mixed_info_table#

View page source

The CppAD Mixed Information Table#

Discussion#

A new version of this table is created each time a fit_command is run. The table contains information about the size of certain cppad_mixed objects. It is useful for seeing how the problem setup affects the size (and hence time) of the problem. For example, how different the size is when one uses a hold_out_command to sub-sample the data.

Format#

Each row of this table has the following information:

mixed_info_id#

This column has type integer and is the primary key for this table. Its initial value is zero, and it increments by one for each row.

mixed_name#

This column has type text and has a different value for every row; i.e., the names are unique and can act as substitutes for the primary key. Some of the possible values for mixed_name are listed below. There may be other (unspecified) rows in this table.

mixed_value#

This column has type text and is the value for the corresponding name.

n_fixed#

If mixed_name is n_fixed , mixed_value is the number of fixed effects.

n_random#

If mixed_name is n_random , mixed_value is the number of fixed effects.

quasi_fixed#

If mixed_name is quasi_fixed , mixed_value is one (zero) if we are a using a quasi-Newton (Newton) method for optimizing the fixed effects.

A_nr#

If mixed_name is A_nr , mixed_value is the number of rows in the liner constraint matrix A (the matrix has n_fixed columns).

A_nnz#

If mixed_name is A_nnz , mixed_value is the number of non-zeros in the liner constraint matrix A.

ran_like_fun.size_var#

If mixed_name is ran_like_fun.size_var , mixed_value is the number of variables in the algorithm that maps the fixed and random effects to the part of the likelihood that depends on the random effects.

fix_like_fun.size_var#

If mixed_name is fix_like_fun.size_var , mixed_value is the number of variables in the algorithm that maps the fixed effects to the part of the likelihood that does not depend on the random effects.