Skip to content

Releases: tanzercakir-commits/organon-mueller

v1.2.0 — The Lorentz face

Choose a tag to compare

@tanzercakir-commits tanzercakir-commits released this 17 Jul 22:43
7c1e719

The Lorentz face (FROZEN-7 complete)

Adds organon_mueller.lorentz — the Σ^μ / Z = α_μΣ^μ / Λ = ZZ* algebra, built from a collaborator's written work order. The Σ^μ basis coincides exactly with the engine's existing Z-matrix basis (the bridge theorem), so this is a new face of the same algebra rather than a bolt-on.

What's in it

  • core — Σ / Σ̄ bases, Z / Z̄ / guarded inverse / Λ, boost & rotation parametrizations with exact textbook anchors, metric-factor theorem ΛᵀgΛ = q q̄ g.
  • identities — the work order's five Task-1 identities (verified as guard-free theorem forms; conventions forced) and the five Task-2 Σ̄ identities (discovered; forced coefficient matrix C = gΛᵀg = Λ(Z̄); two bonus theorems).
  • terms — an 8-letter term language with a self-recovery gate: the machinery re-finds all ten known identities blindly and certifies them before it is trusted to hunt for new ones.
  • discovery — the Task-3 sweep: all 128 sandwiches over the alphabet decided exactly — 40 identities / 88 symbolic negative certificates — with a reproducible sweep table (reports/sweep-lorentz-01.json).

Verification

  • 399 automated tests, green on Linux ×3 + Windows CI.
  • Every identity is an exact symbolic proof over generic complex α. Novelty relative to the literature is deliberately not claimed — that judgement is reserved for domain experts.

Experimental research software, MIT-licensed. Nothing hosted or exposed.

pip install -U organon-mueller

organon-mueller v1.1.1

Choose a tag to compare

@tanzercakir-commits tanzercakir-commits released this 16 Jul 10:59
a258e8c

organon-mueller v1.1.1

Patch release — two fixes from independent field testing of v1.1.0 on
Windows.

Fixed

  • organon_mueller.__version__ now reports the real version. It had
    been left at "0.0.1" since before the first release, so bug reports
    quoting it would name the wrong version. It is now synced with the
    package metadata and guarded by a test that fails any future release
    where the two drift apart.
  • A typo in the first row of a matrix file now names the exact cell.
    Previously a data row containing one non-numeric cell (e.g.
    1,abc,3,4) could be mistaken for a header line and silently skipped,
    surfacing later as a confusing row-count error. The header rule is now
    strict — a line counts as a header only if none of its numeric
    positions contains a number — so the same typo anywhere in the file
    produces the same precise message: cell [row 1][col 2] is not a number: 'abc'. When a genuine header sits over a wrong number of
    rows, the error now says a header was skipped.

Notes

344 tests green on Linux (Python 3.10–3.12) and Windows. No API or
behaviour changes beyond the above.

Install / upgrade

pip install -U "organon-mueller[ui]"

organon-mueller v1.1.0

Choose a tag to compare

@tanzercakir-commits tanzercakir-commits released this 16 Jul 10:21
f8abd5b

organon-mueller v1.1.0

Feature release: load Mueller matrices from files, decompose whole
batches
, plus cross-platform (Windows) hardening driven by field
reports.

New

  • File / batch tab in the local interface (organon-ui): load a
    4×4 matrix from a .csv / .tsv / .txt file straight into the
    editor, or batch-decompose a whole file — one matrix per row, 16
    columns (m00…m33, row-major) or 17 columns (leading label such as
    wavelength + 16 entries). Per-row α₁ or a readable failure reason,
    plus a downloadable results CSV. Delimiters comma/semicolon/tab/
    whitespace, optional header line, # comments; limits 5 MB / 10,000
    matrices. Files are parsed strictly as numbers, locally — nothing is
    uploaded anywhere.
  • When several hypotheses are exact at once, the propose view now says
    so and explains why (composite types subsume fundamental ones; a
    decomposition is a decomposition, not the decomposition —
    selecting the physically meaningful one is a domain judgement).

Fixed (user field reports)

  • The LaTeX report tab crashed on Windows for titles containing Greek
    letters (α₁, β, λ — this project's daily notation): all text output is
    now written as UTF-8 explicitly.
  • The test suite failed on Windows via platform-code-page text reads:
    repo-wide explicit-encoding sweep, locked by a static tree-scanning
    guard and a dynamic EncodingWarning sweep; CI now includes a
    Windows cell alongside Linux (Python 3.10–3.12).
  • organon-ui without the [ui] extra used to die with a raw
    ModuleNotFoundError traceback (the command is installed regardless):
    it now prints one readable line with the fix
    (pip install "organon-mueller[ui]").

Notes

  • Strictness is deliberate: blank cells, trailing delimiters, and
    ragged rows are rejected with the exact row/column named — a batch
    loader must never silently shift columns and return plausible-looking
    wrong physics.
  • 342 tests green (Linux ×3 Python versions + Windows). Every result
    remains evidence-labelled; no novelty or physics claim is made.

Install / upgrade

pip install -U "organon-mueller[ui]"
organon-ui

organon-mueller v1.0.0

Choose a tag to compare

@tanzercakir-commits tanzercakir-commits released this 16 Jul 07:55
bc20856

organon-mueller v1.0.0

First public release.

A symbolic engine for the Stokes–Mueller algebra of polarization optics:
it decomposes depolarizing Mueller matrices into nondepolarizing
components, discovers and certifies algebraic identities of the underlying
state algebra, models coupled oriented dipoles, and reports every result
as evidence-labelled LaTeX — with each mathematical statement gated by a
written verification contract.

Highlights

  • Symmetry-conditioned decomposition — the six fundamental variants
    and three composite types, derived from rank-1 minor conditions (not
    transcribed from the published tables), verified symbolically against
    the literature to exact zero; plus a rank-3 three-term zone beyond the
    published cases (clearly labelled candidate — no novelty claim).
  • Identity discovery — equality saturation (egglog) over a
    covariance-vector term language; every candidate certified by
    engine-independent exact symbolic proof. The known-identity library (21
    entries, each tied to a published equation) doubles as a regression
    suite.
  • Coupled-dipole module — the three-term scattering decomposition as
    a theorem, mode hybridization, direction-general optical activity,
    Perrin reciprocity, and ensemble statistics, bridged back to the
    decomposition layer.
  • Local web interfacepip install "organon-mueller[ui]" then
    organon-ui: an editable 4×4 matrix grid with worked examples,
    decompose / propose-hypotheses / LaTeX-report tabs. Binds to 127.0.0.1
    only; nothing is uploaded anywhere.
  • Deterministic reporting — byte-identical, evidence-labelled LaTeX;
    a printed small number is a measured residual, never an exactness claim.

Verification

306 tests, green on the Python 3.10 / 3.11 / 3.12 CI matrix. No
mathematical statement reaches main without exact symbolic proof,
seeded numeric checks, known-identity regression, engine-independent
certification of every discovery candidate, and an independent
adversarial review (see docs/VERIFICATION.md).

Results are labelled by evidence class: verified (machine-checked
against published equations/tables) or candidate (beyond the
published results; no novelty or physics claim — that judgement is left
to human experts, per docs/novelty-protocol.md).

Install

pip install "organon-mueller[test,ui] @ git+https://github.com/tanzercakir-commits/organon-mueller@v1.0.0"
python -m pytest -q     # verify: full suite green
organon-ui              # local interface at http://127.0.0.1:7860

Requires Python ≥ 3.10 (discovery engine: ≥ 3.11). License: MIT.