Skip to content

HANS Solver fo heterogeneous agents with discrete choice stationary equilibrium models.

License

Notifications You must be signed in to change notification settings

neoabs/HANSsolver.jl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

HANSsolver.jl

HANS Solver is a toolbox for finding stationary equilibria of heterogeneous agents with discrete choice models. HANS stands for Heterogeneous Agents with Noncontinous Strategy. Was made on Julia 1.4.1

The use is straightforward:

] add https://github.com/neoabs/HANSsolver.jl
using HANSsolver

There might be some warnings during compilation: these should be ignored I might address them in the near future.

I uploaded the algorithm description in pdf format. See it if need more deep insight into workings of this toolbox. Take note that after pdf was generated some vital changes were submited to solver so solution times in document might be affected.

In the folder "example" an instance of ModelInit is implemented. On it's own Model does not makes snece (might change in the future) but it can be used as a reference point how to use HANS Solver.

TODO List

Since it is the first public release of the toolbox, many things are not yet perfectly polished while some functionalities are missing. If someone wants to help → let me know.

  • Integrate with Julia 1.6. Should work out of the box but you never know.

  • Allign naming convention with the one in PDF document. This does not affects functionality of HANS Solver, however naming convention in PDF is more Julia-like.

  • Add GPU computing support. In the past I tried using ArrayFire.jl but it turned out to lack some functionalities I needed. I tried AMDGPU.jl but could not install ROCm on my distro. I recently switched to xUbuntu so if I find some spare time I might try that.

  • Dynamic griding. Genrally something that would remove grid points that are not used by agents and move them to area where Ψ is positive.

  • TODO from the example. I realized that there is a need for user to define an custom object after Prices are updated and Additional Prices are calculated.

  • Upgrade example. Implementing economically sound model + adding Model description in PDF would do.

  • Implement Druedahl [2020] and Druedahl and Jørgensen [2017] algorithms. This method is not compatible with multi-threading in it's raw form BUT with few tricks it might work.

Thanks

The development of HANS Solver and corresponding research is financed by National Science Centre, Poland. A project id: 2018/29/N/HS4/01225. Also great thanks to Patrick Kofod Mogensen for advice and help.

About

HANS Solver fo heterogeneous agents with discrete choice stationary equilibrium models.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages