I implemented a parallel rugarch script by looping over 4 input parameters
and writing some output into two files for optimal AIC and BIC,
respectively. My PC has 4 cores.
I write output when the new AIC or BIC are more negative than the current
ones (see the snippet below). But now my question is: can the
parallel sessions be confused about which AIC/BIC are current? And could
ihtis implementation miss some parameter combos?
Thank you, Alec
result <- foreach(ka = ka0:kaa, .packages = 'rugarch') %:%
foreach(kb = kb0:kbb) %:%
foreach(km = km0:kmm) %:%
foreach(kn = kn0:knn) %dopar% {
# print(c(ka, kb, km, kn))
GARCHspec <- ugarchspec( variance.model = list(model = garchType,
garchOrder = c(km, kn)),
mean.model = list(armaOrder = c(ka, kb),
include.mean = TRUE),
distribution.model = dist)
#fit
aic <- infocriteria(fit)[1] #AIC
bic <- infocriteria(fit)[2] #BIC
if(aic < currAIC) {
currAIC <- aic
ka_aic <- ka
kb_aic <-kb
km_aic <- km
kn_aic <- kn
aic_sigma <- sigma(fit)[L]
msg <- sprintf("%d %d %d %d %.3E %.3E\n",
ka, kb, km, kn, currAIC, aic_sigma)
cat(msg, file = aic_garch_outfile, append = TRUE, sep=',')
}
if(bic < currBIC) {
currBIC <- bic
ka_bic <- ka
kb_bic <-kb
km_bic <- km
kn_bic <- kn
bic_sigma <- sigma(fit)[L]
msg <- sprintf("%d %d %d %d %.3E %.3E\n",
ka, kb, km, kn, currBIC, bic_sigma)
cat(msg, file = bic_garch_outfile, append = TRUE, sep=',')
}
[[alternative HTML version deleted]]
_______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should
go.
