How do we determine when to stop the simulation? These is no simple answer to that question. One complication is multimodality of the simulated distribution. The chain could stay near a local mode for a very long time and the simulation could be stopped before the chain visits the major mode. Simulated annealing and related methods improve the mixing of the chain and help avoiding these problems. To decide how many steps we need, it helps to run the sampler on simulated data where we know the answer we are looking for (e.g. the recombination fraction of a gene with a marker). Then, we can see for how long we need to simulate to get the right answer back. There is however no general rule to decide when to stop.