Alternate models
Writing down the change-point model correctly now, starting with the simulation. Done. Implemented in sde_likelihood.R
Implementing Neyman-Pearson comparisons for all three models.
Linearizing directly from the saddle node model
Dynamics:
Slope (alpha) at equilibrium comes from the derivative:
is the slope evaluated at n equal to equilbrium (which is the root of some messy cubic).