High r/m error, branch assertion failed

[Uthara28]

I am trying to do some benchmarking of SINGER using msprime-generated VCF files for different r/m values and the beta coalescent model. The goal is to eventually run it on a species that shows an r/m ratio of 1000 and potentially non-Kingman genealogies.

Simulation Details

• Simulate non-Kingman (beta coalescent) genealogies with constant population size.

• When I run the following command:

  python3 singer-0.1.8-beta-linux-x86_64/singer_master \
      -vcf "${SUBDIR}/${FILENAME_PREFIX}" \
      -output "${OUTPUT_SUBDIR}/${FILENAME_PREFIX}_singerout" \
      -m 1e-9 -start 0 -end 1e7 -Ne $Ne \
      -ratio 100 -polar 0.99 -n 25 -thin 100

• The effective population size (Ne) is dynamically calculated using pi = ts.diversity() from a .ts file generated by the same simulation that produced the VCF. The relationship used is $( \pi = 2Ne \cdot m )$ (haploid sequences).

• A sample Ne for these configurations is 292.

I seem to encounter two types of errors:

1. Error Type 1:

   singer: Tree.cpp:70: void Tree::forward_update(Recombination&): Assertion `prev_size == after_size or r.pos == 0' failed.
   

2. Error Type 2:

   singer: Branch.cpp:16: Branch::Branch(Node_ptr, Node_ptr): Assertion `(l == nullptr and u == nullptr) or l->time < u->time' failed.
   

   • The second error is more common.