nlme Error in MEEM(object, conLin, control$niterEM) : Singularity in backsolve at level 0, block 1

[chansgary]

Hi Dr. Fidler and nlmixr2 users,

I have trouble fitting the following model with the data attached here. Previously, I had no trouble fitting using the same model and the same data. I came up with this error recently with even the simplest model(without any covariate).

This is crucial in one of our research questions in building PK Models. I wonder if you could help me look into this and your help will be greatly appreciated.

`library(dplyr)
sample.data.csv

sample.data = read.csv("sample.data.csv")[,-1]
mod.2CMT <- function() {
ini({
tVc <- c(-2.30, 2.68, 3.91) # log_V1 (L)
tCl <- c(0, 2.21, 4.60) # log_Cl (L/hr)
tVp <- c(-2.30, 1.53, 3.00) # log_V2 (L)
tQ <- c(-2.30, 0.46, 2.30) # log_Q (L/hr)
eta.Vc + eta.Cl ~ c(0.373973,
0.333469, 0.530645) # IIV
Aeff.Cl <- c(0.1,1.178,5) # Effect of A on CL
Beff.Vc <- c(0.1,1.56,2) # Effect of B on Vc
prop.err = c(0,0.133,1)
})
model({
Vc <- exp(tVc + eta.Vc + log(WT70)) * (1 + Beff.Vc * B)
Vp <- exp(tVp + eta.Vc + log(WT70))
Cl <- exp(tCl + eta.Cl + 0.75 * log(WT70) + Aeff.Cl * A)
Q = exp(tQ + 0.75 * log(WT70))
linCmt() ~ prop(prop.err)
})
}

nlmixr2(mod.2CMT)

fit.example = nlmixr2(mod.2CMT,sample.data,"nlme")`

The following is the error it prompts:
`ℹ parameter labels from comments will be replaced by 'label()'

→ loading into symengine environment...
→ pruning branches (if/else) of nlme model...
✔ done
→ finding duplicate expressions in nlme model...
→ optimizing duplicate expressions in nlme model...
✔ done
ld: warning: -single_module is obsolete
ld: warning: -multiply_defined is obsolete

Error in MEEM(object, conLin, control$niterEM) :
Singularity in backsolve at level 0, block 1
Error: Singularity in backsolve at level 0, block 1`

[mattfidler]

Hi @chansgary

I would suggest using focei for this problem; nlme often has this problem. Sometimes you can use a nonlinear model (like the newly added nlm estimation method) to give a better idea of the initial parameters that can get over this problem.

The following example shows the exploration I did with your example:

library(nlmixr2est)
#> Loading required package: nlmixr2data

sample.data = read.csv("~/Downloads/sample.data.csv")[,-1]

mod.2CMT <- function() {
  ini({
    tVc <- c(-2.30, 2.68, 3.91) # log_V1 (L)
    tCl <- c(0, 2.21, 4.60) # log_Cl (L/hr)
    tVp <- c(-2.30, 1.53, 3.00) # log_V2 (L)
    tQ <- c(-2.30, 0.46, 2.30) # log_Q (L/hr)
    eta.Vc + eta.Cl ~ c(0.373973,
                        0.333469, 0.530645) # IIV
    Aeff.Cl <- c(0.1,1.178,5) # Effect of A on CL
    Beff.Vc <- c(0.1,1.56,2) # Effect of B on Vc
    prop.err = c(0,0.133,1)
  })
  model({
    Vc <- exp(tVc + eta.Vc + log(WT70)) * (1 + Beff.Vc * B)
    Vp <- exp(tVp + eta.Vc + log(WT70))
    Cl <- exp(tCl + eta.Cl + 0.75 * log(WT70) + Aeff.Cl * A)
    Q = exp(tQ + 0.75 * log(WT70))
    linCmt() ~ prop(prop.err)
  })
}

