Multivariate Uniform Prior over triangle


Is it possible to implement a multivariate uniform prior in which the parameters, s1 and s2, are uniform in the triangle 0 < s1 < s2 < T?


1 Like

You just declare it that way in the parameters block

parameters {
  real<lower = 0, upper = T> s1;
  real<lower = s1, upper = T> s2;

Then you have the triangle and can use any prior you want over that, including no further modification of the log-kernel, in which case you would be expressing a prior belief of uniformity.