Mark error

[Kerim]

"ERROR numerical convergence never reached. Exitcode ...."

What is cause of this error?
Thak you in advance,
Best wishes,

Kerim

[cooch]

"ERROR numerical convergence never reached. Exitcode ...."

What is cause of this error?
Thak you in advance,
Best wishes,

Kerim

Could be

1. strange data

2. incorrect model structure

3. a few problem parameters.

Try ruling out (3) to start - use the simulated annealing optimization (the 'alternate optimization' checkbox in the numerical run window).

[Tracy]

I thought I would contribute a suggestion, since it seems like this post has gotten a lot of hits. In my experience this error message can show up in MARK when running a highly parameterized model (e.g., Barker, multi-state) with sparse data. Here are two suggestions of things that worked for me after I got that error:

1) as Evan said, try using the simulated annealing optimization. With a complicated model, this could take a long time (mine is taking 20+ minutes to run).

2) Try to simplify your data so that it is not so parsed out between parameters (e.g., combine groups or cohorts, combine capture periods, etc. as long as it makes biological sense to do so, of course; to get the model to run, you may also have to simplify the question you are trying to answer).

3) If possible, try using a model with fewer parameters.