Help using Base SAS procedures

NLMixed: Upper Limit on number of Random Effects

Reply
N/A
Posts: 0

NLMixed: Upper Limit on number of Random Effects

Can anyone tell me what the upper limit is for the number of random effects in NLMIXED? I get a model to run with 2 REs. I see an example in the documentation with 3. I could use 4. The software doesn't object to a submission of 4 but wants to take forever and I've not yet gotten a set of results.

Thanks.

T

Message was edited by: Turgidson Message was edited by: Turgidson
Trusted Advisor
Posts: 2,115

Re: NLMixed: Upper Limit on number of Random Effects

Posted in reply to deleted_user
The technical upper limit is much higher than the practical upper limit.

Check the documentation for computational problems and computational resources. A couple of things are likely to be going on.
1) you have run out of real memory and the system is swapping
2) by adding additional random effects, the model is no longer as well specified and the algorithm is having a hard time converging.

1) invites a hardware solution.
2) requires skull time and data investigation.
Ask a Question
Discussion stats
  • 1 reply
  • 95 views
  • 0 likes
  • 2 in conversation