1 reply to this topic

[richadua]

Where can I possibly start if my model continues to give a Retcode = -1? In other words, what are some of the most common reasons as to why a model will not converge? 

 

I tried a 2-comp and then reduced it to 1-comp. The AIC and BIC improve with a 1-comp. However, no matter what residual error model I use, my graphs for CWRES vs IVAR and CWRES vs IPRED look off by a lot. 

 

If model is not converging, how am I getting theta's and SDs and all diagnostic plots?

 

Thanks,

Richa Dua 

[smouksassi1]

A minimization algorithm can always convert to a local minimum or less than optimal solution.

So you still get parameter estimates  and residuals and other outputs based ont this solution which is not optimal.

 

 

The user manual states:

Negative return codes generally indicate some kind of convergence failure and

often a more definitive reason for the convergence failure can be found in the

core status file in the Text Output section of the Results tab.

 

It will be hard to help without further information about the data and the method used to solve the problem.

Samer