nlmixr2(mod.2CMT)
#> ℹ parameter labels from comments will be replaced by 'label()'
#>  ── rxode2-based solved PK 2-compartment model ────────────────────────────────── 
#>  ── Initalization: ──  
#> Fixed Effects ($theta): 
#>      tVc      tCl      tVp       tQ  Aeff.Cl  Beff.Vc prop.err 
#>    2.680    2.210    1.530    0.460    1.178    1.560    0.133 
#> 
#> Omega ($omega): 
#>          eta.Vc   eta.Cl
#> eta.Vc 0.373973 0.333469
#> eta.Cl 0.333469 0.530645
#>  ── μ-referencing ($muRefTable): ──  
#>   theta    eta level covariates
#> 1   tCl eta.Cl    id  A*Aeff.Cl
#> 
#>  ── Model (Normalized Syntax): ── 
#> function() {
#>     ini({
#>         tVc <- c(-2.3, 2.68, 3.91)
#>         label("log_V1 (L)")
#>         tCl <- c(0, 2.21, 4.6)
#>         label("log_Cl (L/hr)")
#>         tVp <- c(-2.3, 1.53, 3)
#>         label("log_V2 (L)")
#>         tQ <- c(-2.3, 0.46, 2.3)
#>         label("log_Q (L/hr)")
#>         Aeff.Cl <- c(0.1, 1.178, 5)
#>         label("Effect of A on CL")
#>         Beff.Vc <- c(0.1, 1.56, 2)
#>         label("Effect of B on Vc")
#>         prop.err <- c(0, 0.133, 1)
#>         eta.Vc + eta.Cl ~ c(0.373973, 0.333469, 0.530645)
#>     })
#>     model({
#>         Vc <- exp(tVc + eta.Vc + log(WT70)) * (1 + Beff.Vc * 
#>             B)
#>         Vp <- exp(tVp + eta.Vc + log(WT70))
#>         Cl <- exp(tCl + eta.Cl + 0.75 * log(WT70) + Aeff.Cl * 
#>             A)
#>         Q = exp(tQ + 0.75 * log(WT70))
#>         linCmt() ~ prop(prop.err)
#>     })
#> }

fit.example = try(nlmixr2(mod.2CMT,sample.data,"nlme"))
#> ℹ parameter labels from comments will be replaced by 'label()'
#> → loading into symengine environment...
#> → pruning branches (`if`/`else`) of nlme model...
#> ✔ done
#> → finding duplicate expressions in nlme model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in nlme model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> ✔ done
#> 
#> **Iteration 1
#> LME step: Loglik: -79.20682, nlminb iterations: 45
#> reStruct  parameters:
#>       ID1       ID2       ID3 
#>  4.495867 -5.518972  8.059559 
#>  Beginning PNLS step: .. Error in nlme.formula(model = DV ~ .nlmixrNlmeFun(pars = list(tCl = tCl +  : 
#>   Singularity in backsolve at level 0, block 1
#> Error : Singularity in backsolve at level 0, block 1

mod2 <- mod.2CMT |>
  model(Vc <- exp(tVc + log(WT70)) * (1 + Beff.Vc * B)) |>
  model(Vp <- exp(tVp + log(WT70))) |>
  model(Cl <- exp(tCl + 0.75 * log(WT70) + Aeff.Cl * A))
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ! remove between subject variability `eta.Vc`
#> ! remove between subject variability `eta.Cl`

fit.example <- nlmixr2(mod2, sample.data, "nlm", nlmControl(print=0))
#> → loading into symengine environment...
#> → pruning branches (`if`/`else`) of population log-likelihood model...
#> ✔ done
#> → calculate jacobian
#> → calculate ∂(f)/∂(θ)
#> [====|====|====|====|====|====|====|====|====|====] 0:00:04
#> → finding duplicate expressions in nlm llik gradient...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in nlm llik gradient...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → finding duplicate expressions in nlm pred-only...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in nlm pred-only...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> using C compiler: ‘gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0’
#> using C compiler: ‘gcc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0’
#> rxode2 2.1.2 using 8 threads (see ?getRxThreads)
#>   no cache: create with `rxCreateCache()`
#> |-----+---------------+-----------+-----------+-----------+-----------|
#> → calculating covariance
#> ✔ done
#> → loading into symengine environment...
#> → pruning branches (`if`/`else`) of full model...
#> ✔ done
#> → finding duplicate expressions in EBE model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in EBE model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → compiling EBE model...
#> ✔ done
#> → loading llik model into symengine environment...
#> → pruning branches (`if`/`else`) of llik full model...
#> ✔ done
#> → finding duplicate expressions in Llik EBE model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in Llik EBE model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → compiling Llik EBE model...
#> ✔ done
#> → Calculating residuals/tables
#> ✔ done
#> → compress origData in nlmixr2 object, save 4432
#> → compress parHistData in nlmixr2 object, save 9448

