[RNG-134] BoxSampler to sampler uniformly from a box (or hyperrectangle)

• Type: New Feature
• Status: Closed
• Priority: Minor
• Resolution: Implemented
• Affects Version/s: 1.4
• Fix Version/s: 1.4
• Component/s: sampling
• Assignee: Alex Herbert
• Reporter: Alex Herbert
• Created: 30/Apr/21 15:48
• Updated: 13/Sep/21 10:41
• Resolved: 06/May/21 19:56

Description

Create a sampler to sample uniformly within a hyperrectangle.

Alternative names for a hyperrectangle are a box or an orthotope. Below I use the term box for simplicity and to match the BallSampler for an n-ball.
 

public abstract class BoxSampler implements 
        SharedStateSampler<BoxSampler> {
    public static BoxSampler of(double[] a,
                                double[] b,
                                UniformRandomProvider rng);
}

Inputs a and b are the minimum and maximum of the box in each dimension.

Sampling can be performed using the same method as the LineSampler but using a uniform deviate for each dimension (instead of the same deviate for all dimensions):

p = a + u * (b - a)
  = a + ub - ua
  = (1 - u)a + ub

This will produce the same results as using an instance of the the ContinuousUniformSampler for each dimension with all samplers using a common RNG.
 

Issue Links

• is a child of: RNG-132 Package rng.sampling.shape for sampling coordinates from shapes