fastglmm - Fast (Generalized) Linear Mixed Models via the 'glmm' Rust
Kernel
Fits linear models, generalized linear models, and
(generalized) linear mixed models with a compiled Rust backend
(the 'glmm' crate): linear mixed models by REML or maximum
likelihood, and Laplace or adaptive Gauss-Hermite quadrature
for binomial, Poisson, negative-binomial, and Gamma mixed
models, using 'lme4'-style formulas. Estimates are validated
against 'lme4', 'glmmTMB', and 'MixedModels.jl'. Deliberately
scoped to fast fitting - fixed effects, Wald standard errors,
variance components, and tests of fixed terms and of two nested
models (anova(), compare()), with an exact restricted
likelihood-ratio test for one random intercept (Crainiceanu and
Ruppert (2004) <doi:10.1111/j.1467-9868.2004.00438.x>);
anything the backend cannot compute honestly (random-effect
predictions, profiling) is a documented error, never a silently
different answer.