mod3 <- mod2 |>
  model(Vc <- exp(tVc + eta.Vc + log(WT70)) * (1 + Beff.Vc * B)) |>
  model(Vp <- exp(tVp + eta.Vc + log(WT70))) |>
  model(Cl <- exp(tCl + eta.Cl + 0.75 * log(WT70) + Aeff.Cl * A)) |>
  ini(eta.Vc + eta.Cl ~ c(0.373973,
                          0.333469, 0.530645))
#> ℹ add between subject variability `eta.Vc` and set estimate to 1
#> ℹ add between subject variability `eta.Cl` and set estimate to 1
#> ℹ change initial estimate of `eta.Vc` to `0.373973`
#> ℹ add covariance between `eta.Cl` and `eta.Vc` with initial estimate `0.333469`
#> ℹ change initial estimate of `eta.Cl` to `0.530645`

fit.example = try(nlmixr2(mod3, sample.data, "nlme"))
#> → loading into symengine environment...
#> → pruning branches (`if`/`else`) of nlme model...
#> ✔ done
#> → finding duplicate expressions in nlme model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in nlme model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> ✔ done
#> 
#> **Iteration 1
#> LME step: Loglik: -79.20682, nlminb iterations: 45
#> reStruct  parameters:
#>       ID1       ID2       ID3 
#>  4.495867 -5.518972  8.059559 
#>  Beginning PNLS step: .. Error in nlme.formula(model = DV ~ .nlmixrNlmeFun(pars = list(tCl = tCl +  : 
#>   Singularity in backsolve at level 0, block 1
#> Error : Singularity in backsolve at level 0, block 1

fit.exampleF <- nlmixr2(mod.2CMT, sample.data, "focei", foceiControl(print=0))
#> ℹ parameter labels from comments will be replaced by 'label()'
#> calculating covariance matrix
#> [====|====|====|====|====|====|====|====|====|====] 0:00:08 
#> done
#> → Calculating residuals/tables
#> ✔ done
#> → compress origData in nlmixr2 object, save 4432
#> → compress parHistData in nlmixr2 object, save 7960

