PHP algorithm for generating random red envelopes

Basic idea

In terms of random number generation, I learned from this blogger @The miserable uncle Idea:

Original text: For example, if you want to distribute 1 red envelope to N people, it is actually equivalent to getting N percentage data. The condition is that the sum of these N percentages = 100/100. The average of these N percentages is 1/N. And these N percentage data conform to a normal distribution (most values are relatively close to the mean).

Interpretation: For example, if I have 1,000 yuan and send out 50 red envelopes, I will first randomly pick out 50 numbers, and then calculate the average avg of these 50 numbers, using avg/(1/N), and you will get A base number mixrand, and then divide the 50 randomly generated numbers by the mixrand to get the percentage randVal of each number relative to the base, and then multiply randVal by 1,000 yuan to get the specific amount of each red envelope.

Algorithm implementation

Talk is cheap, show me your code!

Core generation algorithm:

rewardMoney = $rewardMoney * 100 - $rewardNum * $min; $this->rewardNum = $rewardNum; // 计算出发出红包的平均概率值、精确到小数4位。 $avgRand = 1 / $this->rewardNum; $randArr = []; // 定义生成的数据总合sum $sum = 0; $t_count = 0; while ($t_count < $rewardNum) { // 随机产出四个区间的额度 $c = rand(1, 100); if ($c < 15) { $t = round(sqrt(mt_rand(1, 1500))); } else if ($c < 65) { $t = round(sqrt(mt_rand(1500, 6500))); } else if ($c < 95) { $t = round(sqrt(mt_rand(6500, 9500))); } else { $t = round(sqrt(mt_rand(9500, 10000))); } ++$t_count; $sum += $t; $randArr[] = $t; } // 计算当前生成的随机数的平均值，保留4位小数 $randAll = round($sum / $rewardNum, 4); // 为将生成的随机数的平均值变成我们要的1/N，计算一下每个随机数要除以的总基数mixrand。此处可以约等处理，产生的误差后边会找齐 // 总基数 = 均值/平均概率 $mixrand = round($randAll / $avgRand, 4); // 对每一个随机数进行处理，并乘以总金额数来得出这个红包的金额。 $rewardArr = array(); foreach ($randArr as $key => $randVal) { // 单个红包所占比例randVal $randVal = round($randVal / $mixrand, 4); // 算出单个红包金额 $single = floor($this->rewardMoney * $randVal); // 小于最小值直接给最小值 if ($single < $min) { $single += $min; } // 大于最大值直接给最大值 if ($single > $max) { $single = $max; } // 将红包放入结果数组 $rewardArr[] = $single; } // 对比红包总数的差异、将差值放在第一个红包上 $rewardAll = array_sum($rewardArr); // 此处应使用真正的总金额rewardMoney，$rewardArr[0]可能小于0 $rewardArr[0] = $rewardMoney * 100 - ($rewardAll - $rewardArr[0]); // 第一个红包小于0时,做修正 if ($rewardArr[0] < 0) { rsort($rewardArr); $this->add($rewardArr, $min); } rsort($rewardArr); // 随机生成的最大值大于指定最大值 if ($rewardArr[0] > $max) { // 差额 $diff = 0; foreach ($rewardArr as $k => &$v) { if ($v > $max) { $diff += $v - $max; $v = $max; } else { break; } } $transfer = round($diff / ($this->rewardNum - $k + 1)); $this->diff($diff, $rewardArr, $max, $min, $transfer, $k); } return $rewardArr; } // 处理所有超过最大值的红包 public function diff($diff, &$rewardArr, $max, $min, $transfer, $k) { // 将多余的钱均摊给小于最大值的红包 for ($i = $k; $i < $this->rewardNum; $i++) { // 造随机值 if ($transfer > $min * 20) { $aa = rand($min, $min * 20); if ($i % 2) { $transfer += $aa; } else { $transfer -= $aa; } } if ($rewardArr[$i] + $transfer > $max) continue; if ($diff - $transfer < 0) { $rewardArr[$i] += $diff; $diff = 0; break; } $rewardArr[$i] += $transfer; $diff -= $transfer; } if ($diff > 0) { $i++; $this->diff($diff, $rewardArr, $max, $min, $transfer, $k); } } // 第一个红包小于0,从大红包上往下减 public function add(&$rewardArr, $min) { foreach ($rewardArr as &$re) { $dev = floor($re / $min); if ($dev > 2) { $transfer = $min * floor($dev / 2); $re -= $transfer; $rewardArr[$this->rewardNum - 1] += $transfer; } elseif ($dev == 2) { $re -= $min; $rewardArr[$this->rewardNum - 1] += $min; } else { break; } } if ($rewardArr[$this->rewardNum - 1] > $min || $rewardArr[$this->rewardNum - 1] == $min) { return; } else { $this->add($rewardArr, $min); } } }

