I have been using the modular lme4 functions to customize some of our problems and I noticed that the control argument seems to be ignored in the modular functions or I am just not placing it in the right place. Here a minimal example based on the rpubs from https://rpubs.com/bbolker/groupmembers

can you help me find what am I missing that I can’t change the optimizer using the modular lme4 functions?

nm <- 20
nobs <- 500
set.seed(101)
## choose items for observations
pres <- matrix(rbinom(nobs*nm, prob=0.25, size=1),nrow=nobs,ncol=nm)
dimnames(pres) <- list(NULL,LETTERS[seq(nm)])
pres[1:5,]

b <- rnorm(nm)  ## item-level effects
## n.b. get in trouble if we don't add residual error
## (theta is scaled relative to residual error)
## here, use theta=sigma=1
y <- c(pres %*% b) +rnorm(nobs,sd=1)

## helpful to specify a factor with the right levels:
## actual values are unimportant since we will specify Zt/Ztlist directly
fake <- rep(LETTERS[seq(nm)],length.out=nobs)
lmod <- lFormula(y~1+(1|fake), control = lmerControl(
  optimizer = "bobyqa",
  optCtrl=list(maxeval=1)
) )
lmod$reTrms$Zt <- lmod$reTrms$Ztlist[[1]] <- Matrix(t(pres))
devfun <- do.call(mkLmerDevfun, lmod)
opt <- optimizeLmer(devfun)
m1 <- mkMerMod(environment(devfun), opt, lmod$reTrms, fr = lmod$fr)
summary(m1)
m1@optinfo$optimizer # does not coincide with requested optimizer
# [1] "nloptwrap"