search

harrispopgen / mushi

[mu]tation [s]pectrum [h]istory [i]nference
https://harrispopgen.github.io/mushi/
MIT License
24 stars 6 forks source link

indexing errors in formulae #5

Closed kamdh closed 5 years ago

kamdh commented 5 years ago

The indices of the y vectors do not exactly match the formulation in the Rosen. y[-1] should be equivalent to eta_k, which is actually eta at time infinity, whereas now we are treating it as the end of our modeled time.

wsdewitt commented 5 years ago

@kharris Can you clarify where there's an indexing error? The modeled time does include the infinite epoch in the current implementation, e.g. we have this assert: https://github.com/WSDeWitt/dement/blob/08c6bcf0aae36daa69fb4dd49a0ef5d71cf07462/dement.py#L23

kamdh commented 5 years ago

Yeah I added that. However, in the paper t[0] = 0 (true for us) but they never use y[0], whereas in our case we do use it.

wsdewitt commented 5 years ago

See commit comment https://github.com/WSDeWitt/dement/commit/cc4fd48151f83cdd706fa3f2fe6d4d291efa501e#r33188986

They do use y[0], but they call it _y1 because python indexing starts at 0.

kamdh commented 5 years ago

What they call y_1 is the time in the first window, from t_0 = 0 to t_1. Anyway, just double check everything. I was thinking there might be a disagreement.