$set1 = randomSet(110000);
$set2 = randomSet(0);
$set3 = randomSet(120000);
echo "Number of words in set 1: " . count($set1) . "\n";
echo "Number of words in set 2: " . count($set2) . "\n";
echo "Number of words in set 3: " . count($set3) . "\n";
echo "------\n";
echo "Cardinailiy of set 1: " . cardinality($set1) . "\n";
echo "Cardinailiy of set 2: " . cardinality($set2) . "\n";
echo "Cardinailiy of set 3: " . cardinality($set3) . "\n";
$intersection = array_intersect($set1, $set2, $set3);
$union = array_merge($set1, $set2, $set3);
$intersectionCount = cardinality($intersection);
echo "Cardinailiy of union: " . cardinality($union) . "\n";
echo "Cardinailiy of intersection: " . $intersectionCount . "\n";
echo "------\nLogLog\n";
$log_logs = array();
foreach (array($set1, $set2, $set3) as $i => $set) {
$log_log = new HyperLogLog\MinHash();
foreach ($set as $word) {
$log_log->add($word);
}
$log_logs[] = $log_log;
echo "Added set " . ($i + 1) . "\n";
}
$count = \HyperLogLog\Utils\MinHashIntersector::count($log_logs);
echo "intersection complete: count: {$count}\n";
if ($count) {
echo "Error: 100% - count should be zero\n";
}
echo "Error: 0% - count is zero\n";
public function test($repeat = 100)
{
while ($repeat--) {
$ll1 = new HyperLogLog\MinHash();
$ll2 = new HyperLogLog\MinHash();
$i = 100000000 + $this->random();
$r = mt_rand(1, 4);
$end = $i + $this->i * $r;
$actual = 0;
$overlap = 0;
while ($i <= $end) {
$ll1->add($i);
if (++$overlap === 2) {
$overlap = 0;
$ll2->add($i);
$actual++;
}
$i += $r;
}
$intersection = \HyperLogLog\Utils\MinHashIntersector::count(array($ll1, $ll2));
$ll1->union($ll2);
$total = $ll1->count();
$this->average[0] += $actual;
$this->average[1] += $intersection;
$this->average[2] += $total;
$this->results[] = array($actual, $intersection, $total);
}
}
public function test($repeat = 100)
{
while ($repeat--) {
$keep1 = array();
$keep2 = array();
$ll1 = new HyperLogLog\MinHash();
$ll2 = new HyperLogLog\MinHash();
$total = 0;
while (1) {
$total++;
$rand = $this->random();
$keep1[$rand] = 1;
$ll1->add($rand);
$rand = $this->random();
$keep2[$rand] = 1;
$ll2->add($rand);
if (($count = count($keep2)) >= $this->i) {
break;
}
}
$intersection = \HyperLogLog\Utils\MinHashIntersector::count(array($ll1, $ll2));
$actual = count(array_intersect_key($keep1, $keep2));
if ($actual == 0 || $intersection == 0) {
continue;
}
$ll1->union($ll2);
$total = $ll1->count();
$this->average[0] += $actual;
$this->average[1] += $intersection;
$this->average[2] += $total;
$this->results[] = array($actual, $intersection, $total);
}
}
include __DIR__ . '/../vendor/autoload.php';
include __DIR__ . '/randomGenerator.php';
$set1 = randomSet(1000);
$set2 = randomSet(1000);
$set3 = randomSet(1000);
echo "Number of words in set 1: " . count($set1) . "\n";
echo "Number of words in set 2: " . count($set2) . "\n";
echo "Number of words in set 3: " . count($set3) . "\n";
$intersection = array_intersect($set1, $set2, $set3);
$union = array_merge($set1, $set2, $set3);
$intersectionCount = cardinality($intersection);
echo "Cardinailiy of union: " . cardinality($union) . "\n";
echo "Number of words in intersection: " . $intersectionCount . "\n";
echo "------\nLogLog\n";
$log_logs = array();
foreach (array($set1, $set2, $set3) as $i => $set) {
$log_log = new HyperLogLog\MinHash();
foreach ($set as $word) {
$log_log->add($word);
}
$log_logs[] = $log_log;
echo "Added set " . ($i + 1) . "\n";
}
list($minHashIntersection, $minHashK, $hllUnion) = \HyperLogLog\Utils\MinHashIntersector::jaccard($log_logs);
$hllUnionCount = $hllUnion->count();
echo "Hll union: " . $hllUnionCount . "\n";
echo "Min hash intersection: " . $minHashIntersection . "\n";
echo "Min hash k: " . $minHashK . "\n";
$count = $minHashIntersection / $minHashK * $hllUnionCount;
echo "intersection complete\n";
echo $count . "\n" . 'error: ' . number_format(($count - $intersectionCount) / ($intersectionCount / 100.0), 3) . '%' . PHP_EOL;