Chapter 2 of thesis:
- Convergence tolerance
In the main file of Larch, several related variables could be traced to the convergence tolerance. In the document of Larch, the author mentioned two variables that might be related to convergence tolerance level, one is ctol, and the other is ftol. I set both to 1e-4.
Also, the model did not converge in the following situation: take log of the average of the disposable income of a place, did not add a quantity of average disposable income of a place, add constant terms to the model.
In one of the model I choose, not sure if it is because of too few variables, or by large ctol level (which I set 1e-3), singularity came out.