A complete understanding and evaluating of this colorspace would take probably another year, because it depends on a set of parameters and equations that I've "invented" without studying their consequences and behavior.

It would be interesting in future work to study a bit what happens to the colorspace when you change the equations and paramethers while keeping the basic conditions.

First try (and last, for now): change the blackening function from

Rb = B_2 * R^2 + B_1 * R + B_0

to

Rb = B_2 / R + B_1 * R + B_0

the mixing capabilities are pretty much the same but I can see a wider spectrum of greens