The hessian matrix at the "mode" is not positive definite!

[qing23185]

POSTERIOR KERNEL OPTIMIZATION PROBLEM!
(minus) the hessian matrix at the "mode" is not positive definite!
=> posterior variance of the estimated parameters are not positive.
You should try to change the initial values of the parameters using
the estimated_params_init block, or use another optimization routine.

I don't know what i should do.
Thanks!

[jpfeifer]

You have an identification problem in your model:

Testing prior mean

All parameters are identified in the model (rank of H).


WARNING !!!
The rank of J (moments) is deficient!

e_sigmaqU is not identified by J moments!
[dJ/d(e_sigmaqU)=0 for all J moments!]
rho1_sigmaqU is not identified by J moments!
[dJ/d(rho1_sigmaqU)=0 for all J moments!]

[e_epsilU,e_lambdafU] are PAIRWISE collinear (with tol = 1.e-10) !
[rho1_epsilU,rho1_lambdafU] are PAIRWISE collinear (with tol = 1.e-10) !

For example, sigmaqU seems to not enter any model equation.

[qing23185]

Thanks!
I already fixed the identification problems.
Now Testing prior mean
All parameters are identified in the model (rank of H).
All parameters are identified by J moments (rank of J)

but i still got.. POSTERIOR KERNEL OPTIMIZATION PROBLEM!
(minus) the hessian matrix at the "mode" is not positive definite!

where I was wrong?

[jpfeifer]

Try a different mode-finder. You are clearly not at a global mode for some parameters.

[ZBCPA]

Try a different mode-finder. You are clearly not at a global mode for some parameters.

Dear Johannes, do you mean if at a global mode, the hessian matrix at the mode must be positive definite?

[jpfeifer]

If it's an interior maximum, yes that must be the case. (More precisely, minus the Hessian must be positive definite, see https://en.wikipedia.org/wiki/Second_partial_derivative_test)