\(\newcommand{\B}[1]{ {\bf #1} }\) \(\newcommand{\R}[1]{ {\rm #1} }\)
mixed_ctor#
View page sourceMixed Class Constructor#
Syntax#
# mixed_obj = cppad_py.mixed(
fixed_init = None,
random_init = None,
quasi_fixed = False,
bool_sparsity = False,
A_rcv = None,
warning = None,
fix_likelihood = None,
fix_constraint = None,
ran_likelihood = None,
# )
mixed_obj#
We refer to the value returned by this constructor as mixed_obj.
fixed_init#
is a numpy vector with float elements.
It specifies a value of the fixed effects for which the
likelihood and prior functions can be evaluated and is used to
initialize mixed_obj.
This vector must be non-empty and the default value None is not valid.
n_fixed#
We use the notation n_fixed for the length of fixed_init .
random_init#
is a numpy vector with float elements.
It specifies a value of the random effects for which the
likelihood and prior functions can be evaluated and is used to
initialize mixed_obj.
The default value for this argument None corresponds
to the empty vector.
n_random#
We use the notation n_random for the length of random_init .
quasi_fixed#
is True (False) if a quasi-Newton method (Newton method) is used to optimize the fixed effects. The Newton method requires computation of second derivatives.
bool_sparsity#
is True (False) if CppAD should use boolean sparsity patterns (set sparsity patterns) for its internal calculations
A_rcv#
Is a sparse_rcv representation of the
random constraint matrix \(A\); i.e.
\(A \cdot \hat{u} ( \theta ) = 0\)
where \(\hat{u} ( \theta )\) is the
optimal random effects as a function of the fixed effects.
The value None corresponds to no random constraint.
warning#
is a python function that gets called when mixed_obj
has a warning to report; see Mixed Class Warnings.
The value None corresponds to ignoring all warning messages.
fix_likelihood#
see Fixed Effects Likelihood .
The value None corresponds to no fixed effects likelihood.
fix_constraint#
see Fixed Effects Constraint Function .
The value None corresponds to no constraint function
for the fixed effects (one can still have bound constraints).
ran_likelihood#
see Random Effects Likelihood .
The value None corresponds to no random effects likelihood.