search

petrelharp / treestats_ms

1 stars 1 forks source link

Benchmarks prose #14

Closed molpopgen closed 5 years ago

molpopgen commented 5 years ago

Rough draft of text re: benchmark fig.

petrelharp commented 5 years ago

I've updated the O( ) stuff. It wasn't right because we had a model for total time to compute a branch stat, while what we were fitting was per-site time to fit a site stat. I adjusted this, and also had it do the fitting on a log-log scale (since previously most of the fit was being determined by the biggest value). Note that the model is the same because in the sims mu = r. The fit is somewhat better, I think.

petrelharp commented 5 years ago

feel free to merge when you want, btw