stefan-jansen/machine-learning-for-trading

Mamba env installation fails

hacky1610 opened this issue · 1 comments

hacky1610 commented

Describe the bug
The installation of the python packages by means of the ml4t.yml file is failing.
I am getting the error:

Encountered problems while solving:

  • package numpy-1.19.1-py36hbc911f0_0 requires mkl >=2019.4,<2021.0a0, but none of the providers can be installed
  • package libsqlite-3.40.0-h753d276_0 requires libzlib >=1.2.13,<1.3.0a0, but none of the providers can be installed

To Reproduce

  1. Create a DOCKERFILE inside of the installation folder.
    The content should be

    Start from a core stack version

    FROM jupyter/datascience-notebook:python-3.8

    Install in the default python3 environment

    COPY ./linux/ml4t.yml /var/tmp/ml4t.yml
    RUN conda create -n ml4t python=3.8
    RUN mamba env update -n ml4t -f /var/tmp/ml4t.yml
    RUN conda activate ml4t

  2. run docker image build --tag jupyter/base-notebook:my_version .

stefan-jansen commented

I'd recommend to just install the packages you'll actually use given the notebooks of interest. Installation without pinned versions allows access to the latest versions but may fail due to conflicts given python version etc.