Mhsim = function(bstart,v,w,x,y,nmc) {
# function to perform MH (Metropolis-Hastings) simulations for a 
#   log-linear Poisson model with 1 covariate and a Gaussian prior
#   on the (intercept,slope) parameter vector.
# INPUTS:
#   bstart: 2-vector of starting values for the parameters
#   v = 2-vector of prior variances
#   w = 2-vector of variances for metropolis-hastings steps
#   x = n-vector of predictors
#   y = n-vector of the (Poisson) responses
#   nmc = positive integer; the number of steps in the MH algorithm
# OUTPUTS:
#   bmat = (nmc+1)x2 matrix of simulated parameter values; bmat[i+1,]
#      is the simulated value of the parameter vector on the i'th step
#      of the MH algorithm
#   lups = (nmc+1) vector of the log of the unnormalized posterior
#########################################################################3
bmat = matrix(NA,nrow=nmc+1,ncol=2)
colnames(bmat) = c("intercept","slope")
lups = rep(NA,nmc+1)
t = c(sum(y),sum(x*y))
bmat[1,] = bstart
lups[1] = Lup(bstart,v,t,x)
for(i in 1:nmc) {
   temp = Mhstep(bmat[i,],lups[i],v,w,t,x)
   bmat[i+1,] = temp$bnew
   lups[i+1] = temp$lupnew
}
mhout = list(bmat=bmat,lups=lups)
return(mhout)
}