Dear Sir/Madam,
i am learning to use a statistical sampling-modle(arms in HI)
But i have a problem with the "indFunc"(bounds of the convex support).
I have found a good example of c code using "arms" function and try to
program the example in R.
If you are interested, you can find the c code from arms usage(P.155-162)
But i always get the error message:
error at convex.bounds(y.start,dir=1,indFunc,...): x not in the support of
indFunc
the R code :
####simulate intercept mu
#####starting values of mu: mu<-c(0,-2,-4,.....)
for (k in 1:K){
for (t in 1:25){
if(k==1){
n_MU[t]<-firmsCCC[t] -----used in the density function(Mudens)
N_MU[t]<-defCCC[t] -----used in the density function(Mudens)
MUl=mu[2]
MUr=50
}
else if(k==2){
n_MU[t]<-firmsB[t]
N_MU[t]<-defB[t]
MUl=mu[3]
MUr=mu[1]
}
else if...
...
}
Muprev<-mu[k]
mu[k]<- arms(Muprev,Mudens,function(xm,...)(xm>MUl)*(xm<MUr),1)
}
#########In the examples of the package the bounds of the convex support are
fest numbers,
######## but here the bounds MUl and MUr are variable and they are updated
after every iteration, what i have
never found in other cases.
best regards!!!!
Yuan Hu
[[alternative HTML version deleted]]