focal_refine: Refine summary likelihood profile in focal parameter values

In Infusion: Inference Using Simulation

Refine summary likelihood profile in focal parameter values

Description

This function refines an SLik_j object in a focused way defined by focal parameter values. It is paticularly useful to check a suspect pattern in a likelihood profile. If there is a suspect dip or peak at value <somepar>=<somevalue>, focal_refine(<SLik_j object>, focal=c(<somepar>=<somevalue>), size=<size>) will define <size> parameter points near c(<somepar>=<somevalue>) and (subject to these points being in the parameter bounds of the object) simulate new samples for these parameter points and refine the object using these new simulations.

Usage

focal_refine(object, focal, size, plotprof = TRUE, ...)

Arguments

object	An object of class SLik_j.
focal	Parameter value(s) (as a vector of named values)
size	Target number of points to add to the reference table
plotprof	Whether to replot a likelihood profile (1D or 2D depending on the dimension of focal).
...	Further arguments passed to profile.SLik_j (but not including argument return.optim) or refine.

Value

The updated object

Examples

## Not run: 
# Using the slik_j object from the toy example in help("example_reftable"):

plot1Dprof(slik_j,"s2")
slik_fix <- focal_refine(slik_j,focal=c(s2=2), size=100)
plot1Dprof(slik_fix,"s2")

# In that case the effect is not spectacular because 
# there is no major problem in the starting profile. 

## End(Not run)

Infusion documentation built on May 3, 2023, 5:10 p.m.