In the Priors panel, a distribution's own internal parameters (e.g. LogNormal's M/S, Gamma's alpha/theta) used to show an "estimate" checkbox on the right, letting a user promote that parameter to a random variable with its own prior (a "hyper prior"). This is currently broken in several ways for the new beast.base.spec distribution/parameter classes.
- Checkbox missing. The "estimate" checkbox next to a distribution's own parameter (e.g. LogNormal's mean/stdev) no longer appears, even though it did previously.
- Hyper prior silently not created. Even in Expert Mode (where the checkbox is visible), checking it, confirming the "Hyper prior" warning dialog, does not add anything to the Priors panel. The parameter's estimate flag gets flipped to true, but no state/operator/prior wiring is created — leaving the model inconsistent.
- Wrong hyper-prior distribution domain. For a PositiveReal-domain parameter (e.g. LogNormal's S), the hyper prior that (should) get created must itself be domain-compatible (e.g. Gamma), not a Real-domain distribution — otherwise it shows up as a mismatched entry outside the compatible-domain dropdown list in its own Priors-panel row.
Reproduction
- Start BEAUti, import an alignment, go to Priors.
- Change the birth rate's prior from Gamma to LogNormal.
- Try to check "estimate" next to M or S. (Currently: no checkbox visible outside Expert Mode; in Expert Mode, checking it and confirming does nothing visible.)

In the Priors panel, a distribution's own internal parameters (e.g. LogNormal's M/S, Gamma's alpha/theta) used to show an "estimate" checkbox on the right, letting a user promote that parameter to a random variable with its own prior (a "hyper prior"). This is currently broken in several ways for the new beast.base.spec distribution/parameter classes.
Reproduction