Dear All: I have an AR(1) example via WinBUGS. Maybe you are interested in. Many thanks to Professor Farrow. Best Wishes, Gary Gau Math & Stat Boston University http://math.bu.edu/people/wgau ----- Original Message ----- >a simulated AR(1) series > > y_t = mu + phi*(y_t-1 - mu) + e_t > > with mu=10, phi=0.8 and var(e)=1. > > The other file contains a (classic) BUGS program which works with this model and > these data. The initial observation y_0 is taken as fixed, y_0=13. I took a > sample of 10000 phi values and, curiously, there was a little hump at the upper > end of the posterior distribution. Otherwise everything seems perfectly OK. > ---------------------------------------------------------------------------- ---- > 13.0826 > 12.0263 > 13.2564 > 12.1181 > 11.3720 > 12.4619 > 11.2712 > 10.3401 > 10.2225 > 9.4269 > 10.8509 > 9.2491 > 8.1869 > 7.9515 > 7.9624 > 9.5700 > 11.4676 > 9.9282 > 10.0481 > 9.0173 > 9.2939 > 8.4095 > 8.0112 > 8.0132 > 9.1525 > 10.0040 > 8.9083 > 9.1437 > 9.6536 > 10.5645 > 11.2686 > 9.5324 > 10.0392 > 9.3548 > 8.5384 > 9.5017 > 10.3454 > 9.1905 > 8.9498 > 9.1033 > 8.2317 > 6.8483 > 8.05571 > 7.50762 > 5.86687 > 6.94436 > 9.50971 > 9.27238 > 9.46818 > 9.20724 > 7.54402 > 7.41957 > 7.38182 > 8.41263 > 8.31616 > 9.12996 > 9.86440 > 8.57547 > 9.81590 > 9.00445 > 9.43477 > 9.13524 > 9.06310 > 8.1533 > 8.2119 > 8.4364 > 8.4518 > 8.6679 > 9.0107 > 9.3537 > 9.8223 > 10.6041 > 8.8178 > 11.1377 > 11.3183 > 11.9031 > 11.1018 > 9.4257 > 8.8812 > 8.9565 > 8.0170 > 8.5344 > 9.1885 > 8.9057 > 10.1016 > 10.5212 > 10.6441 > 10.1745 > 8.4729 > 9.5725 > 8.7555 > 9.6180 > 8.4995 > 8.5272 > 8.1157 > 9.3125 > 10.0596 > 11.3611 > 10.3386 > 10.2119 ---------------------------------------------------------------------------- ---- > model ar1a; > > const N=100, y0=13; > > var y[N], ymean[N], mu, phi, tau; > > data y in "ar1.dat"; > > { > ymean[1]<-mu + phi*(y0-mu); > y[1]~dnorm(ymean[1],tau); > > for (i in 2:100) { > ymean[i]<-mu + phi*(y[i-1]-mu); > y[i]~dnorm(ymean[i],tau); > } > > mu~dnorm(0,0.1); > tau~dgamma(1,0.1); > phi~dunif(-1,1); > > }