is_singular with lme models uses a lot of memory

[alejandrocorbellini-ac]

hi!

I just wanted to report a memory usage issue when calling get_variance with a linear mixed effects models. I'm new to R and mixed effects models in general, so sorry if something doesn't make sense.

I noticed that calling r2_nakagawa from the performance package ends up calling is_singular in helper_functions

In my code, this is the call chain:

r2_nakagawa -> insight::get_variance -> compute_variances -> is_singular

This ends up using a lot of memory when creating the diagonal matrix, although I don't know how to fix it. This also affects sjPlot::tab_model when passing an lme model, since internally they also seem to call r2_nakagawa.

Thanks!

[strengejacke]

Do you have a reproducible example? Is it in general, or with many parameters or large data sets only?