ROADMAP

[MichelJuillard]

Solution methods

• finalize sparse grids
• explore using PATH with sparse grids for models with occasionally binding constraints
• add deep net equilibrium
• add GDSGE (leverage Cao et al. C++ library)
• finalize k-order local approximation
• add stochastic extended path
• extend solution methods for limited rationality models

Interface with *.mod file

• benchmark evaluation of symbolic derivatives produced by preprocessor with Julia AD solutions

Estimation

• compute analytical derivatives of posterior density + compare with automatic differentiation
• finalize ./src/estimation/smc.jl
• investigate and add Hamiltonian Monte Carlo
• add Approximate Bayesian Computation (ABC)

Simulation and Forecast

• add plots with confidence regions

Recursive blocks

• finalize decomposition of a model in recursive blocks
• identify intermediate variables that can be treated as local variables in a function representing the sub-model
• adapt existing solution algorithms to solve the various blocks in a loop

Heterogeneous agent models

• use the techniques developed for recursive blocks to solve heterogeneous agent models

Markov switching models

• explore collaboration with Junior Maih's RISE

Global sensitivity methods

• port Marco Ratto's code from Dynare Matlab