We have an issue where the convergence plot shows convergence after a short # of iterations, but the model is running to the max iterations anyway. I have checked all instances in macros and Ella of preventing convergence artificially, and can’t find the reason. Any ideas?
Liz, two thoughts. (1) are all the controllers converged? I think it might be possible for the model to stop iterating, thinking it's converged, even though a controller hasn't hit its target. Take a look at the controller plots first, then check TOL and FSTOP specifications if something doesn't look right. (2) Look for instances of "kssignalnotcoverged" in KS EbsScripts (in Comp 93s or macros in "Expert" mode with KS). That function forces iterations to continue. Use the method in this post to find them: https://davidkolman.substack.com/p/automate-ebsscript-edits-in-a-suite
Tbh I have never used the convergence plot. I would be glad if you find time to explain it to me and the other readers. Thanks in advance
We have an issue where the convergence plot shows convergence after a short # of iterations, but the model is running to the max iterations anyway. I have checked all instances in macros and Ella of preventing convergence artificially, and can’t find the reason. Any ideas?
Liz, two thoughts. (1) are all the controllers converged? I think it might be possible for the model to stop iterating, thinking it's converged, even though a controller hasn't hit its target. Take a look at the controller plots first, then check TOL and FSTOP specifications if something doesn't look right. (2) Look for instances of "kssignalnotcoverged" in KS EbsScripts (in Comp 93s or macros in "Expert" mode with KS). That function forces iterations to continue. Use the method in this post to find them: https://davidkolman.substack.com/p/automate-ebsscript-edits-in-a-suite
Mike, I give some overview information in this post: https://davidkolman.substack.com/p/zero-is-a-terrible-starting-guess