Andrew Kern

> I'm finding the time and population size conversions in the sweeps model super confusing, and I would _really_ like to decouple the trajectory simulation from the sweep definition. I...

huh interesting... i wonder what i messed up here. i'll dig a bit.

so browsed this quickly-- i think the difference is that discoal is doing a full soft sweep (i.e. conditioning on a neutral trajectory from freq f_0 -> 1/2N) where as...

yeah i think that's it. to get msprime to do something equivalent to the `discoal -f` option we need to add a neutral trajectory model or maybe just a soft...

Kevin are you going to implement the neutral trajectory stuff?

> So we need to be a bit careful here, and will want @jeromekelleher to weigh in on this. We have a 1.0 API that has a `start_frequency` in there,...

Ah okay. I have a branch going with some of the neutral traj code. I'll clean that up and start getting a PR together.

> do you think it could work as a PhD chapter? Or is there more/less work than that. I'd say not great PhD chapter material @jjberg2-- a bunch of low...

kevin note this is just the neutral fixation case, an incremental step towards getting the soft sweep model in

also i should probably add verification against SLiM for patterns of linked polymorphism to this PR. Right now in `verification.py` i've only got tests of the sojourn time of the...