profile.SLik: Compute profile summary likelihood

In Infusion: Inference Using Simulation

Compute profile summary likelihood

Description

Predicts the profile likelihood for a given parameter value (or vector of such values) using predictions from an SLik_j (or older SLik) object (as produced by MSL).

Usage

## S3 method for class 'SLik_j'
profile(fitted, value, fixed=NULL, return.optim=FALSE, 
        init = "default", which="safe", 
        constr_crits=fitted$constr_crits, 
        eq_constr=NULL, ...)
## S3 method for class 'SLik'
profile(fitted, ...)

Arguments

fitted	an SLik object.
value	The parameter value (as a vector of named values) for which the profile is to be computed. Setting in explictly to NULL will maximize likelihood subject only to optional constraints specified by other arguments.
fixed	This argument appears redundant with the value argument, so it will be deprecated and its use should be avoided. When it is non-NULL, the profile is computed for value updated to c(value, fixed).
return.optim	If this is TRUE, and if maximization of likelihood given value and fixed is indeed required, then the full result of the optimization call is returned.
constr_crits	Inequality constraints, by default those provided in the first iteration of the workflow, if any. See constr_crits for details.
eq_constr	Optional equality constraints, provided in the same format as constr_crits (This feature is experimental: in particular the procedure for finding initial values for maximization of likelihood does not use the eq_constr information).
...	For SLik_j method, arguments passed to SLik method. For SLik_j method, currently not used.
init	Better ignored. Either a named vector of parameter values (initial value for some optimizations) or a character string. The default is to call a procedure to find a good initial point from a set of candidates. The source code should be consulted for further details and is subject to change without notice.
which	Better ignored (for development purpose).

Value

If return.optim is FALSE (default): the predicted summary profile log-likelihood, with possible attribute "solution", the optimization solution vector (named numeric vector, missing if no profiling was needed). if return.optim is TRUE, the result of an optimization call, a list including elements solution (solution vector) and objective (log-likelihood).

See Also

example_reftable

Examples

## see e.g. 'example_reftable' documentation

Infusion documentation built on Sept. 30, 2024, 9:16 a.m.