Closed skhilton closed 5 years ago
Let's use d=10000 for this. In trying to understand the Goldman-Yang behavior, I coded up a calculator for the percent of substitutions (on site-pairs) that should be doublet substitutions, and at d=100 it's 91.5% (for our exchange rates and doublet frequencies) whereas at 10000 it's 99.9%.
We did this, we found a bug, it was useful.
maybe d=100?