converged: Test a model for convergence

In cvoeten/buildmer: Stepwise Elimination and Term Reordering for Mixed-Effects Regression

Test a model for convergence

Description

Test a model for convergence

Usage

converged(model, singular.ok = FALSE, grad.tol = 0.1, hess.tol = 0.01)

Arguments

model	The model object to test.
singular.ok	A logical indicating whether singular fits are accepted as ‘converged’ or not. Relevant only for lme4 models.
grad.tol	The tolerance to use for checking the gradient. This is currently only used by mgcv, glmmTMB, and clm(m) models.
hess.tol	The tolerance to use for checking the Hessian for negative eigenvalues. This is currently only used by mgcv, glmmTMB, and clm(m) models.

Value

Logical indicating whether the model converged.

Examples

library(buildmer)
library(lme4)
good1 <- lm(Reaction ~ Days,sleepstudy)
good2 <- lmer(Reaction ~ Days + (Days|Subject),sleepstudy)
bad <- lmer(Reaction ~ Days + (Days|Subject),sleepstudy,control=lmerControl(
            optimizer='bobyqa',optCtrl=list(maxfun=1)))
sapply(list(good1,good2,bad),converged)

cvoeten/buildmer documentation built on March 3, 2023, 3:25 p.m.