Ben Bolker
[This issue](https://github.com/glmmTMB/glmmTMB/issues/803) describes speedups available by enabling a new AD library inside of TMB. According to [this commit](c66c8cc8200076a840ef2) all we need to do is Depend: on TMB >= 1.8.0 (DESCRIPTION...
The only (?) purpose of `fixed_pars` at present (at least within the `run_sim_breaks` framework) is to specify fixed values of the dispersion parameters. Is there a cleaner way to do...
Need to be fancy if we don't to break automatic redirect for the main repo
There's some kind of weird interaction between the `testthat` and the negative-state warnings. With the current version of the `bmb_devel` branch, if I ```r source("tests/testthat/test-methods.R", echo=TRUE) ``` in a clean...
if `opt_pars$time_params` is named, substitution into `params_timevar$Relative_value` should be done by name, not position, following the convention that the names will be `[Symbol].[Date]` (which should be unique). Also, we should...
We should be more consistent/come up with a formal system for version numbering. **Semantic versioning** (add link) is a nice idea but I have found in the past that I...
it would be nice to have `[ci skip]` in commit message exit with an "OK" status from the whole job. If we use a zero error code for this step...
Lots to say here. ## mechanistic model Incorporate testing, including lags and allowing use of negative-test data, in compartmental model (or renewal-equation model??) in order to account for changes in...
This will be a catch-all for minor cosmetic/methods issues. It can be updated by ~crossing off~ items in the list, or by commenting. * make sure state vars are factorized...
Go through parameters (especially `stanford_estimates.csv` and `midas_estimates.csv` ; see `notes/midas_estimates.R`) and try to assemble some principled thoughts about how to construct priors for model parameters ...