Initiate the algorithm with an arbitrarily selected. Correspondingly, set. Letbe the vector of dimensionin which the firstelements store the distinct value ofandbe an-vector to represent the numbers ofoccurring in.
(3)
Now assume that we have gotand the corresponding auxiliary vectorsand. The sampled valuesare generated by the following two-step procedure:
Step 1. For ,
(1) set ,;
(2) draw a new value for from according to the distribution law
and then, in case , draw a value for from with replaced by , where is the normalization constant subject to ; and then
(3) set .
Step 2. For all which are obtained by above Step 1, draw from the conditional density
