clone $variable->getValue();
$oldMessage = clone $message->getValue();
@ -55,15 +57,16 @@ class GaussianWithinFactor extends GaussianFactor
$denominator = 1.0 - TruncatedGaussianCorrectionFunctions::wWithinMargin($dOnSqrtC, $epsilonTimesSqrtC);
$newPrecision = $c / $denominator;
$newPrecisionMean = ($d +
$sqrtC *
TruncatedGaussianCorrectionFunctions::vWithinMargin($dOnSqrtC, $epsilonTimesSqrtC)) /
$denominator;
$newPrecisionMean = ( $d +
$sqrtC *
TruncatedGaussianCorrectionFunctions::vWithinMargin($dOnSqrtC, $epsilonTimesSqrtC)
) / $denominator;
$newMarginal = GaussianDistribution::fromPrecisionMean($newPrecisionMean, $newPrecision);
$newMessage = GaussianDistribution::divide(
GaussianDistribution::multiply($oldMessage, $newMarginal),
$oldMarginal);
$oldMarginal
);
// Update the message and marginal
$message->setValue($newMessage);