non-zero status code 1

3 replies [Last post]
userzht's picture
Offline
Joined: 02/19/2011

Hi, all

I was going to make a multivariate ACE model. But I got this warning when fit saturated model.

In model 'Phenotypic_Cholesky' NPSOL returned a non-zero status code 1. The final iterate satisfies the optimality conditions to the accuracy requested, but the sequence of iterates has not yet converged. NPSOL was terminated because no further improvement could be made in the merit function (Mx status GREEN).

What should I do with error like this?

Thanks in advance!
Ting

tbates's picture
Online
Joined: 07/31/2009
what to do

my workflow is as follows

1. run it again: mxRun returns a runnable model, so, if your model is called, say, "m1", just say
m1 = mxRun(m1). If the code green goes away, it just needed another hack at getting to a good solution.

2. If it reliably code-greens, check the stability of the model by running a few times (preferably with different start values), and seeing if the parameter estimates move around or not.

Mostly code green is not a problem. Of course, completely wrong models can run just fine: so ensure that you are not doing something silly by checking what is hooked up to what. Can post your script here if you wish.

Also, the wiki (linked from the top of every page here) is helpful

userzht's picture
Offline
Joined: 02/19/2011
further problem

Thanks for your reply. I rerun the model and the code green disappeared.

Here is another problem. I made saturated model and three multivariate models, but all the three models were significantly different from the saturated one, and the diffdf between saturated and Common Pathway models did not equal to their difference of ep.

tableFitStatistics(Phenotypic_Cholesky_Fit, Nested)
Name ep -2LL df AIC diffLL diffdf p
Model 1 : Phenotypic_Cholesky 28 25935.8 3994 17947.8 - - -
Model 2 : ACE_Cholesky 11 36272.17 4011 28250.17 10336.37 17 0
Model 3 : ACE_Common 13 36273.33 4010 28253.33 10337.53 16 0
Model 4 : ACE_Independent 14 36272.17 4008 28256.17 10336.37 14 0

I am confused. There must be something wrong. Could you please instruct me?

AttachmentSize
mv.R 9.11 KB
neale's picture
Offline
Joined: 07/31/2009
Nothing wrong.

Equality constraints specified with mxConstraint() automatically give you df credit, because they effectively reduce the number of free parameters. In the present case, there are mxConstraints in the Common Pathway model but not the Cholesky, so the difference in df and the difference in ep are not the same.