No model meta

[athowes]

In the case that there is no model on a parameter we can either

1. Work around it by setting appropriate priors (Automate setting prior in case where there is no model #305)
2. Enforce there being a model by putting ~ 1 internally (Consider setting intercept on any distributional parameters without a model #266)

This is a meta issue for deciding between these options, or suggesting new ways to handle this.

[seabbs]

I strongly think we should do 2 for simplicity

[kgostic]

Agreed in meeting that we're doing 2.

[athowes]

• Also shift to brms style formula input.
• If given string then add on all other formula as constant
• Require formula on everything if given brms::bf()
• Have to also deal with the fixing stuff to a constant case

[seabbs]

sounds good

I think the default brms input is

bmr(mu ~ 1 +w wefwef) vs a string (see here: https://paulbuerkner.com/brms/)

[athowes]

bmr(mu ~ 1 +w wefwef)

What's this? Ok got it as in brm(formula, ...) where formula isn't a string. So it's not being given a string it's being given a formula (that isn't a bf?)

[seabbs]

yes exactly sorry should have been clearer

[athowes]

Closing as outdated / no longer relevant.