search

petrelharp / ftprime_ms

4 stars 2 forks source link

more explanation in complexity section #73

Closed petrelharp closed 6 years ago

petrelharp commented 6 years ago

This is following on from the discussion in #70. The complexity section may be a bit wordy now, but should be closer to understandable. I'm citing Wakeley, even though it's out of print, because it's the best explanation, I think.

petrelharp commented 6 years ago

oh, dammit, I haven't removed the 2s. Doing this.

petrelharp commented 6 years ago

Thanks for forcing me to actually explain this bit. I was being lazy.

petrelharp commented 6 years ago

I've rewritten this bit substantially, to flow better and include more explanation. In doing this, I could not reconstruct how I got to the previous bound: too bad, because it was a bit closer to the simulations, but I'm forced to conclude that it was in error. We should expect substantial discrepancy between coalescent calculations and the whole-population genealogy, so I'm not worried about the current gap. The form of the dependency (T log N) is right.

petrelharp commented 6 years ago

Possibly in bad form, I've merged to master here, but am leaving this open for discussion.

petrelharp commented 6 years ago

Thanks for the comments. I've made changes for all of them. Except the one about floor.

jeromekelleher commented 6 years ago

Thanks for the comments. I've made changes for all of them. Except the one about floor.

LGTM. I'm sure the one about the floor is irrelevant. Merge away.

petrelharp commented 6 years ago

I've merged this manually.