Problem 1: aborting without error message or results
While running standard robust-design models with MCMC, the MCMC process will abort with 0 kb files. I've watched it part of the way, and a number of iterations (at least 1000) are run on chain 1 before the window just closes. Some models I can get to complete, but others do not. My attempts are to model survival group by year (this worked) or full time (this did not work) using a hyperdistribution.
Problem 2: odd .csv file results for detection parameters.
For p, the mean values match what I am getting without MCMC, but the median and other percentiles are formatted as
, for example. The logit link parameters are not bonkers, but the real estimates are. For example, here's a 95% CI for p session 89, time 3:' 0.6577915E-37
-0.2477196 0.0419403
Here's what is reported for the real estimate:
0.6486637E-44 0.5652921E-34
Also note that the logit link parameters are reported in a different order compared to the real ones. Not sure if this matters at all but thought I'd mention it. Here's parameters 119 through 121:
Logit:
119 p:session73.9:time3
120 p:session75.9:time3
121 p:session78.9:time3
Real:
119 p g1 s86.9 t1
120 p g1 s86.9 t2
121 p g1 s86.9 t3