Stan recently switched to weighting acceptance rates by their probability. An explanation is [here](https://discourse.mc-stan.org/t/request-for-volunteers-to-test-adaptation-tweak/9532/29). The change process was controversial, but the theory is appealing — DynamicHMC should have this as an option and test it for existing models.
Stan recently switched to weighting acceptance rates by their probability. An explanation is here.
The change process was controversial, but the theory is appealing — DynamicHMC should have this as an option and test it for existing models.