search

AdamCSmithCWS / JAGS-version-compare

Comparing JAGS 4.3 and 4.2 with the model used by the CWS and USGS to analyse population status and trends from the North American Breeding Bird Survey
0 stars 0 forks source link

is there a solution? #1

Closed DWLinden closed 4 years ago

DWLinden commented 4 years ago

Hi Adam,

Very happy to have found your reply on Martyn Plummers blog, and the testing you did here. It seems you may have found a solution to this problem?

I recently upgraded JAGS to 4.3 (after a complete R upgrade) and started having trouble fitting old model structures that previously worked. Models never got past initializing. I've reverted back to 4.2 as a short term solution. But honestly, I'm surprised more people have not had this issue and chirped about it since the release of 4.3 almost 3 years ago. I have 2 Windows laptops and 1 Linux (Ubuntu), all with the problem after upgrading.

Any insights would be appreciated!

Thanks, Dan

AdamCSmithCWS commented 4 years ago

Thanks @danlinden . I'm sorry I didn't notice your issue back in the spring. I've added a band-aid solution - don't load the glm module. It now works with JAGS 4.3. It's an imperfect solution thought, because the glm module is otherwise a very helpful package of samplers.

My long-term solution, is to transition to using Stan and Nimble (sigh).