private static function mutate($indiv)
{
// Loop through genes
for ($i = 0; $i < $indiv->size(); $i++) {
if (algorithm::random() <= algorithm::$mutationRate) {
$gene = individual::$characters[rand(0, strlen(individual::$characters) - 1)];
// Create random gene
$indiv->setGene($i, $gene);
//substitute the gene into the individual
}
}
}
/**
* Glycemic Control Algorithm Test bench
*
* Creates algorithm object, sets algorithm parameters, runs algorithm and returns the algorithm object.
*
* @param $testNumber -
* @param $a1c - current a1c value
* @param $a1clast - last a1c value
* @param $a1cTarget - a1c target value
* @param $symptoms - Type II diabetes symptoms (true/false)
* @param $allergies - patient drug allergies
* @param $med1 - previous medicine 1
* @param $med2 - previous medicine 2
* @param $med3 - previous medicine 3
* @param $medhistory - patient medical history
*/
function algTest($testNumber, $a1c, $a1clast, $a1cTarget, $symptoms, $allergies, $med1, $med2, $med3, $medhistory)
{
echo "<br>";
echo "Test Number ";
echo $testNumber;
echo "<br>";
$alg1 = new algorithm();
// set A1c values
$alg1->setA1C($a1c);
$alg1->setA1Clast($a1clast);
$alg1->setA1CTarget($a1cTarget);
$alg1->setSymptoms($symptoms);
// Set allergies
$alg1->setAllergies($allergies);
// Set medicines
$alg1->setMedicine1($med1);
$alg1->setMedicine2($med2);
$alg1->setMedicine3($med3);
// set medical history
$alg1->setMedhistory($medhistory);
// print a1c values
$alg1->printA1C();
$alg1->printAllergies();
echo "<br>";
// run algorithm and print results
$alg1->gcAlgorithm();
echo "Results:";
echo "<br>";
$alg1->printResults();
return $alg1;
}
echo "\nmutationRate (what % of genes change for each mutate) :" . algorithm::$mutationRate;
echo "\nPoolSize (crossover # of individuals to select in each pool ):" . algorithm::$poolSize;
echo "\nInitial population # individuals:" . $initial_population_size;
echo "\nelitism (keep best individual each generation true=1) :" . algorithm::$elitism;
// Set a candidate solution static class
fitnesscalc::setSolution($solution_phrase);
echo "\nMax Fitness is :" . fitnesscalc::getMaxFitness();
echo "\n-----------------------------------------------";
// Create an initial population
$time1 = microtime(true);
$myPop = new population($initial_population_size, true);
// Evolve our population until we reach an optimum solution
while ($myPop->getFittest()->getFitness() > fitnesscalc::getMaxFitness()) {
$generationCount++;
$most_fit = $myPop->getFittest()->getFitness();
$myPop = algorithm::evolvePopulation($myPop);
//create a new generation
if ($most_fit < $most_fit_last) {
// echo " *** MOST FIT ".$most_fit." Most fit last".$most_fit_last;
echo "\n Generation: " . $generationCount . " (Stagnant:" . $generation_stagnant . ") Fittest: " . $most_fit . "/" . fitnesscalc::getMaxFitness();
echo " Best: " . $myPop->getFittest();
$most_fit_last = $most_fit;
$generation_stagnant = 0;
//reset stagnant generation counter
} else {
$generation_stagnant++;
}
//no improvement increment may want to end early
if ($generation_stagnant > algorithm::$max_generation_stagnant) {
echo "\n-- Ending TOO MANY (" . algorithm::$max_generation_stagnant . ") stagnant generations unchanged. Ending APPROX solution below \n..)";
break;
