https://github.com/pyamg/pyamg

Algebraic Multigrid Solvers in Python
https://github.com/pyamg/pyamg

amg conda krylov multigrid pyamg python solver sparse-matrix

Last synced: about 2 months ago
JSON representation

Algebraic Multigrid Solvers in Python

• Host: GitHub
• URL: https://github.com/pyamg/pyamg
• Owner: pyamg
• License: mit
• Created: 2013-06-22T14:24:24.000Z (about 11 years ago)
• Default Branch: main
• Last Pushed: 2024-04-12T12:14:29.000Z (3 months ago)
• Last Synced: 2024-04-12T19:55:58.530Z (3 months ago)
• Topics: amg, conda, krylov, multigrid, pyamg, python, solver, sparse-matrix
• Language: Python
• Homepage: http://pyamg.github.io/
• Size: 10.9 MB
• Stars: 534
• Watchers: 20
• Forks: 110
• Open Issues: 36
• Metadata Files:
  • Readme: README-DEV.md
  • Changelog: changelog.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE.txt
  • Citation: CITATION.bib

Lists

• awesome-scientific-computing - GitHub
• awesome-stars - pyamg/pyamg - Algebraic Multigrid Solvers in Python (Python)

README

        
Release Steps:

- suppose the current tag is 4.2.2 and the next is 4.2.3

- with no staged commits and a clean status...

- meld a summary of `git log $(git tag --sort version:refname | tail -n 1)..HEAD --oneline` (all commits since last tag) with whatever hash with `[4.2.3]` in changelog

- commit, push

- `mkvirtualenv releasetest`

- check that `pip install .` and `python -c "import pyamg; pyamg.test()"` pass (outside source directory)

- remove untracked files `git clean -xdf`

- the following can be done with a pre-release, `v4.2.3-alpha.6`, for testing.  It will not become the default on pypi and `gh release create` can be marked with `--prerelease` (below)

- commit, push

- `git tag -a v4.2.3 -m "version 4.2.3"`

- `git push`

- `git push --tags`

- then release the version on Github: `gh release create v4.2.3 --notes "see changelog.md"`

  - This will trigger the GHA `.github/workflows/wheels.yml` which builds wheels and a source distribution, and publishes to pypi

Testing notes:

- do not use seeds such as 0, 1, 42, 100

- for each needed seed, generate a "random" int

- `python3 -c "import numpy as np; np.random.seed(); seeds = np.random.randint(0, 2**32 - 1, 5); print(seeds)"`