Error after Theta reset due to ID being a factor

[billdenney]

This may be related to #454

I'm fitting a model oddly while working on likelihood profiling. This did not converge (and it may never converge due to the tv value being far from the correct value), but the reason it didn't converge may be something that should be fixed where the error of "Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'." did not seem like it should be relevant.

library(nlmixr2)
#> Loading required package: nlmixr2data

mod <- function() {
  ini({
    tka <- 0.457920524576555
    tcl <- 1.01463922110082
    tv <- fix(7.42588307259747)
    add.sd <- c(0, 0.693895455003917)
    eta.ka ~ 0.406807327013617
    eta.cl ~ 0.0684420360035659
    eta.v ~ 0.019933306613011
  })
  model({
    ka <- exp(tka + eta.ka)
    cl <- exp(tcl + eta.cl)
    v <- exp(tv + eta.v)
    cp <- linCmt()
    cp ~ add(add.sd)
  })
}

nlmixr2(mod, data = theo_sd, est = "focei")
#> rxode2 2.1.3.9000 using 8 threads (see ?getRxThreads)
#>   no cache: create with `rxCreateCache()`
#> Key: U: Unscaled Parameters; X: Back-transformed parameters; G: Gill difference gradient approximation
#> F: Forward difference gradient approximation
#> C: Central difference gradient approximation
#> M: Mixed forward and central difference gradient approximation
#> Unscaled parameters for Omegas=chol(solve(omega));
#> Diagonals are transformed, as specified by foceiControl(diagXform=)
[ deleted a bunch of lines ]
#> |  266|     622.72340 |    -1.053 |     1.706 |     8.741 |    -1.545 |
#> |.....................|   -0.9078 |     10.09 |...........|...........|
#> |    U|      622.7234 |    0.4050 |     3.215 |     3.999 |    0.2409 |
#> |.....................|     1.307 |     6.077 |...........|...........|
#> |    X|      622.7234 |     1.499 |     24.91 |     3.999 |    0.2409 |
#> |.....................|     1.307 |     6.077 |...........|...........|
#> Theta reset (ETA drift)
#> |  267|     7364.7264 |    -1.000 |   -0.4947 |   -0.7858 |   -0.2791 |
#> |.....................|    0.3589 |     1.000 |...........|...........|
#> |    U|     7364.7264 |    0.4579 |     1.015 |    0.6939 |     1.252 |
#> |.....................|     1.955 |     2.661 |...........|...........|
#> |    X|     7364.7264 |     1.581 |     2.758 |    0.6939 |     1.252 |
#> |.....................|     1.955 |     2.661 |...........|...........|
#> Error in .foceiFitInternal(.env) : Starting values violate bounds
#> Restart 1
#> Error in .foceiFitInternal(.ret) : 
#>   Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'.
#> Restart 2
#> Error in .foceiFitInternal(.ret) : 
#>   Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'.
#> Restart 3
#> Error in .foceiFitInternal(.ret) : 
#>   Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'.
#> Error : Could not fit data
#>   Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'.
#> Error: Could not fit data
#>   Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'.

^{Created on 2024-08-20 with reprex v2.1.1}

[billdenney]

I think that there may be two things happening here:

1. Something around the reported error of Assertion on 'fitEnv$etaObj$ID' failed: Must be of type 'integer', not 'factor'. I think that error is happening here:

   nlmixr2est/R/focei.R

   Line 1411 in 21c4721

   checkmate::assertInteger(.ret$etaObf$ID, any.missing=FALSE, min=1, .var.name="fitEnv$etaObj$ID")

2. A theta reset should not occur on fixed parameters.