problematic mpi when creating new environment

[Jannes Nys]

On our HPC, I was using conda as a way to manage my environments and have a new python=3 available.

Hereby, i used only "conda create -n env_name python=3.9" to create new environments, and used pip for all the rest.

Using pip for all the rest is necessary to use all the compilers, mpi, etc.. available on (and optimized for) the cluster.

Since recently, something has changed and creating a new environment now also installs some mpi packages from the start.

How can I avoid this to happen? Or how can I make sure that I only install python, without conda installing its own MPI (or alternatively, how can I link to an available MPI package when creating a new conda environment)?

Thanks in advance, J

[宋玉成]

I do not think conda environment is a better solution in your case than `venv`.