Ready the wiki for the v3.3.4.4 release
Checked every page against what is actually public now, not against the devel
branch it was written for. The substantive fix is the Python module: the
released code builds `hiclassy` with a `HiClass` object (python/hiclassy.pyx,
Makefile target `hiclassy`) and provides no `classy` module at all, so every
`from classy import Class` on the wiki was about to document an import that
does not exist. Python-wrapper.md and Installation.md now use hiclassy
throughout, the PyPI section no longer claims the import name differs from the
local build, and the samplers note says plainly that anything importing
`classy` needs pointing at `hiclassy`.
Also corrected against the released tree:
- PyPI is an sdist, not a wheel: pip compiles the C code locally. The old text
promised an install "without compiling anything locally".
- `pip install hiclassy` no longer needs numpy and scipy spelled out; they are
declared from 3.3.4.3 on.
- the zip link pointed at archive/hi_class.zip, a branch that no longer exists
-> the v3.3.4.4 tag archive
- `mkdir -p output` dropped from the quick starts: output/ now ships
- version lines -> hi_class v3.3.4.4 (built on CLASS v3.3.4)
- Home.md logo now comes from the repository, not miguelzuma.github.io
- the cccg worked example links to commit 31382d65 in the public repository,
where it is now reachable, instead of naming a devel commit
PUBLISHING.md removed as the release-day step says; it is preserved in this
history and copied to ~/code/hi_class_notes/code_worklog/wiki_PUBLISHING.md.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QY2a7349dKv2NdiDi2aDJx
Write out the Horndeski and GLPV Lagrangians
The covariant page named G2...G5 in every model section without ever
saying what they multiply, so a reader could not check a model's
Lagrangian against a paper, or against their own. It now carries the
action and L2...L5 in the conventions of the code paper (1605.06102
eqs 2.1-2.5), the M*^2 definition, and the note that the G's already
carry 1/8*pi*G_N -- which is why every G4 on the page starts from 1/2.
The beyond-Horndeski section gains the two GLPV terms it was already
discussing, from Traykova, Bellini & Ferreira (1902.10687 eq 2.2), in
the same normalisation.
These are the first equations in the wiki. GitHub renders $$ blocks
with KaTeX in wikis as well as in Markdown; each block was checked
against KaTeX (and LaTeX) before committing, and PUBLISHING.md now says
to look at the rendered section once after the release force-push.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QY2a7349dKv2NdiDi2aDJx
Rename Models pages to colon-free filenames
GitHub's markdown renderer parses a colon in a relative link target as a
URI scheme and sanitizes the link away, so [x](Models:-Covariant-theories)
rendered as plain text. Pages renamed to Covariant-theories,
Parametrized-alphas, Expansion-history; all internal references updated.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Rebuild the wiki for the next release
Written against devel hi_class @ 90e58bcd (shooting fix, smg precision
defaults, BD attractor IC, cccg_exp): all 13 gravity_models/*.ini verified
to run out of the box before documenting them as such.
New pages: Home, Installation, Running-hi_class, Models (covariant /
parametrized-alphas / expansion-history), Output-files,
Stability-and-precision, Python-wrapper, Extending-hi_class,
Physics-and-papers, plus sidebar and footer. Every parameter name, option
string, default and formula was checked against the source; first written
documentation of gravity_submodel, wowa_w and the output_background_smg
column map. The 2017 CLASS-internals pages are dropped (the CLASS wiki is
linked instead).
PUBLISHING.md holds the release-day push instructions and is not a wiki
page.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>