search

hcp4715 / dockerHDDM

A docker image for running HDDM with parallel processing in jupyter notebook.
https://hub.docker.com/r/hcp4715/hddm
25 stars 6 forks source link

predefined 'thin' argument in HDDM_0.8_tutorial image #8

Closed XiaoyuZeng closed 10 months ago

XiaoyuZeng commented 1 year ago

Hi all. Many thanks for your time and effort in making dockerHDDM.

When using the HDDM_0.8_tutorial image (https://github.com/hcp4715/dockerHDDM_Guide/blob/master/tutorial/dockerHDDM_tutorial.ipynb), I noticed that the thinning doesn't work. Specifically, I've set thin=5 to improve the model convergence, but it turned out all of the samples were saved.

Then I noticed that the thin argument has been predefined in the model-specification function so that thin=1 in all cases: image

Feel free to close this issue if this has been corrected in the latest version of dockerHDDM.

Best, Xiaoyu

panwanke commented 1 year ago

This probelm has been fixed in new version of dockerHDDM. Now you could easily fit a model with m.sample(500, burn = 100, thin = 4).