The variables and are given by the distributions and with common support. Assume that We generate the variable as follows:

Step 1. Generate and independently.

Step 2. If then set otherwise return to step 1.

Indeed, It is a part of the above computation that Hence, for the procedure to be effective, has to be large.