BetaProposal Class Reference
[Metropolis-Hastings Proposals]

Beta proposal for Metropolis-Hastings. More...

#include <mcmc.h>

List of all members.

Public Member Functions

 BetaProposal ()
 Empy constructor.
 BetaProposal (double idenominator)
 Main constructor.
double Draw (double starting_value)
 Draw from Beta proposal.
double LogDensity (double new_value, double starting_value)
 Log probability of proposal new value, given starting value.


Detailed Description

Beta proposal for Metropolis-Hastings.

Useful proposal type for parameters with support between 0 and 1.

The beta proposal is parameterized using the mean and the inverse denominator. The denominator is a measure a scale and can be thought of as the number of observed trials bernoulli. We use the inverse of the denominator so that larger values imply bigger steps, consistent with NormalProposal. For the Beta distribution the mean is

\[ m = \frac{\alpha}{\alpha + \beta} \]

and the denominator is

\[ d = \alpha + \beta. \]

Therefore

\[ \alpha = m \cdot d \]

and

\[ \beta = d \cdot (1-m). \]


Constructor & Destructor Documentation

BetaProposal::BetaProposal (  )  [inline]

Empy constructor.

BetaProposal::BetaProposal ( double  idenominator  )  [inline]

Main constructor.

Scale of steps are parameterized using the inverse denominator of a Beta distribution:

\[ \frac{1}{d} = \frac{1}{\alpha + \beta} \]

Parameters:
idenominator Tuning parameter equal to the inverse denominator $1/d$ of a Beta distribution.


Member Function Documentation

double BetaProposal::Draw ( double  starting_value  )  [inline]

Draw from Beta proposal.

Parameters:
starting_value Starting value (mean of Beta distribution).

double BetaProposal::LogDensity ( double  new_value,
double  starting_value 
) [inline]

Log probability of proposal new value, given starting value.

Parameters:
starting_value Starting value.
new_value Proposed new value.


The documentation for this class was generated from the following file:

Generated on Thu Jun 18 03:11:16 2009 for Scythe MCMC by  doxygen 1.5.9