Changelog

Source: NEWS.md

fastglmm 0.4.1

• Feb 26, 2026
• fix hypothesis testing

fastglmm 0.4.0

• Feb 17, 2026
• Hypothesis test uses Satterthwaite denominator degrees of freedom
  • updated summary(), anova() and linearHypothesis()
  • implemented at C++ level, return A.sat, B.sat, hessian.vc

fastglmm 0.3.9

• Feb 11, 2026
• improved get_rdf() for fastlmm in C++
• clean up code, consts
• add mutex to response models
• fix bug in dispersion(), uses Pearson residuals
• NB-QL
• negative.binomial(NA) fixes dispersion to 1
• negative.binomial(theta) estimates dispersion from pearson residuals and scales vcov and se
• in C++ NB() now uses QL dispersion of theta is given

fastglmm 0.3.8

• Jan 16, 2026
• more flexable isCountModel() and isNB()
• fix simulate() for NB models

fastglmm 0.3.7

• performance improvements
• add ridge regression in fastglmm and fastlmm
• in variance partitioning analysis, add faster approximation of baseline rate for count models

fastglmm 0.3.6

• add isCountModel() in C++

fastglmm 0.3.5

• Dec 8, 2025
• Fixed but in C++ code for fastglmm for mu and residuals

fastglmm 0.3.4

• Nov 24, 2025
• update varpart() for NB models
• fastglmm() and glm.nb()
• handle singular models for NB fit
• add Cox-Reid option for NB models
• looser convergence criteria for GLMM

fastglmm 0.3.3

• Oct 17, 2025
• when model fit fails return NaN values
• Update docs

fastglmm 0.3.2

• Oct 7, 2025
• add checks and compatibility with BatchRegression

fastglmm 0.3.1

• Sept 16, 2025
• add and check generics
• additional testing

fastglmm 0.3.0

• May 6, 2025
• rename
• add fastglmm() in C++
• pull code from BatchRegression

fastglmm 0.2.1

• April 24, 2025
• refactor for compatibility across ecosystem

fastglmm 0.2.0

• March 6, 2025
• Major refactor

fastglmm 0.1.6

• Nov 19, 2024
• fix API for linear and mixed model regression

fastglmm 0.1.5

• Nov 7, 2024
• linearRegression.h supports weighted regression with preprojection

fastglmm 0.1.4

• Sept 10, 2024
• move fastlmmLib code to inst/include for accessible header-only library

fastglmm 0.1.3

• Aug 28, 2024
• changes to allow PQL with fastglmm()

fastglmm 0.1.2

• Aug 7, 2024
• multivariate model is run in parallel

fastglmm 0.1.1

• Aug 6, 2024
• fix inconsistent merge

fastglmm 0.1.0

• Aug 6, 2024
• fix bottleneck in
  • model.frame() for matrix response
  • log-likelihood in Rcpp since weights are constant across iterations
  • memory usage
• pass R CMD check
• RcppArmadillo code works except doesn’t consider varying weights across responses