SAS9.4 Proc Mixed hangs when requesting solution for random effects

[DaiW]

I am running SAS9.4TS1M4 on Windows 7. I am able to run following proc mixed code without any problem.  

proc mixed data=InDat; 
    class ID;
    model Y = T T*T T*T*T T*T*T*T T*T*T*T*T T*T*T*T*T*T/ddfm=kr solution outpm=predictm;
    random int T/subject=ID type=un;
run;

However, when I add in the option to get estimates for the random effects (see code below), proc mixed hangs without giving any results. The last log message before it hangs is "NOTE: Convergence criteria met". 

proc mixed data=InDat; 
	class ID;
	model Y = T T*T T*T*T T*T*T*T T*T*T*T*T T*T*T*T*T*T/ddfm=kr solution outpm=predictm;
	random int T/subject=ID type=un solution;
run;

My company recently switched from SAS9.2 to SAS9.4. I was able to run the same code and get the random effect estimates using SAS9.2 (installed on the same computer) without any problem.

Any idea what the problem is? Thanks in advance.

[Reeza]

Questions about things working in one version and not another should be directed to SAS tech support, since they can talk to the developers if necessary. Do you know what SAS/STAT versions you were using - PROC MIXED is under SAS/STAT which has different versioning that SAS Base. The current version is 14.3.

I'm not seeing anything wrong syntax wise, but I'm not that familiar with PROC MIXED.

http://documentation.sas.com/?docsetId=statug&docsetTarget=statug_mixed_syntax13.htm&docsetVersion=1...

[DaiW]

Thanks for your reply. I'm using SAS/STAT 14.2. 

[pau13rown]

see if the results are sent to an ods ie before the proc mixed statement put: "ods output solutionr=solutionr;" (see here: https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_mixed_sect0...)

since the code seems tp run but just not throw the results(?)

[DaiW]

Thanks for your reply. 

No. No result was sent to ods with ods output statement added.

[pau13rown]

i know you shouldnt need to do this, if it ran on an earlier version, but try a simpler cov structure rather than type=UN, UN will demand the estimation of many parameters, something simpler eg CS might run, and then you can add complexity until it breaks again

[DaiW]

I only have 2 random effects here. Using "type=un" would not add a lot of parameters. I just tried to use "type=vc". The SAS system hangs the same way.

[pau13rown]

this may sound stupid but have you closed the session and restarted?: https://communities.sas.com/t5/General-SAS-Programming/SAS-exhaustion/m-p/469235#M57636

[DaiW]

Yes. Each time it hangs, I am not able to interrupt proc mixed execution from inside of the SAS session.  I have to force the SAS session to terminate and restart it.

[pau13rown]

this sounds familiar to me, a "quit" statement shouldnt make any difference but worth a try