Issue with loading Dedalus in conda environment

[Yiteng Dang]

I successfully installed dedalus using the standard conda installation steps (bash install_conda.sh) and the installation seems to work except the known issue with the Hermite basis Issue 110 when I run python3 -m dedalus test. This issue disappears when I install the newest version from source using pip.

Then I added the conda environment dedalus to jupyter using ipython kernel install --user --name=dedalus. I can run Jupyter with this environment, but somehow it is unable to find dedalus even though the package is apparently installed (when I run conda list -n dedalus it shows dedalus version 2.2006). However, when I run import dedalus it gives the usual ModuleNotFoundError: No module named 'dedalus' error for not properly installed packages. Other packages in the conda environment seem to be fine though, as I can import numpy, scipy, etc.

Any idea on how to solve this?

[Daniel Michael Lecoanet]

Hi,

When I’ve come across this issue in the past, it has been because the version of jupyter I was using was not connected to the version of python I had used to install dedalus. I’m gotten around this by installing jupyter in the dedalus conda environment (i.e., running "pip install jupyter” when in the dedalus conda environment). It sounds like you’re doing things differently, so you might be having an orthogonal problem.

Daniel

--
You received this message because you are subscribed to the Google Groups "Dedalus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dedalus-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dedalus-users/e12e82dc-bc13-46e8-a125-5954af871f2cn%40googlegroups.com.

[Y. Dang]

Solved by the suggestion above (installing Jupyter in dedalus conda env). Thanks!

For reference, the issue seemed to be with Jupyter only and not with conda or dedalus. Here is a great post on setting up jupyter kernels using conda environments:

https://stackoverflow.com/questions/58068818/how-to-use-jupyter-notebooks-in-a-conda-environment

Unfortunately, troubleshooting through all steps still didn't solve the issue for me, but it seems that some settings in the kernel configuration probably got messed up (these can be manually edited by changing the .json file for the Jupyter kernel).

Thus, the easier solution is indeed to just install Juypter in the dedalus environment.