Consequently drop “occbin” option to “model”.
Incidentally, allow more values in equation tag names (previously some keywords
such as “alpha” were disallowed).
Ref. #68
As the name implies, this option allows contemporaneous variables on the RHS.
The A0 matrix for contemporaneous variables is added as a second (optional)
output to the generated var_ar.m file. Note that for reduced-form VAR, this
matrix will be the identity.
Also, the user is now allowed to write the VAR models in a more flexible form:
the LHS must still be a single variable, but the RHS can be an arbitrary
expression (as long as it is linear, obviously). Internally, the preprocessor
now uses derivation to compute the coefficients of the AR and A0. This change
applies to both reduced-form and structural VAR models.
Ref. dynare#1785
For the time being, the preprocessor will refuse that this option be used with
any command other than estimation.
By the way, remove occbin_likelihood and occbin_smoother options to estimation.
Ref. dynare#569
— rename the files with camel case
— encapsulate the functions with modules
— change the signature of function <MODFILE>_dynamic_set_series!, by removing
the output argument and appending the exclamation mark, since this function
modifies one of its arguments, and for symmetry with the static version
Ref. DynareJulia/Dynare.jl#1
– Fix order of items in this structure. Previously, items were ordered
according to the declaration order of parameters. Now, items are order
according to lag order (first lag appears first)
– Gracefully handle the case where there is no autoregressive part
(Closes: #52)
Also be more strict on the form of the target (must now be X(-1) or log(X(-1))
where X is *not* an auxiliary variable).
By the way, improve some comments in SymbolTable.
The detection of the target EC variable to be used when constructing the
forward-looking expectation variable is rather fragile.
When the PAC model is written with an (non-)optimizing share of agents,
restrict the identification of the target variable to the optimizing
expression, to minimize the risk of wrong identification.
By the way, add a few comments, and a small simplification.
By default, the preprocessor is supposed to only do the “unary ops”
transformation in the equations of VAR/PAC/trend component models.
However, the implementation was slightly different so far. It would detect
candidates to this transformation in the chosen equations, but it would then
perform the substitution in *all* equations.
This could lead for crashes, for example if the chosen equation contains
log(X(-1)), but another (non-chosen) equation has log(X(-2)). Then this latter
expression, even though it belongs to the same lag-equivalence class, is not
properly handled, causing a segfault.
Also do a few related cosmetic changes.
Use 3-column format before calling sparse().
Ensure that the 3-column matrix is constructed in column-major order: data
locality will greatly improve performance once we implement use_dll.
Rather use a single vector as in non-block mode.
By the way, change the order of output arguments in static functions, to be
closer to the dynamic ones.
Temporary terms need to be computed per equation (as was done previously), and
not simply per block.
It’s necessary to track temporary terms per equation, because some equations
are evaluated instead of solved, and an equation E1 may depend on the value of
an endogenous Y computed by a previously evaluated equation E2; in this case,
if some temporary term TT of equation E2 contains Y, then TT needs to be
computed after E1, but before E2.
In particular, in dynamic models, temporary terms are now computed for
derivatives w.r.t. exogenous, and also w.r.t. endogenous variables that do not
belong to the block.
Also remove a message about elements below the cutoff that was no longer
correct (elements below the cutoff have no impact on the incidence matrix
outside of normalization).
By the way, remove the BlockType stuff which was purely informative (and it’s
not worth carrying over prologue and epilogue information just for that).
Previously, the cutoff option would also impact the block decomposition itself,
since it would had an influence on the incidence matrix used for computing the
blocks and their derivatives.
The problem is that, in the general case, it’s quite possible that an element
of the numerical Jacobian be zero at the evaluation point, while being quite
different from zero along the simulation path. A typical example is an
expression of the type x*y, where y is an endogenous and x is an exogenous not
present in the initval block (and hence initialized to zero).
It has been superseded by ModelTree::blocks_derivatives.
By the way, fix the initial number of non-zero elements in sparse Jacobian.
Also avoid computing suboptimal temporary terms.
— use a std::map for storing block derivatives
— remove redundant ModelTree::first_chain_rule_derivatives structure
— remove unused codepaths in StaticModel
— DynamicModel: simplify code that determines the type of derivatives in a
block. We now use a slightly different categorization.
— by the way, fix the max lead/lag information for blocks that are obtained via
merging. A workaround was previously implemented in
DynamicModel::get_Derivative(), but it is no longer needed with this fix.