24 days later, so much for end of the week. Finally finished rolling out the Tan improvement. Phew! [Not all pushed yet, but will be in near future. Improvements are everywhere, especially nice with fastOLG mode in OLG transition paths.]
A quick comment on using Tan improvement in portfolio choice models (or any model where an iid shock occurs between this period decisions and next period endogenous state; what VFI Toolkit calls a u shock). Notice that when you have a u shock you will sometimes get the same grid index for lower grid point from one value of u and upper grid point of another value of u [upper and lower refer to interpolating back onto the grid]. Turns out Matlab’s sparse() command automatically sums the two values when an index is repeated, which is anyway exactly what you want to do to create Gamma for step 1 of the Tan improvement. Got lucky 