从Java中的数组生成对数正态分布

Question

I didn't find any help on the subject so I'm posting a new question about it. I have a variable containing three values {min, average, standard dev.}. How do I generate a lognormal distribution of this array that would randomly give me a value of the time from the lognormal distribution. Also If I would run it 1000 times I would want to randomly get a value of the time from the lognormal distribution each of the 1000 times. How would I write this in java code? Also I guess running it 1000 times would give me an average of the average in the array?

Answer 1

First generate standard normal values and convert them to normal distribution with given parameters. Finally raise to exponential to get log-normal distribution with given mean and std dev.

Random rng = new Random(0);
double[] lognormalValues = new double[1000];

for (int i = 0; i < 1000; ++i)
{
    stdNormal   = rng.nextGaussian();
    normalValue = stddev*stdNormal + mean;

    logNormalValues[i] = Math.exp(normalValue);
}

Answer 2

you can also get a built in function here. in your above example, you won't actually end up with the stddev and mean noted -- they change when you exp() the values.

http://commons.apache.org/proper/commons-math/apidocs/org/apache/commons/math3/distribution/LogNormalDistribution.html

Answer 3

or this

public static double LogNormal(double mean, double stddev) {
    Random randGen = new Random();
    double varx = Math.pow(stddev, 2);
    double ess = Math.log(1.0 + (varx/Math.pow(mean,2)));
    double mu = Math.log(mean) - (0.5*Math.pow(ess, 2));
    return Math.pow(2.71828, (mu+(ess*randGen.nextGaussian() ) ) );

}