Description Usage Arguments Value Author(s) Examples
View source: R/multivspost2blog.R
Function to evaluate marginalized and reparameterized joint posterior.
1 | multivspost2blog(smat, alpha, beta, D, y, By, k)
|
smat |
|
alpha |
|
beta |
|
D |
|
y |
|
By |
|
k |
vector of the log of the posterior marginalized density of each set of candidate kappa values
Kate Cowles
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 | ##---- Should be DIRECTLY executable !! ----
##-- ==> Define data, use random,
##-- or do help(data=index) for the standard data sets.
## The function is currently defined as
function(smat, alpha, beta, D, y, By, k )
{
# computes posterior marginal of S in He/Hodges model
# called by optimization function
# alpha = c(alphae, alphaz_1 to alphaz_{F-1})
# beta = c(betae, betaz_1 to betaz_{F-1})
# D: matrix of diagonals of the diagonal matrices from diag of Q matrices
# k is rank deficiency of sum(Q's); i.e., rank of Q is n-k
# smat is matrix with F-1 cols; each row is s_1 to s_{F-1}
smat <- as.matrix(smat)
n <- length(y)
F <- ncol(D) + 1
sums <- apply(D,1,sum)
#k <- length( sums[sums==0] )
logpostvect <- numeric()
tausqtot <- numeric()
sumalpha <- sum(alpha)
s <- smat
s0 <- 1-sum(s)
neweigennumer <- s[1] * D[,1]
if (F > 2)
for(j in 2:(F-1))
neweigennumer <- neweigennumer + s[j] * D[,j]
neweigendenom <- neweigennumer + s0
neweigen <- s0 * neweigennumer / neweigendenom
# corrected to (alpha-1) 09/18/09
#logpostdensnumer <- sum( log(c(s0,s)) * alpha ) +
logpostdensnumer <- sum( log(c(s0,s)) * (alpha-1) ) +
sum( log(neweigen[ neweigen > 0 ]) ) / 2
# whole <- sum( neweigen * Bysq )
whole <- sum( neweigen * By^2 )
newbeta <- whole/2 + sum( c(s0,s) * beta )
newalpha <- (sumalpha + (n-k)/2 )
logpostdensdenom <- log( newbeta)* newalpha
logpostvect <- logpostdensnumer - logpostdensdenom
logpostvect
}
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.