print(fit.exampleF)
#> ── nlmixr² FOCEi (outer: nlminb) ──
#> 
#>           OBJF      AIC      BIC Log-likelihood Condition#(Cov) Condition#(Cor)
#> FOCEi 1498.389 1593.742 1610.878       -786.871        5.733959        1.112537
#> 
#> ── Time (sec $time): ──
#> 
#>            setup optimize covariance table compress
#> elapsed 0.000926 8.136331   8.136333 0.031     0.03
#> 
#> ── Population Parameters ($parFixed or $parFixedDf): ──
#> 
#>                  Parameter  Est.      SE  %RSE Back-transformed(95%CI) BSV(CV%)
#> tVc             log_V1 (L)   2.7  0.0029 0.107         14.9 (14.8, 15)         
#> tCl          log_Cl (L/hr)  2.23 0.00304 0.137       9.29 (9.24, 9.35)     83.7
#> tVp             log_V2 (L)  1.78 0.00227 0.127        5.92 (5.9, 5.95)         
#> tQ            log_Q (L/hr)  0.46 0.00127 0.277       1.58 (1.58, 1.59)         
#> Aeff.Cl  Effect of A on CL  1.18 0.00176  0.15       1.18 (1.17, 1.18)         
#> Beff.Vc  Effect of B on Vc  1.56 0.00195 0.125       1.56 (1.56, 1.56)         
#> prop.err                   0.133                                 0.133         
#> eta.Vc                                                                     67.3
#>          Shrink(SD)%
#> tVc                 
#> tCl           24.2% 
#> tVp                 
#> tQ                  
#> Aeff.Cl             
#> Beff.Vc             
#> prop.err            
#> eta.Vc        92.6% 
#>  
#>   Covariance Type ($covMethod): r
#>   Correlations in between subject variability (BSV) matrix:
#>     cor:eta.Cl,eta.Vc 
#>            0.749  
#>  
#> 
#>   Full BSV covariance ($omega) or correlation ($omegaR; diagonals=SDs) 
#>   Distribution stats (mean/skewness/kurtosis/p-value) available in $shrink 
#>   Information about run found ($runInfo):
#>    • gradient problems with initial estimate and covariance; see $scaleInfo 
#>    • using R matrix to calculate covariance 
#>    • S matrix non-positive definite 
#>    • ETAs were reset to zero during optimization; (Can control by foceiControl(resetEtaP=.)) 
#>    • initial ETAs were nudged; (can control by foceiControl(etaNudge=., etaNudge2=)) 
#>   Censoring ($censInformation): No censoring
#>   Minimization message ($message):  
#>     false convergence (8) 
#>   In an ODE system, false convergence may mean "useless" evaluations were performed.
#>   See https://tinyurl.com/yyrrwkce
#>   It could also mean the convergence is poor, check results before accepting fit
#>   You may also try a good derivative free optimization:
#>     nlmixr2(...,control=list(outerOpt="bobyqa"))
#> 
#> ── Fit Data (object is a modified tibble): ──
#> # A tibble: 41 × 23
#>   ID     TIME     DV  PRED   RES  WRES IPRED   IRES IWRES CPRED  CRES CWRES
#>   <fct> <dbl>  <dbl> <dbl> <dbl> <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 101    16.3  0.436  208. -207. -1.25  16.1 -15.7  -7.32  56.6 -56.2 -4.71
#> 2 101    16.7 42.3    225. -183. -1.06  25.9  16.4   4.75  79.3 -37.0 -2.06
#> 3 101    17.3 22.5    220. -197. -1.17  17.1   5.40  2.37  64.3 -41.8 -3.17
#> # ℹ 38 more rows
#> # ℹ 11 more variables: eta.Vc <dbl>, eta.Cl <dbl>, Vc <dbl>, Vp <dbl>,
#> #   Cl <dbl>, Q <dbl>, tad <dbl>, dosenum <dbl>, A <dbl>, B <dbl>, WT70 <dbl>

^{Created on 2024-02-04 with reprex v2.1.0}

However, I find the assumption curious that you have exactly the same between subject varaibility between central volume and peripheral volume curious. I have not seen this very often is something to think a bit more about if you want to find a solution in nlme. It isn't purely mu-referenced and add some instability into the model as far as nlme concerned (in fact because of this I don't think it can run in saem currently.

If you wish to use saem you would have to fix this, as in the following extended example:

library(nlmixr2est)
#> Loading required package: nlmixr2data

sample.data = read.csv("~/Downloads/sample.data.csv")[,-1]

mod.2CMT <- function() {
  ini({
    tVc <- c(-2.30, 2.68, 3.91) # log_V1 (L)
    tCl <- c(0, 2.21, 4.60) # log_Cl (L/hr)
    tVp <- c(-2.30, 1.53, 3.00) # log_V2 (L)
    tQ <- c(-2.30, 0.46, 2.30) # log_Q (L/hr)
    eta.Vc + eta.Cl ~ c(0.373973,
                        0.333469, 0.530645) # IIV
    Aeff.Cl <- c(0.1,1.178,5) # Effect of A on CL
    Beff.Vc <- c(0.1,1.56,2) # Effect of B on Vc
    prop.err = c(0,0.133,1)
  })
  model({
    Vc <- exp(tVc + eta.Vc + log(WT70)) * (1 + Beff.Vc * B)
    Vp <- exp(tVp + log(WT70))
    Cl <- exp(tCl + eta.Cl + 0.75 * log(WT70) + Aeff.Cl * A)
    Q = exp(tQ + 0.75 * log(WT70))
    linCmt() ~ prop(prop.err)
  })
}

