search

harrispopgen / mushi

[mu]tation [s]pectrum [h]istory [i]nference
https://harrispopgen.github.io/mushi/
MIT License
24 stars 5 forks source link

Optimization #31

Closed kamdh closed 4 years ago

kamdh commented 4 years ago

Mostly a change to the nomenclature in the document & code. Now we are using "cost" to describe the optimization objective function, "loss" for misfit term, and "regularization" for the rest.

Also should fix a small bug in the accelerated prox-gradient algorithm with regards to how the next iterate is formed.

wsdewitt commented 4 years ago

Ok this all looks good to me, and the Nesterov code indeed seems to correspond to the Accelerated proximal gradient method section beginning p.3-26 in these notes