Binomial random numbers are generated via the perfect sampling algorithm. At each iteration dual markov chains are generated and coalescence is checked. In case coalescence occurs, the resulting number is outputted. In case not, then the algorithm is restarted from T(t)=2*T(t) until coalescence occurs.