牛顿迭代法应用

时间:2023-01-07 21:48:05

转载地址:http://www.cnblogs.com/javathread/archive/2011/12/26/2634731.html
在MIT公开课《计算机科学与编程导论》的第五讲中,讲到编写求解平方根的函数sqrt时,提到了牛顿迭代法。今天仔细一查,发现这是一个用途很广、很牛的计算方法。
首先,考虑如何编写一个开平方根的函数sqrt(float num, float e)。参数num是要求开平方根的实数,参数e是计算结果可以达到多大误差。这是一个无法得到精确解,只能求出近似解的问题。该如何编写呢?

1. 传统的二分法
我们可以先猜测一个值作为解,看这个值是否在误差范围内。如果没有达到误差要求,就构造一个更好的猜测值,继续迭代。猜测值可以简单地初始化为num/2,但怎样在下一轮迭代前构造一个更好的猜测值呢?我们不妨参照二分查找算法的思路,解的初始范围是[0, num],用二分法逐渐缩小范围。
     private static float sqrt(float num, float e) {                            float low = 0F;              float high = num;              float guess, e0;              int count = 0;                            do {                     guess = (low + high) / 2;                     if (guess*guess > num) {                           high = guess;                           e0 = guess*guess - num;                     } else {                           low = guess;                           e0 = num - guess*guess;                     }                                          count++;                     System.out.printf("Try %f, e: %f\n", guess, e0);              } while (e0 > e);
              System.out.printf("Try %d times, result: %f\n", count, guess);                            return guess;       }
在区间[low, high]内取中点(low+high)/2作为猜测值。如果guess*guess大于num,说明猜测值偏大,则在区间[low, guess]进行下一轮迭代,否则在区间[guess, high]中继续。当误差值e0小于能够接受的误差值e时停止迭代,返回结果。
取num=2, e=0.01进行测试,运行结果如下:Try 1.000000, e: 1.000000Try 1.500000, e: 0.250000Try 1.250000, e: 0.437500Try 1.375000, e: 0.109375Try 1.437500, e: 0.066406Try 1.406250, e: 0.022461Try 1.421875, e: 0.021729Try 1.414063, e: 0.000427Try 8 times, result: 1.414063可见尝试了八次才达到0.01的误差。

2. 神奇的牛顿法
仔细思考一下就能发现,我们需要解决的问题可以简单化理解。从函数意义上理解:我们是要求函数f(x) = x²,使f(x) = num的近似解,即x² - num = 0的近似解从几何意义上理解:我们是要求抛物线g(x) = x² - num与x轴交点(g(x) = 0)最接近的点。
我们假设g(x0)=0,即x0是正解,那么我们要做的就是让近似解x不断逼近x0,这是函数导数的定义:
牛顿迭代法应用
牛顿迭代法应用
可以由此得到
牛顿迭代法应用
牛顿迭代法应用
从几何图形上看,因为导数是切线,通过不断迭代,导数与x轴的交点会不断逼近x0。
牛顿迭代法应用


3. 牛顿法的实现与测试
     public static void main(String[] args) {              float num = 2;              float e = 0.01F;              sqrt(num, e);              sqrtNewton(num, e);                            num = 2;              e = 0.0001F;              sqrt(num, e);              sqrtNewton(num, e);                            num = 2;              e = 0.00001F;              sqrt(num, e);              sqrtNewton(num, e);       }
     private static float sqrtNewton(float num, float e) {                            float guess = num / 2;              float e0;              int count = 0;                            do {                     guess = (guess + num / guess) / 2;                     e0 = guess*guess - num;                                          count++;                     System.out.printf("Try %f, e: %f\n", guess, e0);              } while (e0 > e);
              System.out.printf("Try %d times, result: %f\n", count, guess);                            return guess;       }
与二分法的对比测试结果:
Try 1.000000, e: 1.000000Try 1.500000, e: 0.250000Try 1.250000, e: 0.437500Try 1.375000, e: 0.109375Try 1.437500, e: 0.066406Try 1.406250, e: 0.022461Try 1.421875, e: 0.021729Try 1.414063, e: 0.000427Try 8 times, result: 1.414063
Try 1.500000, e: 0.250000Try 1.416667, e: 0.006945Try 2 times, result: 1.416667
Try 1.000000, e: 1.000000Try 1.500000, e: 0.250000Try 1.250000, e: 0.437500Try 1.375000, e: 0.109375Try 1.437500, e: 0.066406Try 1.406250, e: 0.022461Try 1.421875, e: 0.021729Try 1.414063, e: 0.000427Try 1.417969, e: 0.010635Try 1.416016, e: 0.005100Try 1.415039, e: 0.002336Try 1.414551, e: 0.000954Try 1.414307, e: 0.000263Try 1.414185, e: 0.000082Try 14 times, result: 1.414185
Try 1.500000, e: 0.250000Try 1.416667, e: 0.006945Try 1.414216, e: 0.000006Try 3 times, result: 1.414216
Try 1.000000, e: 1.000000Try 1.500000, e: 0.250000Try 1.250000, e: 0.437500Try 1.375000, e: 0.109375Try 1.437500, e: 0.066406Try 1.406250, e: 0.022461Try 1.421875, e: 0.021729Try 1.414063, e: 0.000427Try 1.417969, e: 0.010635Try 1.416016, e: 0.005100Try 1.415039, e: 0.002336Try 1.414551, e: 0.000954Try 1.414307, e: 0.000263Try 1.414185, e: 0.000082Try 1.414246, e: 0.000091Try 1.414215, e: 0.000004Try 16 times, result: 1.414215
Try 1.500000, e: 0.250000Try 1.416667, e: 0.006945Try 1.414216, e: 0.000006Try 3 times, result: 1.414216
可以看到随着对误差要求的更加精确,二分法的效率很低下,而牛顿法的确非常高效。可在两三次内得到结果。
如果搞不清牛顿法的具体原理,可能就要像我一样复习下数学知识了。极限、导数、泰勒展开式、单变量微分等。

4. 更快的方法
在Quake源码中有段求sqrt的方法,大概思路是只进行一次牛顿迭代,得到能够接受误差范围内的结果。因此肯定是更快的。
float Q_rsqrt( float number )
{
  long i;
  float x2, y;
  const float threehalfs = 1.5F;

  x2 = number * 0.5F;
  y  = number;
  i  = * ( long * ) &y;  // evil floating point bit level hacking
  i  = 0x5f3759df - ( i >> 1 ); // what the fuck?
  y  = * ( float * ) &i;
  y  = y * ( threehalfs - ( x2 * y * y ) ); // 1st iteration
  // y  = y * ( threehalfs - ( x2 * y * y ) ); // 2nd iteration, this can be removed

  #ifndef Q3_VM
  #ifdef __linux__
    assert( !isnan(y) ); // bk010122 - FPE?
  #endif
  #endif
  return y;
}