SCM

Forum: gecon

Monitor Forum | Start New Thread Start New Thread
RE: Steady_state equilibrium cannot be found [ Reply ]
By: Karol Podemski on 2021-03-14 13:37
[forum:48830]
The Jacobian based methods may indeed get stuck when the Jacobian is ill-conditioned. You cannot make such methods work, however you can change the solver to Broyden and set use_jac = FALSE in the steady_state function. If this does not help, you may try to perturb the initial values for variables and parameters (if you are using the previous equilibrium values) or reset them to the default initial values.

Steady_state equilibrium cannot be found [ Reply ]
By: Daniel Marais on 2020-09-10 16:10
[forum:48826]
I have implemented the IFPRI (International Food Policy Research Institute) standard CGE model from the GAMS-version into Gecon and can achieve calibration equilibrium at the initial variable values.

However, as soosn as I introduce a minor shock to one of the parameters the steady_state function cannot generate a new equilibrium state: always a problem with the jacobian being ill-conditioned.

I made sure that the IFPRI model is exactly replicated in Gecon and also turned on the "tryreduce" block to enable a reduction in variables. However, there is no improvement in the condition of the jacobian and no equilibrium can be achieved while the GAMS version handles this without any issue.

Is there perhaps a way that one can switch on the "allowSingular" control condition in the nleqslv package from gecon?

Any advice will be appreciated.

Regards
Daniel Marais


Thanks to:
Vienna University of Economics and Business Powered By FusionForge