ikr.simlib.distributions.continuous

Class EmpiricalDistribution

java.lang.Object

ikr.simlib.distributions.Distribution

ikr.simlib.distributions.continuous.ContinuousDistribution

ikr.simlib.distributions.continuous.EmpiricalDistribution

public class EmpiricalDistribution
extends ContinuousDistribution

EmpiricalDistribution allows to read empirical distributions obtained e.g.from measurements into a simulation program.

This is useful if no mathematical fit can be found to describe a distribution or if a measured distribution has to be integrated into a simulation very fast.

EmpiricalDistribution is a continuous distribution and requires 2-tuples as input data: value and cdf-probability. This allows to have arbitrarily spaced data.

Similar to GeneralDistribution it relies on the DistributionVector. This reads the 2-tuple-vector and generates 2 separate vectors for values and probabilities. Using the same search algorithm as ProbabilityVector it determines the upper bound of the determined bin and then does a linear interpolation between that bin and the next lower bin (if any) to determine the fraction of the bin the search value points to. The return value is determined by calculating that same fraction between the corresponding vector elements of the value vector.

      |                                           x
      |                               x       
  RND>| -  -  -  -  -  -  -  -  -  . `                                  
      |                         . `| 
      |               x   x   x    |               
      |       x                    |            
      ----x---+---+---+---+---+---+---+---+---+---+-->
                                   v
                                  RV
 

The simulation par-file should look like the following:

  EmpiricalDist {
    CDF = {
      0.000 ,   0.00000000 ,
      1.000 ,   0.05000000 ,
      2.000 ,   0.10000000 ,
      10.000 ,  0.25000000 ,
      20.000 ,  0.30000000 ,
      30.000 ,  0.50000000 ,
      40.000 ,  0.60000000 ,
      50.000 ,  0.80000000 ,
      100.000 , 0.95000000 ,
      110.000 , 1.00000000
    };
  }
 

Notes:

• It is best always give the final value of 1.0.
• The first entry must be (x|0.0), where x is an arbitrary double.
• Avoid identical values in the last 2 entries.

See Also:

GeneralDistribution

Field Summary

Fields inherited from class ikr.simlib.distributions.Distribution

rng

Constructor Summary

Constructors

Constructor and Description
EmpiricalDistribution(double[] vector)
EmpiricalDistribution(double[] vector, RandomNumberGenerator rng)
EmpiricalDistribution(java.util.List<java.lang.Double> vector)
EmpiricalDistribution(java.util.List<java.lang.Double> vector, RandomNumberGenerator rng)

Method Summary

Methods

Modifier and Type	Method and Description
double	next() Create random numbers

Methods inherited from class ikr.simlib.distributions.Distribution

getDefaultRNG, getRandomNumberGenerator, reset

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

EmpiricalDistribution

public EmpiricalDistribution(double[] vector,
                     RandomNumberGenerator rng)

EmpiricalDistribution

public EmpiricalDistribution(double[] vector)

EmpiricalDistribution

public EmpiricalDistribution(java.util.List<java.lang.Double> vector,
                     RandomNumberGenerator rng)

EmpiricalDistribution

public EmpiricalDistribution(java.util.List<java.lang.Double> vector)

Method Detail

next

public double next()

Description copied from class: ContinuousDistribution

Create random numbers

Specified by:

next in class ContinuousDistribution