Trouble Installing

[Kane York]

Hey all,

It seems I am not the first with this issue. But, I have tried following the github instructions to install HDDM, but to no avail. It seems there is an on-going issue with getting 2.3.8 pymc to install as well based on what I am seeing from other threads. cython and kabuki install just fine. But, 2.3.8 pymc and hddm are the main things that are returning errors. I tried downgrading to 3.7 from the latest version of Python, but still no dice. I would paste the errors into this thread, but there are many, and I fear that it would just took too much space as a meaningless wall of text. 

Has anyone been able to do a clean install as of recent?

[Michael J Frank]

Alex (cc'd) may be able to help but we are quite close to an alpha release of a new completely overhauled toolbox, so if you are just getting started I would recommend waiting to use that.

MJF

Michael J Frank, PhD | Edgar L. Marston Professor

Director, Carney Center for Computational Brain Science 

Laboratory of Neural Computation and Cognition  

Brown University
website 

--
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/8dc6107a-154e-41bf-888d-4d197a93b814n%40googlegroups.com.

[alekhya...@gmail.com]

Dear Kane

This set of commands did the trick for me with python=3.6 in conda

conda activate yourenv conda install conda-build conda install pymc=2.3.8 -c conda-forge conda install pandas patsy export MACOSX_DEPLOYMENT_TARGET=10.11 pip install cython pip install hddm
Alekhya

[Sean Devine]

Alekhya, 

Your solution works on my machine (MacOS Monterey, 12.5.1, M1 chip), but installs hddm v. 0.8.0. Maybe this what Kane wanted, but for versions of python > 3.6, has anyone managed to install locally to Mac or Linux? Much like Kane, I haven't had much success following the steps here and would need to leverage the new angle model for a project. Happy to post error messages if they would be of help. 

thanks for any help, 

Sean  

[Kianté Fernandez]

Here is the solution that has been worked out in previous threads:

#Create an enviroment via the terminal: conda create -vv -n LANHDDM37 python=3.7 jupyter conda install -c anaconda ipykernel conda activate LANHDDM37 python -m ipykernel install --user --name=LANHDDM37 #Install HDDM inside the env: conda activate LANHDDM37 conda install conda-build conda install pymc=2.3.8 -c conda-forge conda install pandas patsy export MACOSX_DEPLOYMENT_TARGET=12.5.1 pip install git+https://github.com/hddm-devs/kabuki pip install git+https://github.com/hddm-devs/hddm conda install jupyter pip install torch torchvision torchaudio

Hope that helps.

Kia

_____

Kianté Fernandez, M.S. M.A.

kiantefernandez.com

Doctoral Student, Decision Psychology Program

Department of Psychology 

[Kane York]

Hey all,

I think I have tried all the solutions so far, varying the version of Python between 3.6, 3.7, and 3.8. Would an older version of HDDM work? Also, if it helps I am on PC (Windows 11).

[Alexander Fengler]

You might get some mileage out of this: https://hub.docker.com/r/hcp4715/hddm.

However we suggest to check out the HSSM toolbox, developed to be HDDMs successor, and currently in alpha release.

Install (and many other things) should be much easier with HSSM.

Best,

Alex

[Zeynep Özkan]

Hello all, 

As Kane, I guess I tried every possible solution in the internet to install hddm package. Nothing worked. I would be appreciate any kind of help in this point. 

Best wishes,

Zeynep

23 Ekim 2023 Pazartesi tarihinde saat 05:03:33 UTC+2 itibarıyla Alexander Fengler şunları yazdı:

[Alexander Fengler]

I will try to look into this towards the end of the week, and  try to fix some dependency versions.

That might help. The basic problem of HDDM, which is dependency on the deprecated PyMC 2, 

won't be resolved however, and this is  the main reason why we have to slowly let it die :/.

However in the context of other projects I realize that e.g. new versions of numpy wrecked havoc across package ecosystems,

so in general it probably makes sense to aggressively constrain package versions once now, and allow HDDM to persist in 

a semi-stable state.

In principle, we suggest to switch over to HSSM. 

HSSM is build on a backend that will make it last far into the future.

There are a few growing pains initially, but that package is maturing quickly, 

including plotting support, support for modern model comparison, within AND between subject regressions etc. etc.

Best,

Alex