HPC/Applications/python

Python variants

Interpreter-only releases from the Python Software Foundation

Contains

• The Python interpreter only, from the python.org sources.

Module nomenclature

python/pyMajor.pyMinor/compilerName-compMajor.compMinor/pyMajor.pyMinor.pyPatch-moduleBuild

Python interpreters and packages as bundled with the OS

Contains

• The Python interpreter that comes with the operating system is required for many system-provided and add-on packages.

Module nomenclature

python-osname/pyMajor.pyMinor/compilerName-compMajor.compMinor/pyMajor.pyMinor.pyPatch

Notes

Such a module normally does not need to be loaded. It is included here for visibility in the module system, and as a base for (usually older) modules which provide Python packages outside of Conda or Pip. Such packages are often tied to the version of the Python interpreter under which they were installed.

Conda-based Python suites

Contains

• Wide-ranging Python-based software suites, rather than specific HPC applications.

Module nomenclature

python-distributor/pyMajor.pyMinor/distMajor/distributor_defined_version[…]-moduleBuild

The suites typically contain:

• A Python interpreter.
• The conda/pip package management systems.
• Idle – Python Integrated DeveLopment Environment
• A rich set of Python-native packages, extensions, and add-on executables, as selected by the distributor, usually including
  • iPython and juPyter – Interactive computation system
  • Cython – Compiler for a superset of Python and C/C++
• Underlying libraries and supporting binaries, e.g.:
  • Tcl – Interpreter for the Tool Command Language
  • BLAS/LAPACK – Linear algebra libraries
  • MPI – Runtime environment for parallel applications using Message Passing Interface, providing the mpirun or mpiexec commands, plus related binaries and libraries.

Caveat: An MPI Runtime within a Python suite makes parallel computing easily accessible under Python, but this means a "mere python module" can clash with other MPI implementations on Carbon.

Avoid loading both a Python suite and an MPI implementation at the same time, especially in your dot-files. Doing so may lead to non-obvious failures with either the bundled parallel Python applications or (worse) unrelated compiled applications on Carbon.

Custom Conda environments

Contains

• Conda environments derived from one of the distributed Conda suites, with the package selection customized for Carbon.

Module nomenclature

python-env-distributor/pyMajor.pyMinor/distMajor/pyVersion[…]-moduleBuild

Possible extension for alternative package selections:

python-env-distributor-purpose/pyMajor.pyMinor/distMajor/pyVersion[…]-moduleBuild