nlmixr2(mod.2CMT)
#> ℹ parameter labels from comments will be replaced by 'label()'
#>  ── rxode2-based solved PK 2-compartment model ────────────────────────────────── 
#>  ── Initalization: ──  
#> Fixed Effects ($theta): 
#>      tVc      tCl      tVp       tQ  Aeff.Cl  Beff.Vc prop.err 
#>    2.680    2.210    1.530    0.460    1.178    1.560    0.133 
#> 
#> Omega ($omega): 
#>          eta.Vc   eta.Cl
#> eta.Vc 0.373973 0.333469
#> eta.Cl 0.333469 0.530645
#>  ── μ-referencing ($muRefTable): ──  
#>   theta    eta level covariates
#> 1   tVc eta.Vc    id           
#> 2   tCl eta.Cl    id  A*Aeff.Cl
#> 
#>  ── Model (Normalized Syntax): ── 
#> function() {
#>     ini({
#>         tVc <- c(-2.3, 2.68, 3.91)
#>         label("log_V1 (L)")
#>         tCl <- c(0, 2.21, 4.6)
#>         label("log_Cl (L/hr)")
#>         tVp <- c(-2.3, 1.53, 3)
#>         label("log_V2 (L)")
#>         tQ <- c(-2.3, 0.46, 2.3)
#>         label("log_Q (L/hr)")
#>         Aeff.Cl <- c(0.1, 1.178, 5)
#>         label("Effect of A on CL")
#>         Beff.Vc <- c(0.1, 1.56, 2)
#>         label("Effect of B on Vc")
#>         prop.err <- c(0, 0.133, 1)
#>         eta.Vc + eta.Cl ~ c(0.373973, 0.333469, 0.530645)
#>     })
#>     model({
#>         Vc <- exp(tVc + eta.Vc + log(WT70)) * (1 + Beff.Vc * 
#>             B)
#>         Vp <- exp(tVp + log(WT70))
#>         Cl <- exp(tCl + eta.Cl + 0.75 * log(WT70) + Aeff.Cl * 
#>             A)
#>         Q = exp(tQ + 0.75 * log(WT70))
#>         linCmt() ~ prop(prop.err)
#>     })
#> }

fit.example = nlmixr2(mod.2CMT,sample.data,"saem", saemControl(print=0))
#> ℹ parameter labels from comments will be replaced by 'label()'
#> → loading into symengine environment...
#> → pruning branches (`if`/`else`) of saem model...
#> ✔ done
#> → finding duplicate expressions in saem model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in saem model...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> ✔ done
#> rxode2 2.1.2 using 8 threads (see ?getRxThreads)
#>   no cache: create with `rxCreateCache()`
#> NaN in prediction; Consider: relax atol & rtol; change initials; change seed; change structural model
#>   warning only issued once per problem
#> Calculating covariance matrix
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → loading into symengine environment...
#> → pruning branches (`if`/`else`) of saem model...
#> ✔ done
#> → finding duplicate expressions in saem predOnly model 0...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → finding duplicate expressions in saem predOnly model 1...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in saem predOnly model 1...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → finding duplicate expressions in saem predOnly model 2...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> → optimizing duplicate expressions in saem predOnly model 2...
#> [====|====|====|====|====|====|====|====|====|====] 0:00:00
#> ✔ done
#> → Calculating residuals/tables
#> ✔ done
#> → compress origData in nlmixr2 object, save 4432
#> → compress phiM in nlmixr2 object, save 152256
#> → compress parHistData in nlmixr2 object, save 22064
#> → compress saem0 in nlmixr2 object, save 30560

