Right now pi is only generated at the tips using abundance and therefore assuming no incomplete lineage sorting. Since we have the tree we could construct a giant coalescent model and simulate data at the tips in more integrated fashion, but this involves some potentially nasty housekeeping to build the msprime demographic model.
This is now done in PIED.nucleotide_diversity_ILS(), but it returns much different info than PIED.nucleotide_diversity(), so it hasn't been worked into the full model yet
Right now pi is only generated at the tips using abundance and therefore assuming no incomplete lineage sorting. Since we have the tree we could construct a giant coalescent model and simulate data at the tips in more integrated fashion, but this involves some potentially nasty housekeeping to build the msprime demographic model.