Class BNGibbsSampler

java.lang.Object

BNGibbsSampler

public class BNGibbsSampler
extends java.lang.Object

An implementation of the Gibbs Sampling Stochastic Simulation method for estimating Bayesian Network probabilities with/without evidence. You should only modify the simulate method. Algorithm from Section 4.4.3 of Pearl, Judea. "Probabilistic Reasoning in Intelligent Systems"

Field Summary

Fields

Modifier and Type	Field and Description
static int	iterations total iterations; each non-evidence variable is updated in each iteration
static int	reportFrequency iteration frequency of progress reports

Constructor Summary

Constructors

Constructor and Description
BNGibbsSampler()

Method Summary

Modifier and Type	Method and Description
static void	main(java.lang.String[] args) Initialize parameters, parse input, display BN information, and perform Gibbs sampling.
static void	simulate() Perform Stochastic Simulation as described in Section 4.4.3 of Pearl, Judea.

Methods inherited from class java.lang.Object

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

Field Detail

reportFrequency

public static int reportFrequency

iteration frequency of progress reports

iterations

public static int iterations

total iterations; each non-evidence variable is updated in each iteration

Constructor Detail

BNGibbsSampler

public BNGibbsSampler()

Method Detail

main

public static void main(java.lang.String[] args)
                 throws ParseException

Initialize parameters, parse input, display BN information, and perform Gibbs sampling. You should not modify this method

Parameters:

args - an array of command-line arguments

Throws:

ParseException - standard input does not match grammar for Bayesian network specification. (See assignment documentation for BNF grammar.)

simulate

public static void simulate()

Perform Stochastic Simulation as described in Section 4.4.3 of Pearl, Judea. "Probabilistic Reasoning in Intelligent Systems". The enclosed file pearl.out shows the output format given the input: java BNGibbsSampler 1000000 200000 < sample.in > sample.out This is the only method you should modify.