tests/seeded-results.R

In cbreto/panelPomp: Inference for Panel Partially Observed Markov Processes

# check reproducibility over a range of randomly chosen seeds
if (file.exists("_options.R")) source("_options.R")
library(panelPomp,quietly=TRUE)

ppo <- panelRandomWalk(U=3,N=4)

seeds <- c(21125715L,81902071L,86680005L,29971689L,73376915L)

for (s in seq_along(seeds)) {
  set.seed(seeds[s])
  pf <- pfilter(ppo,Np=10)
  print(unitlogLik(pf))
  print(logLik(pf))
}

for (s in seq_along(seeds)) {
  set.seed(seeds[s])
  mf <- mif2(ppo,Np=10,rw.sd=rw_sd(sigmaX=0.05,X.0=0.5),
             cooling.fraction.50=0.5,cooling.type="geometric")
  print(mf@shared)
  print(mf@specific)
  print(unitlogLik(mf))
  print(logLik(mf))
  print(traces(mf))
}