Simply assigning one observation to each component a priori may effectively eliminate the problem. Using a very simple example—a univariate sample from a mixture of two exponentials—we evaluate the ...

Label switching is a well-known problem occurring in MCMC outputs in Bayesian mixture modeling. In this article we propose a formal solution to this problem by considering the space of the artificial ...