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;
}
}
99905660