Skip to content

Real libraries - #79

Merged
saidctb merged 6 commits into
mainfrom
real_libraries
Sep 28, 2026
Merged

saidctb merged 6 commits into
mainfrom
real_libraries

Conversation

@saidctb

@saidctb saidctb commented Sep 28, 2026

Copy link
Copy Markdown
Contributor

No description provided.

saidctb and others added 6 commits September 28, 2026 13:56
Lead with what the example builds, a two-command quick start with a usage
snippet, and a key-files table; explain the single shared-library build
with a diagram before the build scripts; collect the PRIK and f2py
return-value differences in one table; and move environment setup,
versions, and platforms to the end. The quick start and snippet were run
against a fresh build (167 tests passed).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Lead with what the example builds (1,936 wrapped procedures, 127 validated
routines), a quick start, and a key-files table; explain the single
shared-library build with a diagram before the build scripts; collect the
PRIK, f2py, and SciPy calling differences in one table; and move setup,
versions, and platforms to the end. Match the one-line changelog style on
main for the BLAS and LAPACK guide entries.

The LAPACK build and tests were not run locally, per the repository's
instruction to leave LAPACK coverage to CI.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Lead with what the example builds and the 22 procedures by family, a quick
start that shows the minpack_module import, and a key-files table; explain
the single-command build before the script; add verified hybrd1 and lmdif1
examples with work-array sizes and info meanings; and correct the claim
that results are checked against SciPy, which the suite does not use.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SciPy's root(method="hybr") and least_squares(method="lm") are built on
MINPACK, so hybrd, hybrd1, hybrj, hybrj1, lmdif, lmdif1, lmder, and lmder1
can be cross-checked through SciPy's public API; the other 14 procedures
have no public SciPy counterpart. One parametrized test solves a nonlinear
root-finding system and a nonlinear curve fit with each solver, checks
PRIK's answer against the known solution first, then checks agreement with
SciPy. It skips without SciPy; CI already installs scipy==1.18.0 in the
MINPACK job. A perturbed-data control confirmed the agreement check fails
when the two differ.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Lead with the 31 procedures by family, a quick start with the fftpack
import, and a key-files table; explain the three source roles with a
diagram before the build script; and add a verified table of where
FFTPACK's conventions differ from numpy.fft (integer fftfreq indices, an
unnormalized ifft, and packed rfft output). The quick start and every page
snippet were run against a fresh build (33 tests passed).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Lead with the two namespaces and their public surface, a quick start, and a
key-files table; explain the three-source build with a diagram; add a table
of what PRIK maps from modern Fortran (abstract base, extensions, deferred
and inherited bindings, generic constructors, private members, constants);
and extend the API section with printed values, a derivative, a 2-D
surface, and status codes. The quick start and every snippet were run
against a fresh build (39 tests passed).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@saidctb
saidctb merged commit ac424d5 into main Sep 28, 2026
4 checks passed
@saidctb
saidctb deleted the real_libraries branch September 28, 2026 13:53
@codecov

codecov Bot commented Sep 28, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant