Live-dead multistate - problem fixing r=1 for known fate

[polmvand]

Dear all,

I am trying to model known fate data from multiple states. I realize that in principle I do not need MARK for known fate analysis, but I nevertheless do so to benefit from the (multistate) framework in MARK.
We have determined in many years the state of an individual during the breeding season, and we know whether individuals survive each winter in between or not.
Thus I am using 'Multi-strata live-dead Enc.' model. My LD-data looks e.g. like this M00000B0P000B0B100, where M, B, P are the 3 states. Furthermore, p<1 as we only know the state of an individual when we capture it during a certain period of the breeding season.

As this is known fate data I want to fix r to 1, but MARK then reports:
"ERROR - Numerical convergence never reached at G=0.2295007E+08"
I have checked (and double-checked) my input file and that seems all fine. Interestingly, when I fix r=0.999 the model converges fine, and the S- and psi-parameter estimates are similar as in previous analyses.

Does this mean that the ML function for this type of models cannot handle values of r=1 (e.g. because numbers go to infinity in the ML function)?

Martijn