My native Patterson function calculations will take two years of CPU time to complete.

So are mine, I'm afraid. Actually, you may well find that the program will stop much earlier than the two years, with an error message ``Failed to reach convergence ...''. I do not think that there is a native-Patterson-specific bug in GraphEnt. Rather, it is probably that the data for a native Patterson are usually complete and of high quality. The higher the data quality, the longer GraphEnt will take to fit the constraints imposed by them (see also quotation on page ).