Error message: "Numerical convergence never reached"

[Enock Kale]

Hallo there,

I am trying to run Jolly-Seber models and get this error message "numerical convergence never reached". My paper is due this month and I need to get through this quickly. I will really appreciate any advice on this.

Thank you in advance.

Enock Kale.

[bacollier]

Hallo there,

I am trying to run Jolly-Seber models and get this error message "numerical convergence never reached". My paper is due this month and I need to get through this quickly. I will really appreciate any advice on this.

Thank you in advance.

Enock Kale.

Hello Enock,

I think you probably need to read (or at least review if you have already read) Faq 6 (viewtopic.php?f=39&t=9). Then, you might try searching the forum as you might get some idea's on how others have addressed similar issues (e.g., different link functions, setting initial parameter values, etc). Then, you should look at better explaining to those of us who might be able to help you what your data structure looks like and what model structure you are trying to run (again, see Faq 6) and then someone might be able to help you out.

Bret

[jlaake]

Your post is not specific enough as you haven't told us which specific Jolly-Seber model you are using. Is it the one listed as Jolly-Seber or one of the Pradel or POPAN models? If it is the one listed as Jolly-Seber, it is not a particularly good parameterization and routinel has problems with numerical convergence. Try using POPAN or one of the Pradel models to analyze your data.

--jeff

[lucette]

Hi, I have the same problem,
I want to run models on multistates data and after a moment of running it writes the error: " numerical convergence never reached".
Have you find the solution?

Thanks for your help

[Fish_Boy]

Sometimes it is as simple as changing the link function. However, as Jeff suggests other JS options often provide better parameterization routines.