Raftery and Lewis’s Diagnostic method to infer iterations
Hello,
Recently, I'm tring to use diffsel to detect convergent signals. After ruuning 200 interations, I use Raftery and Lewis’s Diagnostic implemented in the R package coda to estimate the number of necessary iterations, then run as many iterations as 120% of the estimated number.
When I use "md1 <- mcmc(mydata) test = raftery.diag(md1, r=0.01)" to estimate the number of necessary iterations, a total of 1124 iterations are needed. However when i use " md1 <- mcmc(mydata) test = raftery.diag(md1, q=0.025, r=0.005, s=0.95, converge.eps=0.001)", a total of 4495 iterarions are needed. The sites identified by these two iterarions are a little different, do you mind explaing which one to choose?
Best wishes, kangli