spaMM (version 3.5.0)

div_info: Information about numerical problems

Description

This function displays information about parameter values for which some numerical problems have occurred. It is prelimnary, has yet no good example of application, but some warnings might suggest its use.

Usage

div_info(object, ...)

Arguments

object

An object of class HLfit, as returned by the fitting functions in spaMM.

…

Currently not used

Value

Used mainly for the side effects (printed output) but returns invisibly either a single parameter vector (if a single numerical problem occurred) or a matrix of parameter ranges.

Examples

# NOT RUN {
if (spaMM.getOption("example_maxtime")>25) {
  set.seed(1L)
  d <- data.frame(y = rbinom(100, 1, 0.5), time = 1:100)
  (logLik(numpb <- fitme(y ~ 1 + AR1(1|time), data = d, 
                         family = binomial(link=cauchit), method = "PQL")))
  div_info(numpb) # High lambda is often part of the problem,
                  # here only for negative AR parameter:      
  fitme(y ~ 1 + AR1(1|time), data = d, family = binomial(link=cauchit), 
        method = "PQL", lower=list(ARphi=0))
  # Less successful fix (or cover-up):
  fitme(y ~ 1 + AR1(1|time), data = d, family = binomial(link=cauchit), 
        method = "PQL", upper=list(lambda=20))
}
# }