### This script provdes several initial values.
rm(list = ls())
suppressMessages(library(pbdMPI, quietly = TRUE))
init(set.seed = FALSE)
suppressMessages(library(cubfits, quietly = TRUE))
### Set environment.
source("00-set_env.r")
set.seed(simulation$seed)
fn.in <- paste(prefix$data, "pre_process.rda", sep = "")
load(fn.in)
### Initial.
init.function(model = model, parallel = run.info$parallel)
### Run PM initial.
fitlist <- fitMultinom(reu13.df.obs, phi.Obs, y, n)
phi.init.PM <- estimatePhi(fitlist, reu13.df.obs.list, y.list, n.list,
E.Phi = median(phi.Obs),
lower.optim = min(phi.Obs) * 0.01,
upper.optim = max(phi.Obs) * 2.0)
phi.init.PM <- as.double(phi.init.PM)
names(fitlist) <- names(reu13.df.obs)
names(phi.init.PM) <- names(phi.Obs)
### Check NA of phi.init.PM.
id.na <- which(is.na(phi.init.PM))
phi.init.PM[id.na] <- phi.Obs[id.na]
### Run scuo initial.
phi.init.SCUO <- scuo.random(SCUO, meanlog = -simulation$sdlog^2 / 2,
sdlog = simulation$sdlog)
names(phi.init.SCUO) <- names(phi.Obs)
### Run more scuo initial.
if(.CF.CONF$scale.phi.Obs){
phi.init.SCUO.emp <- log(SCUO / mean(SCUO))
} else{
phi.init.SCUO.emp <- log(SCUO)
}
phi.init.SCUO.emp <- exp(phi.init.SCUO / sd(phi.init.SCUO) * simulation$sdlog)
names(phi.init.SCUO.emp) <- names(phi.Obs)
### Scale to Mean 1, since here phi is for initial values.
if(.CF.CONF$scale.phi.Obs || .CF.CONF$estimate.bias.Phi){
### No scaling for initial values, but may estimate.bias.Phi
phi.init.PM <- phi.init.PM / mean(phi.init.PM)
phi.init.SCUO <- phi.init.SCUO / mean(phi.init.SCUO)
phi.init.SCUO.emp <- phi.init.SCUO.emp / mean(phi.init.SCUO.emp)
}
### Save.
if(comm.rank() == 0){
comm.size <- comm.size()
ret.time <- proc.time()
print(ret.time)
fn.out <- paste(prefix$data, "init_", model, ".rda", sep = "")
list.save <- c("fitlist", "phi.init.PM", "phi.init.SCUO",
"phi.init.SCUO.emp",
"ret.time", "comm.size")
save(list = list.save, file = fn.out)
warnings()
}
finalize()
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.