I've attached a prototype diff-proposal where I've changed AbstractContinuousDistribution and AbstractIntegerDistribution, and implemented the nextExponential on the ExponentialDistribution as an example of overwriting the nextSample-method from the Abstract-distributions.
Only AbstractContinuousDistribution and AbstractIntegerDistribution extends AbstractDistribution, and both AbstractIntegerDistribution and AbstractContinuousDistribution has an inverseCumulativeProbability-function. But in AbstractContinuousDistribution the inverse cdf returns a double, and at AbstractIntegerDistribution it - naturally - returns an integer. Therefor the nextSample is not put on AbstractDistribution, but on each extension with different return types. RandomGenerator as parameter instead of getting a RNG inside the nextSample, because one typically wants to use the same RNG because often several random samples are wanted. Another option is to have a RNG as a field in the class, but that would be more ugly and also result in several RNGs at runtime. I found an error as well. In RandomDataImpl, mean < 0 throws exception, whereas in the distribution implementation an exception is thrown if mean <= 0. RandomDataImpl is corrected to <= in because that is the correct definition. The function nextExponential couldn't - without further notice - be removed from RandomDataImpl, because nextPoisson and the ValueServer uses the function directly. So here is an issue. nextPoisson can create a ExponentialDistribution and sample from that. ValueServer could do the same, but what to you think about it? As mentioned, the diff is only a prototype hence no formal patch submitted. And no tests has been made neither. But should I report a bug or supply a separate patch on the small mean-problem in nextExponential in RandomDataImpl? Cheers, Mikkel.
--------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
