-
Notifications
You must be signed in to change notification settings - Fork 23
Open
Description
Hi Spencer and thanks for the awesome package!
When I took macro w/ Jesus FV we learned there are 3 broad approaches to solving econ models:
| Method | Examples | Toolkits |
|---|---|---|
| Fixed Point Iteration | VFI, PFI, EE iteration | VFI toolkit, CLMM |
| Perturbation | Linearization, LQ ... | Dynare, SolveDSGE.jl, DSGE.jl |
| Projection | Chebyshev, Smolyak, PEA, FEM | SolveDSGE.jl, TaylorProjection.m |
1 SolveDSGE.jl is the only package that allows both Perturbation and Projection (& their hybrids)
2 VFI toolkit originally planned a Julia version. It doesn't look like its coming soon.
3 Your package (CLMM) seems to be the best Julia option for FPI (allowing 7 methods!!!)
Is there any way CLMM can be automated into a toolkit?
I mean where the user creates:
ncgm: a set of parameters "ncgm"
u: a return function (flow utility)
g: a transition function for the state vector
And then something like:
solve_fpi(VFI(), ncgm, u, g, guess; maxiter=1000, verbose=false)
solve_fpi(VFI_EGM(), ncgm, u, g, guess; maxiter=1000, verbose=false)
solve_fpi(VFI_ECM(), ncgm, u, g, guess; maxiter=1000, verbose=false)
solve_fpi(dVFI_ECM(), ncgm, u, g, guess; maxiter=1000, verbose=false)
solve_fpi(PFI(), ncgm, u, g, guess; maxiter=1000, verbose=false)
solve_fpi(PFI_ECM(), ncgm, u, g, guess; maxiter=1000, verbose=false)
solve_fpi(EulEq(), ncgm, u, g, guess; maxiter=1000, verbose=false)Science would be a lot better off if this were possible...
Thanks again for your awesome package
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
No labels