We want the stationary distribution
π(x)
to be the posterior distribution
P(x)
. So we set
P(x)P(x→y)=P(y)P(y→x)
Rearranging, we get
P(x)P(y)=P(y→x)P(x→y)
We split the transition probability into separate proposal q and acceptance
A(x→y)A(y→x)=P(y)q(y→x)P(x)q(x→y)
An acceptance probability that meets this conidtion is
A(x→y)=min[P(y)q(y→x)P(x)q(x→y),1]
since A <script type="math/tex" id="MathJax-Element-890">A</script> in the numerator and denominator are both bounded above by 1.