For discussion: @llorracc-git , @sbenthall . We discussed over skype the need to have nonregular grids. I started to implement two approaches:
using multilinear interpolation on cartesian products of strictly increasing sets of points
using warp functions mapping a regular grid to a nonregular grid.
I documented a bit what I did here: https://github.com/EconForge/dolark/blob/albop/new_grids/experiments/new_grids_discussion.ipynb
From these small experiments, I'm half convinced that warp functions are useless even though they are very simple to implement.
Note that most of the code is in dolo_improvements.py files. It will ultimately replace the decision rule code in dolo but for now, it is not used anywhere.
For discussion: @llorracc-git , @sbenthall . We discussed over skype the need to have nonregular grids. I started to implement two approaches:
dolo_improvements.pyfiles. It will ultimately replace the decision rule code in dolo but for now, it is not used anywhere.