becarioprecario / inlamcmc_examples Goto Github PK

I'm trying to simulate a spatially autocorrelated random data, and I draw both
co-ordinates of 40 points,by using Voronoi tessellation to construct units and get rook type spatial weight matrix w（row-standardisation ).
#generating data N=40 set.seed(1) beta=c(1,3) rho=0.7 alpha=1 x1=rnorm(N,40,2) x2=rnorm(N,36,5) Irhow.t=Diagonal(nrow(w))-rho*w Irhow2.t=crossprod(Irhow.t) sigma=1 sigma2=sigma*solve(Irhow2.t) epsilon=mvrnorm(1,rep(0,N),sigma2) y=solve(Irhow.t)%*%(x1*beta[1]+x2*beta[2]+alpha)+epsilon

I don't konw what's the zero.variance use for ,so I try to change its value,the simulation either crashes (INLA crashes in early iteration xxx)or its result not working well，even I start point that very close to true values.

Thank you for any help in advance.