print(fit.example)
#> ── nlmixr² SAEM OBJF by FOCEi approximation ──
#> 
#>  Gaussian/Laplacian Likelihoods: AIC() or $objf etc. 
#>  FOCEi CWRES & Likelihoods: addCwres() 
#> 
#> ── Time (sec $time): ──
#> 
#>           setup covariance  saem table compress    other
#> elapsed 0.00137   0.011007 8.335 0.075    0.026 1.423623
#> 
#> ── Population Parameters ($parFixed or $parFixedDf): ──
#> 
#>                  Parameter  Est.    SE %RSE Back-transformed(95%CI) BSV(CV%)
#> tVc             log_V1 (L)  2.57   0.6 23.4         13 (4.02, 42.3)     8.75
#> tCl          log_Cl (L/hr)  2.41 0.134 5.56       11.1 (8.57, 14.5)     13.4
#> tVp             log_V2 (L)  1.12 0.515 45.9       3.07 (1.12, 8.42)         
#> tQ            log_Q (L/hr) 0.875 0.191 21.8        2.4 (1.65, 3.49)         
#> Aeff.Cl  Effect of A on CL  2.18 0.208 9.56       2.18 (1.77, 2.58)         
#> Beff.Vc  Effect of B on Vc  0.79 0.836  106     0.79 (-0.849, 2.43)         
#> prop.err                   0.295                              0.295         
#>          Shrink(SD)%
#> tVc          -21.9% 
#> tCl          -21.8% 
#> tVp                 
#> tQ                  
#> Aeff.Cl             
#> Beff.Vc             
#> prop.err            
#>  
#>   Covariance Type ($covMethod): linFim
#>   Correlations in between subject variability (BSV) matrix:
#>     cor:eta.Cl,eta.Vc 
#>            -1.00  
#>  
#> 
#>   Full BSV covariance ($omega) or correlation ($omegaR; diagonals=SDs) 
#>   Distribution stats (mean/skewness/kurtosis/p-value) available in $shrink 
#>   Information about run found ($runInfo):
#>    • log-scale mu referenced time varying covariates (Aeff.Cl) may have better results on no log-transformed scale (https://github.com/nlmixr2/nlmixr2est/issues/348), check results for plausibility 
#>   Censoring ($censInformation): No censoring
#> 
#> ── Fit Data (object is a modified tibble): ──
#> # A tibble: 41 × 19
#>   ID     TIME     DV  PRED   RES IPRED   IRES IWRES  eta.Vc eta.Cl    Vc    Vp
#>   <fct> <dbl>  <dbl> <dbl> <dbl> <dbl>  <dbl> <dbl>   <dbl>  <dbl> <dbl> <dbl>
#> 1 101    16.3  0.436  114. -114.  103. -102.  -3.37 -0.0684  0.105  5.12 0.438
#> 2 101    16.7 42.3    144. -102.  134.  -91.6 -2.32 -0.0684  0.105  5.12 0.438
#> 3 101    17.3 22.5    129. -106.  116.  -93.4 -2.73 -0.0684  0.105  5.12 0.438
#> # ℹ 38 more rows
#> # ℹ 7 more variables: Cl <dbl>, Q <dbl>, tad <dbl>, dosenum <dbl>, A <dbl>,
#> #   B <dbl>, WT70 <dbl>

^{Created on 2024-02-04 with reprex v2.1.0}

[mattfidler]

Also, you could try to use babelmixr2 to streamline these comparisons. However, the conversion may not exactly be 1:1. It changes the model to mu-referenced (if it can) as well as adding expressions to avoid dividing by zero. It then runs it in NONMEM and compares the results to what rxode2 would give as a table step with the same results.

If you have any issues there you can also add them to that package.

[chansgary]

May I also ask what version of nlmixrest/nlmixr2/rxode2 that you are currently using? I think I use all the development latest version of nlmixr2 packages as stated in (https://nlmixr2.org/). I am using 2.2.1 version of nlmixr2est.

