Power model add #13
Power model add #13
Conversation
There was a problem hiding this comment.
Nice work Tess. Looks great. However, I suggest we write power law as dx/dt = a (x/x_c)^b. Your implementation has x_c=1. a general guide would be that x_c ~= mean(x), but rather than estimating from the data, it would be better to set this in the model setup as a specified quantify. Without this, the slope and intercept become heavily correlated and hard to converge.
… references in helper functions for testing and data generation.
|
Changes to power law model done. @dfalster I don't know if the data generation process is sane as I produce y_bar values and pass them along with the other DE parameters to used to determine true sizes over time. Those true values also get passed to the model, but I assume in practice these would be estimated from the data or something. |
…_bar, including the data generation process. Currently fails to properly estimate coefficient parameter for single individual.
…er functions to use map rather than for loop.
Added the power law model.