Java内部实现：标准正态分布。即：Polar Method方法

public double nextDouble() {

return (((long)(next(26)) << 27) + next(27))

/ (double)(1L << 53);

}

private double nextNextGaussian;

private boolean haveNextNextGaussian = false;

synchronized public double nextGaussian() {

// See Knuth, ACP, Section 3.4.1 Algorithm C.

if (haveNextNextGaussian) {

haveNextNextGaussian = false;

return nextNextGaussian;

} else {

double v1, v2, s;

do {

v1 = 2 * nextDouble() - 1; // between -1 and 1

v2 = 2 * nextDouble() - 1; // between -1 and 1

s = v1 * v1 + v2 * v2;

} while (s >= 1 || s == 0);

double multiplier = StrictMath.sqrt(-2 * StrictMath.log(s)/s);

nextNextGaussian = v2 * multiplier;

haveNextNextGaussian = true;

return v1 * multiplier;

}

}