## Identification

### Local identification

- keyword + option in preprocessor
- entry function: dynare_identification (move to special directory)
- polishing
change default of prior_mc to 1 => set parameter values at which to check identification: posterior mode, prior mode, posterior mean... (we should probably introduce a syntax that will be reused for stoch_simul after estimation, for the smoother, for forecast after estimation...)

- add declaration of parameter list (after the list of options, like the list of variables for stoch_simul)
rename options_ident.prior_range -> use_prior_shape (+ preprocessor)

- streamline output
- rank condition
- if not satisfied, why ?

- needs documentation + test

- further developments:
- SVD decomposition of Jacobian matrix
- derivative of Kalman filter for derivative of information matrix
- (also useful for optimization)

- References: Helsinki Conference paper + papers cited there in

### Global identification

- exploration stage
- consequences for unicity of steady state
provide 2nd derivatives of the static model Ticket #127

## Steady State

- check global/local use of M_.params when some parameters are changed in the computation of the steady state (partially done)
- steady_state_model:
- explore possibility of providing 1st and 2nd derivatives of analytical steady state with respect to parameters
- provide for call to fsolve() with Dynare model sub-blocs (equation tag ?) (to be implemented)
- exceptions (return with penalty, alternative functional forms such as Cobb-Douglas as a degenerate case of CES) for some parameter and endogenous variable values
- allow # expressions in steady_state_model (derivatives are transparent)
- allow for modification of parameters in steady_state_model (! derivatives)

STEADY_STATE operator: derivative with respect to parameter is NOT zero in dynamic model. Derivative of steady state with respect to parameter needs to be provided to dynamic model derivator (current bug, Ticket #128)