MCMC to estimate a jump-diffusion model

We use Winbugs to do MCMC through R, the code is below:

# winbugs program#
modle {
     for (i in 1:N) {
              y ~dnorm(r,tau)    # tau corresponding to 
       r<-mu+Xi*J            # Xi corresponding to  #
       Xi[t]~dnorm(Jmu,Jtau)
        J[t]~dbern(lamda)
}
         Sigma<-1/sqrt(tau)
         JSigma<-1/sqrt(Jtau)
           mu~dnorm(0,1)       # prior distributions for parameters
           tau~dgamma(2, 2)
           Jmu~dnorm(0,1)
           Jtau~dgamma(2,2)
          lamda~dbeta(2,2)

}

# R code with interface to Winbugs

library(R2WinBUGS)  
y=read.table("c:/data.txt")    #import data#
N=length(N)
Xi0=rep(0,N)   # Give initial values to state variables#
J0=rep(0,N)
data=list("N","y") 
# Give initial values to the parameters for winbugs #
inits = function() {list ( tau=2,
                   Jtau=2,
                   mu=0,
                   Jmu=0,
                   lamda = 0.05, Xi=Xi0,J=J0)}
parameters<- c("Sigma","JSigma","mu","Jmu","lamda","Xi","J")
# quote the winbugs#
jump.sim<-bugs(data,inits,parameters,"c:/mcmc/model.txt",n.chains=1,n.iter=1000)

无论怎么调整初始值和先验参数，跳跃的强度 lamda 始终在0.2以上，请问为什么 是不是我模型有问题？data 已经上传：）期待指教：）