Consider details

The code below is used to control specific business logic, and set aside fixed maximum and minimum red envelope amounts according to specific needs; When calling splitReward(total,num,max−0.01,min) in the code to generate red envelopes, the maximum value I passed in was reduced by 0.01. This ensures that the maximum value of the red envelope generated will never exceed the maximum value we set. value.

splitReward($total, $num, $max - 0.01, $min); sort($result_merge); $result_merge[1] = $result_merge[1] + $result_merge[0]; $result_merge[0] = $max * 100; foreach ($result_merge as &$v) { $v = floor($v) / 100; } return $result_merge; } }

Example test

Basic code

First set various initial values.

      
Performance test
      
Because of the limitation of memory_limit, we only measured the average value of 5 times, and the results were all around 1.6s.
      
for ($i=0; $i<5; $i++) { $time_start = microtime_float(); $reward_arr = $create_reward->random_red($total, $num, $max, $min); $time_end = microtime_float(); $time[] = $time_end - $time_start; } echo array_sum($time)/5; function microtime_float() { list($usec, $sec) = explode(" ", microtime()); return ((float)$usec + (float)$sec); }
      
Run results:
      
Data check
      
1) Whether the value is incorrect
      
Detected yes There are no negative values, is there a maximum value, how many maximum values are there, and is there any value smaller than the minimum value.
      
$reward_arr = $create_reward->random_red($total, $num, $max, $min); sort($reward_arr);//正序，最小的在前面 $sum = 0; $min_count = 0; $max_count = 0; foreach($reward_arr as $i => $val) { if ($i<3) { echo "
第".($i+1)."个红包，金额为：".$val."
"; } if ($val == $max) { $max_count++; } if ($val < $min) { $min_count++; } $val = $val*100; $sum += $val; } //检测钱是否全部发完 echo '
已生成红包总金额为：'.($sum/100).';总个数为：'.count($reward_arr).'
'; //检测有没有小于0的值 echo "
最大值:".($val/100).',共有'.$max_count.'个最大值，共有'.$min_count.'个值比最小值小';
      
Running results:
      
2) Normal distribution
      
Note that when drawing the picture, do not give too many red envelopes. Otherwise, the page cannot be rendered and will crash.
      
$reward_arr = $create_reward->random_red($total, $num, $max, $min); $show = array(); rsort($reward_arr); // 为了更直观的显示正态分布效果,需要将数组重新排序 foreach($reward_arr as $k=>$value) { $t=$k%2; if(!$t) $show[]=$value;; else array_unshift($show,$value); } echo "设定最大值为:".$max.',最小值为:'.$min.'
'; echo ""; foreach($show as $val) { // 线条长度计算 $width=intval($num*$val*300/$total); echo ""; } echo "
红包金额
图示
 {$val} 
";
      
Running results:
      

       
PS: A friend asked me whether the generated data has been verified mathematically to see if it conforms to the standard normal distribution, because I am not good at mathematics. I haven’t really thought about this, I just looked at it and thought it was him. Now that I have encountered this problem, I must solve it, so I used PHP's built-in function to calculate it. The calculated result is still relatively close to a normal distribution when the amount of data is small, but when the amount of data increases, it cannot Look, I don’t quite understand this. If you are interested, you can find out the reason.
       
Four functions of php: stats_standard_deviation (standard deviation), stats_variance (variance), stats_kurtosis (kurtosis), stats_skew (skewness). Using the above functions requires installing the stats extension.
      
      
The above is the detailed content of PHP algorithm for generating random red envelopes. For more information, please follow other related articles on the PHP Chinese website!