[ 
https://issues.apache.org/jira/browse/RNG-179?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17541883#comment-17541883
 ] 

Alex Herbert commented on RNG-179:
----------------------------------

For reference the API is:
{code:java}
UniformRandomProvider rng = ...;

// Represented exactly
long[] frequencies = {1, 1, 2, 3, 1};
DiscreteSampler sampler1 = FastLoadedDiceRollerSampler.of(rng, frequencies);

// Represented exactly using relative weights
double[] weights = {1, 0.5, 0.125, 3.5, 1};
DiscreteSampler sampler2 = FastLoadedDiceRollerSampler.of(rng, weights);
{code}
When using frequencies the maximum number of categories is approximately ((2^31 
- 1) / 63) = 34,087,042 and the sum of frequencies must be below Long.MAX_VALUE.

When using weights the maximum number of categories is approximately 1,000,000 
if all weights are Double.MAX_VALUE with the exception of 1 weight of 
Double.MIN_VALUE. In practice this range of weights is unrealistic and more 
categories should be possible. The construction algorithm does not require the 
weights to sum to a finite double value.

> The Fast Loaded Dice Roller: A Near-Optimal Exact Sampler for Discrete 
> Probability Distributions
> ------------------------------------------------------------------------------------------------
>
>                 Key: RNG-179
>                 URL: https://issues.apache.org/jira/browse/RNG-179
>             Project: Commons RNG
>          Issue Type: New Feature
>          Components: sampling
>            Reporter: Vladimir Sitnikov
>            Priority: Major
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> It might make sense to implement Fast Loaded Dice Roller sampler.
> See https://arxiv.org/pdf/2003.03830v2.pdf, 
> https://github.com/probcomp/fast-loaded-dice-roller
> The authors claim that Fast Loaded Dice Roller is faster than the Alias 
> Method.



--
This message was sent by Atlassian Jira
(v8.20.7#820007)

Reply via email to