Removes unnecessary model preparation steps for parameter_expansion="translation" and factorization_method="strong". Improves speed on default settings for models with many random effects.
Updated references in documentation.
Small adjustment to tests to prevent failure from update to waldo package
** Add gKRLS as an option for smoothing multiple (continuous) covariates. Chang and Goplerud (2024; https://doi.org/10.1017/pan.2023.27) provides more details.
Adjust predict.vglmer to allow for faster predictions on large datasets by not copying and filling in a large sparse matrix. Thank you to Michael Auslen for pointing out this issue.
Add the option for terms to predict to allow for predictions for each random effect separately
Address a bug where predictions with NA in new data.frame would fail for certain splines or for cases where newdata had a single row.
vglmer to not throw deprecation messages with Matrix 1.5. Thank you to Mikael Jagan for suggestions on how to adapt the code.