How to extract converge information in depmixs4 fit.mod in the form of 1 & 0?

Question

In package depmixs4, converge information of the HMM is shown in print(fit.mod). However I'm trying to extract the value. For instance, when it converges I get 1. When it doesn't I get 0.

In this way when it doesn't converge I can automatically change to HMM with less states. But I didn't find a solution in the manual.

    for(i in 3000:(length(p[,2])-252*5)){
  mod <- depmix(p[(i):(252*5-1+i),2] ~ 1, data = p[(i):(252*5-1+i),], nstates = 3, family = gaussian())
  #set.seed(1)
  fit.mod <- fit(mod)}

print(fit.mod) will get

print(fit.mod)
Convergence info: 'maxit' iterations reached in EM without convergence. 
'log Lik.' 3478.027 (df=14)
AIC:  -6928.053 
BIC:  -6856.109 

The problem is to extract the information "iterations reached in EM without convergence" as a number.

Answer 1

The convergence return value is assigned in function fit. You can see exactly where in the code with

getMethods("fit", signature = "mix")

and look for the optimizers Rdonlp2::dolnp2, rsolnp and solnp.

A hack can be to extract the slot message from the object returned by fit and grepl an appropriate string.

converged <- function(object, numeric = TRUE){
  msg <- object@message
  if(numeric) as.integer(grepl("converged", msg)) else msg
}

converged(fm)
#[1] 1

As a test object fm I have used an example from help('fit').

library(depmixS4)

data(speed) 
mod <- depmix(list(rt~1,corr~1),data=speed,nstates=2,
              family=list(gaussian(),multinomial("identity")),ntimes=c(168,134,137))
# print the model, formulae and parameter values
mod
set.seed(1)
# fit the model by calling fit
fm <- fit(mod, verbose = TRUE)