WORK IN PROGRESS

• Julia 64bit >= v1.6
• Recommended version: Julia 64 bit v1.10

• https://DynareJulia.github.io/Dynare.jl

using Pkg
pkg"add Dynare"

If you already have a version of Dynare installed (not in development mode):

using Pkg
pkg"update"

Often, it is necessary to restart Julia to load the new version of Dynare.

Note that the version of the Dynare package is printed immediately after running Dynare:

julia> context = @dynare "example1";
Dynare version: 0.9.6
...

Example (to be run in the directory Dynare.jl):

using Dynare
context = @dynare "./test/models/example1/example1.mod";

The results are in the context structure.

1. calib_smoother
2. deterministic_trends
3. endval
4. estimated_params
5. estimated_params_init
6. estimation
7. histval
8. homotopy
9. initval
10. initval_file
11. perfect_foresight_setup (only some options)
12. perfect_foresight_solver (only some options, includind lmmcp)
13. planner_objective
14. ramsey_constraints
15. ramsey_model
16. shocks
17. steady (including numerical solution)
18. stoch_simul (only order=1)

1. The context structure is saved in the directory <path to modfile>/<modfilenane>/output/<modfilename>.jls. It can be loaded with

   using Serialization
   DD = Serialization.deserialize("<path to modfile>/<modefilename>/output/<modefilename>.jls")``
   

2. Graphs are saved in <path to modfile>/<modfilenane>/graphs
3. A log of the session is kept in <filename>.log.

• https://www.dynare.org

1. PATH algorithm for MCP problems

• You need to load PATHSolver before running Dynare (once per Julia session)

    using PATHSolver
    context = @dynare ...

1. PARDISO is high-performance solver for very large sparse linear systems. It can be used with perfect foresight simulation of very large models. You need to load MKL and Pardiso before running Dynare (once per Julia session)

   using MKL, Pardiso
   context = @dynare ...

Dynare uses the PATH software by S. Dirkse, M.C. Ferris and T. Munson (https://pages.cs.wisc.edu/~ferris/path.html), as provided by PATHSolver.jl to solve prefect foresight models with occasionally binding constraints.

In order to use it, you need to add the free licence available at https://pages.cs.wisc.edu/~ferris/path/LICENSE in your file ~/.julia/config/startup.jl

ENV["PATH_LICENSE_STRING"] = "licence number provided in the above link"

1. Create a new folder in your .julia directory called config
2. In the .julia\config folder create a new text file: Right click → New → Text Document
3. Inside the new text document, type:

ENV["PATH_LICENSE_STRING"] = "licence number provided in the above link"

1. In the text document, go File → Save as
2. In the dropdown Save as type option at the bottom of the pop-up window select All files
3. In the field File name write startup.jl
4. Press Save

1. Go to File → Open Folder, navigate through your folders and choose the one where you want to create the startup.jl file
2. Go to File → New Text File
3. In the new text file, type

ENV["PATH_LICENSE_STRING"] = "licence number provided in the above link"

1. Go to File → Save
2. In the dropdown Save as type option at the bottom of the pop-up window select All files
3. In the field File name write startup.jl.
4. Press Save

1. With older Julia versions, in some circumstances, it is necessary to install first Pardiso, then build it, and finally install Dynare.jl
2. Perfect foresight simulations for purely backward models (without any forward look isn't supported yet

• Banque de France
• DSGE-net