High-ish level comments

[ashander]

I like the organization.

(I need to add some comments to Results and Discussion still)

[ashander]

I will be adding a touch more shortly, @petrelharp but I think this gets at my major structural comments/suggestions

[ashander]

ok I've added everything.

[petrelharp]

I maybe heavily edited these edits. Didn't address the question of Alg F; will think again later. Any other opinions?

[ashander]

@molpopgen @jeromekelleher any thoughts on:current state of this? especially

• changing the way alg W is introduced
• feelings ( +/- 1s) on putting in something like alg F, or other ways to make 'our method' more explicit

[jeromekelleher]

Regarding Alg F, I'm not sure I see the need for it. Adding a simplification step to Alg W really is very simple, and we could just explain this in the text (see this implementation for example). Basically, all we have to do when we want to simplify is (a) sort the tables; (b) call simplify with samples=P; and (c) replace the current population P with [0, 1, ..., N-1].

I've avoiding adding table objects to Alg W as I thought it would be simpler to use lists/sets. But we could easily change Alg W to record directly into node and edge tables, and then describing this extra step of simplification is trivial. We could even add this step in to Alg W I guess.

[jeromekelleher]

I've made a quick stab at updating Alg W over in #24 to use the table notation. Under this formulation we can easily include calls to sort_tables() and simplify. We shouldn't need the proposed Alg F at all then, as we have a fully detailed implementation of the method in one place.

Thoughts?

[petrelharp]

I like this solution. If you do, too, @ashander, go ahead and merge jerome's branch into here?

[ashander]

This is great @jeromekelleher and I think will make things clearer. I pulled in your changes to alg w and left off all the unrelated whitespace changes from #24

[petrelharp]

lmk when you're good for me to merge.

[ashander]

ok. good to merge whenever