Kevin R. Thornton

@hadley @asokolsky Copy _layouts/default.html from this project into yours. Edit sidebar manually. Example from one of my projects: https://github.com/molpopgen/fwdpp/blob/dev/_layouts/default.html#L20

Does it make sense to deprecate arguments named "samples" and replace them with "sample_nodes" and "sample_individuals"?

> It does, but I hate to churn everyone's code like that... We can raise FutureWarning when samples is used for a release or 2, then remove it. I think...

There's a slight caveat here: the denominator only holds if all samples have the same birth time (even when all of the other assumptions are met).

Upon some further reflection: this caveat is no worse than applying the Tajma's D stat to a sample set of heterogeneous ages. So feel free to ignore me!

It is not always the case, though, that the unit of a tree sequence is a "base pair". While that is probably true for inferred tree sequences, it often isn't...

> In principle, simulated ones that aren't in base pairs are likely to be from zero to one, I suspect? Not necessarily. I've done things that are 10 "units" long...

If removing all nodes ancestral to a certain time is decapitating, then this would be to `amputate` or `dismember` the tree: ```py import tskit.gory_ops ```

This would be cool to have implemented. There are some subtleties that arise for handling mutations that don't occur in the case of `decapitation`. For any parent/child edge that you...

Exactly. For each parent/child combo spanning the census time, you can pass the parent time, child time, census time, and the index of the mutation on the child node to...