[
https://issues.apache.org/jira/browse/RNG-145?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17362980#comment-17362980
]
Alex Herbert commented on RNG-145:
----------------------------------
I am OK with this. I think this is valid:
{code:java}
public static SharedStateContinuousSampler of(UniformRandomProvider rng,
double lo,
double hi,
boolean excludeBounds) {
{code}
Currently the underlying nextDouble() method typically returns in the interval
[0, 1). So the current ContinuousUniformSampler will return in the interval
[lower, upper). The sampler is robust to a swap of the arguments so you can get
the alternative (lower, upper] by swapping the arguments to the constructor.
The sample is generated using:
{code:java}
final double u = rng.nextDouble();
return u * hi + (1 - u) * lo;
{code}
So if the nextDouble() does return either 0 or 1 you get one of the bounds.
The current method in Commons Math trusts the implementation of nextDouble does
not return 1 and only checks for 0.
To ensure we generate in the open interval (lower, upper) I think requires not
using (trusting) nextDouble() and instead using nextLong() and eliminating 0
from the call to (nextLong() >>> 11). From the 2^53 -1 numbers that are left
you can create a double in (0, 1).
> "ContinuousUniformSampler" with both bounds excluded
> ----------------------------------------------------
>
> Key: RNG-145
> URL: https://issues.apache.org/jira/browse/RNG-145
> Project: Commons RNG
> Issue Type: New Feature
> Components: sampling
> Reporter: Gilles Sadowski
> Priority: Major
> Fix For: 1.4
>
>
> In class {{RandomUtils}}, Commons Math provides
> {code}
> public double nextUniform(double lo, double hi) { /* ... */ }
> {code}
> where both bounds are excluded.
> Should the feature be added to {{ContinuousUniformSampler}}?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
