I’ve generated three training data sets, which will save you around 5 days if you decide to run Jaatha, a molecular demography program. It uses the joint site frequency spectrum of two populations to model various aspects of population history (split time, population size and growth, migration). Here’s the paper: Naduvilezhath et al 2011.
1. Using the default model, with the following maxima: tmax=20, mmax=5, qmax=10.
2. Alternative maxima: tmax=5, mmax=20, qmax=20.
3. Alternative maxima: tmax=5, mmax=20, qmax=5.
They can’t be uploaded because they’re compressed R data structures, but let me know if you’d like to give them a whirl.