What do the error messages "invalid floating point operation", "singular matrix", "diverging parameters" mean?

The model cannot be estimated due to sparse data, a too elaborate model or other reasons. A simpler model should be tried, by removing one or more covariates, covariate categories and/or time points. Often TRIM inidcates the year that causes the model estimation problem. Remove that year from the time-points-selection list, and run the model again. If you were using the time effect model, you have to switch to the linear trend model to be able to do this. Also set the option to automatically delete changepoints that cannot be estimated.