-
Notifications
You must be signed in to change notification settings - Fork 17
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
mincLmer will converge when specifying summary_type = "fixef", yet will not when summary_type = "ranef" or "both" #298
Comments
Based on testing, the fact that this is failing even with |
With Chris no longer actively working on this code I can't promise a quick fix. I'll look into it when I have a bit of time (unless there's a patch coming from you guys). Quick workaround is to use mincApply with your own wrapped lmer function. Will be slower than mincLmer, but should otherwise give the same answer. |
Looking again at the code, the summary function does appear to be wrapped:
:(
That makes certain assumptions about R ability. Will investigate. |
A fortuitous discovery by another student doing the same thing. Her code runs with "|" as the ranef slope (correlated slopes) and fails with the same error as the initial report if run with "||" (uncorrelated slopes). Update: It works with the original dataset with "|"! |
Hello,
I'm trying to use the random slopes of a mixed-effects model as an input for another statistical test, however I'm having issues with getting the mincLmer to run when specifying the summary_type to be "ranef". The models will converge, with a few non-converging voxels, and give outputs when the the summary_type is set to "fixef" (i.e., model1), however, the model will not run and will be sent into a fail state when the summary_type is set to "ranef" (i.e., model2). Photos of failure and from logs are below.
In trying to work through the issue with @gdevenyi, we've checked the mask, we've checked for any low level background in the average and they are both fine. We then used mincWriteVolume to save the "converged" column from model1. The model ran with few non-converging voxels, so we then reran model 1 with an optimizer "Nelder_Mead". We then had fewer points of convergence failure. We tried to apply this method to the "ranef" model (model2), including the safely = True, however the model still will still not run.
Would anyone have any insights into possible solutions to get a model with summary_type = "ranef" to run?
Please let me know if more information is needed.
Tristin
The text was updated successfully, but these errors were encountered: