Only compute the hybrid model based on second and fourth order
perturbation. Higher orders would necessitate smaller tolerance
parameters when solving the stochastic perfect foresight (auxiliary)
models (we use the default value of 1e-5).

Evaluate accuracy with hybrid SEP(2) for different approximation
orders for the perturbation.

Compute share of the gap between SEP and exact solution closed by increaseing the approximation order.

- Introduce fourth-order perturbations
- Remove SEP(10) with the complete tree
- Cosmetic adjustments

Also display results as soon as possible.

Use latexindent Perl script.

Also:

 - Fix font size in Tikz plots
 - Add colors (provisions for IRFs with stochastic perfect foresight models)
 - Change notations for the paths of endogenous variables

With and without irreversible investment.

With SEP order=0:10.