Skip to content

Latest commit

 

History

History
78 lines (60 loc) · 3.74 KB

README.md

File metadata and controls

78 lines (60 loc) · 3.74 KB

MISU Computer-Lab Environments

  • Ezra Eisbrenner (August 2024)
  • Ezra Eisbrenner (December 2023)

Windows

  1. for FORTRAN install a Linux in a virtual machine and follow the instructions for Linux
    • but, you are welcome to give feedback on the use of WSL v2 (Windows Subsystem for Linux version 2). The steps for Linux below should work 1-to-1 on WSLv2.
  2. for Python you can
    • install the Anaconda suite, it is slow but works (you can find a list of recommended packages in the Python environment section below)
    • or follow this miniforge installation guide for Windows
    • or work within the virtual maschine too, then follow the instructions for Linux

MacOS

  1. install Homebrew (https://brew.sh) and run the brew commands, given here and in the other sections, in your terminal
  2. if you want python, then
    • brew install --cask mambaforge
    • see https://formulae.brew.sh/cask/mambaforge for more information
    • If in doubt, look for the conda/mamba executable with brew info mambaforge
  3. if you want FORTRAN, follow the FORTRAN instructions below.

Linux

  1. if you want python, then
  • if you want FORTRAN, install Homebrew (https://brew.sh) and follow the FORTRAN instructions below.

Python Environment

Create the MISU conda environment with the shell (bash) command below

mamba create         `# mamba is faster than conda` \
      -n misu_pyenv  `# name of new conda environment` \
      -c conda-forge `# channel from where to pull packages` \
      "python<3.10"  `# proplot does not support python > 3.10 as of December 2023` \
                     `# cartopy does not support python > 3.9 as of January 2024 with these dependencies` \
      dynaconf       `# advanced project settings` \
      jupyter        `# for notebooks and interactive vscode` \
      dask           `# for parallel workflows like xarray.open_mfdataset` \
      h5netcdf       `# another backend for NetCDF` \
      numpy          `# math` \
      scipy          `# math` \
      xarray         `# dimensional data` \
      pandas         `# tabular data` \
      f90nml         `# fortran namelist handling` \
      proplot        `# plotting with sensible defaults (restricts matplotlib<3.5 as of January 2024)` \
      matplotlib     `# plotting` \
      cartopy        `# map projections` \
      colormaps      `# provides many colormaps` \
      black          `# a formatter, non-negotiable`

activate the environment with

mamba activate misu_pyenv or conda activate misu_pyenv

FORTRAN dependencies

  1. brew install gfortran
  2. brew install netcdf
  3. brew install netcdf-fortran

You can find the path to the executables with

  1. brew info gfortran
    • find the path to the library, for example /home/linuxbrew/.linuxbrew/Cellar/gcc/13.2.0/bin
  2. brew info netcdf-fortran
    • find the path to the library, for example /usr/local/Cellar/netcdf-fortran/4.6.1

Unsupported

Other ways are possible, on all systems, from Linux, to MacOS, to Windows. Including official Linux repositories which may hold libnetcdff-dev or using WSL on Windows. However, none of those are supported by the teaching assistant.