This program built for i386-apple-darwin11.3.0
✔ The package 'rxode2' is installed and seems to be up to date, version 2.1.2
✔ The package 'rxode2et' is installed and seems to be up to date, version 2.0.12
✔ The package 'rxode2parse' is installed and seems to be up to date, version 2.0.18
✔ The package 'rxode2ll' is installed and seems to be up to date, version 2.0.11.9000
✔ The package 'rxode2random' is installed and seems to be up to date, version 2.0.13
✔ The package 'lotri' is installed and seems to be up to date, version 0.4.4
✔ The package 'nlmixr2' is installed and seems to be up to date, version 2.1.1
✔ The package 'nlmixr2est' is installed and seems to be up to date, version 2.2.1
✔ The package 'nlmixr2data' is installed and seems to be up to date, version 2.0.9
✔ The package 'nlmixr2extra' is installed and seems to be up to date, version 2.0.9
✔ The package 'nlmixr2plot' is installed and seems to be up to date, version 2.0.8
✔ The package 'nlmixr2lib' is installed and seems to be up to date, version 0.2.0.9000
✖ The package 'nonmem2rx' is installed but is not the current version, installed version: 0.1.2, current version: 0.1.3
✖ The package 'babelmixr2' is installed but is not the current version, installed version: 0.1.1, current version: 0.1.2

[mattfidler]

> nlmixr2CheckInstall()
ℹ Operating system: Linux 6.6.10-76060610-generic #202401051437~1704728131~22.04~24d69e2 SMP PREEMPT_DYNAMIC Mon J
Loading required namespace: devtools
✔ The 'make' command was found:  /usr/bin/make
ℹ GNU Make 4.3
Built for x86_64-pc-linux-gnu
Copyright (C) 1988-2020 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
✔ The package 'rxode2' is installed and seems to be up to date, version 2.1.2
✔ The package 'rxode2et' is installed and seems to be up to date, version 2.0.12
✔ The package 'rxode2parse' is installed and seems to be up to date, version 2.0.18
✔ The package 'rxode2ll' is installed and seems to be up to date, version 2.0.11
✔ The package 'rxode2random' is installed and seems to be up to date, version 2.0.13
✔ The package 'lotri' is installed and seems to be up to date, version 0.4.3
✔ The package 'nlmixr2' is installed and seems to be up to date, version 2.1.1
✔ The package 'nlmixr2est' is installed and seems to be up to date, version 2.2.1
✔ The package 'nlmixr2data' is installed and seems to be up to date, version 2.0.9
✔ The package 'nlmixr2extra' is installed and seems to be up to date, version 2.0.9
✔ The package 'nlmixr2plot' is installed and seems to be up to date, version 2.0.8
✔ The package 'nlmixr2lib' is installed and seems to be up to date, version 0.2.0
✔ The package 'nonmem2rx' is installed and seems to be up to date, version 0.1.3
✔ The package 'babelmixr2' is installed and seems to be up to date, version 0.1.2

I also have nlme_3.1-164.

There could be platform differences; these are common for R on the mac platform (which you appear to be running).

If you have the version difference for when nlme was working, we can look further into what is going on.

[chansgary]

Hi Dr. Fidler @mattfidler ,

I have the same versions of all of the packages listed above(both nlmixr2-related and nlme), while the "nlm" still did not work on my end. Also for the "nlme", I am also unsure why it previously worked out smoothly three months ago, and all of a sudden it failed. Under this circumstance, do you have any suggestions? e.g., downgrade both nlme and nlmixr2 to a previous version?

[mattfidler]

Hi @chansgary I do not have a mac personally, so it is difficult to diagnose. @john-harrold, @kestrel99 and @RichardHooijmaijers can you run the nlm examples?

[mattfidler]

This is more of a question than a bug so I am recategorizing this.

[mattfidler]

@chansgary I am sure that the other folks are busy; If you reach out to my email address maybe I can setup a MS Teams call to help you with your setup issue. (matt dot fidler at novartis dot com)