*******************************************************
README - Changes in LearnBayes 2.0 (from LearnBayes 1.0)
December 15, 2008
******************************************************

1. Changes to function laplace

This function now uses the more robust Nelder-Mead algorithm 
implemented in the optim function in the base package.
The function has only three inputs, logpost, starting value, 
and data used in the logpost function.   There is one additional
output -- converge (TRUE or FALSE) that indicates if the Nelder-Mead algorithm
converged.

2. Changes to the coding of the log posterior functions

The definitions of the sample log posteriors have been simplified to
accept vector (instead of matrix) inputs for the parameter.  With this 
change, it is not necessary to use loops in the function definition.
The following functions have been changed:  betabinexch0, betabinexch,
bfexch, cauchyerrorpost, groupeddatapost, howardprior, lbinorm, logctablepost,
logisticpost, normchi2post, poissgamexch, transplantpost, weibullregpost.

3. Changes to the functions that operate on posteriors

Generally these functions have been made more user friendly by allowing
the starting value to be a vector instead of a row matrix.  Changes have
been made to laplace, indepmetrop, and rwmetrop

4.  Change to function groupeddatapost

There was a small error in this function in LearnBayes 1.0 that has been corrected.  
Also the data input to this function is now a list with three components, the
vector of left bin boundaries int.lo, the vector of right bin boundaries
int.hi, and the vector of bin boundaries f.

5.  New functions 

New functions have been added and there have been changes to current functions -- 
most of the new functions and current functions are described in the second edition 
of Bayesian Computation Using R.

----------------------------------------------
Chapter 2 -- Introduction to Bayesian Thinking
-----------------------------------------------

beta.select -- for binomial sampling, finds the shape parameters of a beta density 
that matches knowledge of two quantiles of the distribution. 

predplot -- for binomial sampling and a beta prior, this will graph the
prior predictive distribution and show the observed data value

triplot -- this will produce the popular "likelihood, prior, posterior" plot
for binomial sampling with a beta prior

------------------------------------
Chapter 3 -- Single Parameter Models
------------------------------------

binomial.beta.mix -- this performs the posterior calculations for binomial
sampling and a prior that is a mixture of beta densities

poisson.gamma.mix -- this performs the posterior calculations for Poisson
sampling and a prior that is a mixture of gamma densities

normal.normal.mix -- this performs the posterior calculatiosn for normal
sampling (known variance) with a prior that is a mixture of normal densities

normal.select -- Finds the mean and standard deviation of a normal density 
that matches knowledge of two quantiles of the distribution. 

rigamma --  Simulates from a inverse gamma (a, b) distribution with density 
proportional to $y^{-a-1} exp(-b/y)$ 

----------------------------------
Chapter 4 -- Multiparameter Models
----------------------------------

normpostsim -- this produces a sample from the posterior of (mean, variance)
for normal sampling with a noninformative prior

normpostpred -- this simulates the posterior predictive distribution of a 
statistic for a normal sampling problem

bayes.probit -- this produces a simulated sample from the posterior for a 
probit regression model with a normal prior on the regression coefficient

------------------------------
Chapter 9 -- Regression Models
------------------------------

bayes.model.selection -- given a regression model with multiple covariates,
this will compute the marginal density for each possible model using Zellner's
G priors

blinreg -- Computes the log posterior of (beta, log sigma) for a normal regression model. 
This function nows allows for the input of Zellner's g prior with parameters beta0 and c0

bradley.terry.post -- Computes the log posterior density of the talent parameters 
and the log standard deviation for a Bradley Terry model with normal random effects 

reg.gprior.post -- Computes the log posterior of (beta, log sigma) for a normal 
regression model with a g prior with parameters beta0 and c0.
 
----------------------------
Chapter 10 -- Gibbs Sampling
----------------------------

bayes.probit -- this function now allows the input of a multivariate normal
prior and will compute the log marginal likelihood when an informative prior
is used

rtruncated -- this produces a sample of size n from a known distribution 
(normal, beta, gamma, etc) that is truncated between two values

******************************************************************************