autocorrP {MasterBayes} | R Documentation |
Function for assesing mixing of the Markov chain with respect to parentage assignment.
autocorrP(postP, ...)
postP |
JOINT posterior distribution of parentage |
... |
further arguments to be passed |
For each offspring the proportion of transitions is calculated at lags 1, 2, 5, 10, 50 and 100 (i.e. the proportion of times that the parentage assignment at time t is different from the parentage assignement at time t+lag). The difference between these proportions and the proportion at lag 1 is then calculated, and the mean over offspring given. Samples are independent when this difference is non-decreasing with increasing lag. When the parentage assignemnets in each MCMC iteration are independent these autocorrelation metrics should be randomly distributed about zero.
matrix
Jarrod Hadfield j.hadfield@ed.ac.uk
data(WarblerP) data(WarblerG) GdP<-GdataPed(WarblerG) var1<-expression(varPed(c("lat", "long"), gender="Male", relational="OFFSPRING")) # paternity is to be modelled as a function of distance # between offspring and male territories res1<-expression(varPed("offspring", restrict=0)) # indivdiuals from the offspring generation are excluded as parents res2<-expression(varPed("terr", gender="Female", relational="OFFSPRING", restrict="==")) # mothers not from the offspring territory are excluded PdP<-PdataPed(formula=list(var1,res1,res2), data=WarblerP, USsire=FALSE) tP<-tunePed(beta=30) model1<-MCMCped(PdP=PdP, GdP=GdP, tP=tP, nitt=3000, thin=1, burnin=0, write_postP="JOINT") autocorrP(model1$P)