AssertionError: Step-out procedure failed

[Casper Schmidt]

Dear HDDM users

i'm trying to run a the

model.sample(1000, burn=500)

with

hddm.__version__

Out[44]: '0.5.3'

kabuki.__version__

Out[46]: '0.5.3.dev'

on my data that looks like this;

where i get the error 

AssertionError: Step-out procedure failed

model.sample(1000, burn=500)

Traceback (most recent call last):

  File "<ipython-input-39-4df8dbe68bb3>", line 1, in <module>

    model.sample(1000, burn=500)

  File "C:\Users\imaging\Anaconda\lib\site-packages\kabuki\hierarchical.py", line 651, in sample

    self.mc.sample(*args, **kwargs)

  File "C:\Users\imaging\Anaconda\lib\site-packages\pymc\MCMC.py", line 276, in sample

    Sampler.sample(self, iter, length, verbose)

  File "C:\Users\imaging\Anaconda\lib\site-packages\pymc\Model.py", line 249, in sample

    self._loop()

  File "C:\Users\imaging\Anaconda\lib\site-packages\pymc\MCMC.py", line 312, in _loop

    step_method.step()

  File "C:\Users\imaging\Anaconda\lib\site-packages\kabuki\step_methods.py", line 383, in step

    assert iter < self.maxiter, "Step-out procedure failed"

AssertionError: Step-out procedure failed

[Casper Schmidt]

Hi everyone.

The script does not take empty values so it is sorted out. Anyway just replaced them with a mean of all measures in rt and there it was.

Best C 

[Kitti Ban]

Hi everyone, 

I get exactly the same error messages at random times during the iteration process and I don't have missing values in my data. The model.sample() command used to work for the my data set but it keeps giving the above error message more and more frequently, which doesn't allow the sampling process to finish.

I use the following versions of packages

-  HDDM version 0.9.7

- Numpy 1.23.2

- kabuki 0.6.5

Any advice would be really appreciated. 

Thank you,

Kitti

[Hitchcock, Peter]

Hi Kitti,

It looks like this comes from exceeding the max allowed iterations (https://github.com/hddm-devs/kabuki/blob/master/kabuki/step_methods.py) and I remember getting this error when there was something structurally wrong with my model/data including cases when it was more subtle than missing data. 

Are your RTs on the correct scale? (they should be in seconds; see the tutorial; https://hddm.readthedocs.io/en/latest/tutorial_basic_hddm.html)

If that's not it, you may want to try reproducing the tutorial models using the tutorial datasets and see if you get the same error. You seem to be suggesting the error may be related to your HDDM setup; this would be a good way to test that possibility against whether the error is due to the data and/or model you're trying to fit.

HTH,

Pete

--
You received this message because you are subscribed to the Google Groups "hddm-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hddm-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hddm-users/6aa73946-e6be-46e4-8694-df53335f6ecdn%40googlegroups.com.

--

Peter Hitchcock, Ph.D.

Postdoctoral Research Associate

LNCC Lab | Google Scholar

[Kitti Ban]

Hi Pete, 

Thank you so much for the fast reply - it was indeed the RTs that I forgot to change to seconds in this version of my analyses. Adjusting them instantly solved the problem (also dramatically improved run time and solved convergence issues). 

Thank you so much, you truly helped me out!

Best,

Kitti

To view this discussion on the web visit https://groups.google.com/d/msgid/hddm-users/CAONwh2WWwmRmxXTrEmPmHBsUWD48jXb%3DrOVNV-mKP7bkrEkUkg%40mail.gmail.com.