star-magic-program 0.402.0__tar.gz → 0.403.0__tar.gz

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/CHANGELOG.md

@@ -7,6 +7,63 @@ Versioning: [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ---
 
+## v0.403.0 — 2026-08-27 — THE PRODUCT SHIP
+
+The independent evaluation's finish sequence, **executed** (`uqff_downhole_simulator` v1.41.0–v1.48.0):
+
+**Finish sequence (1–5) = FINISHED OFFLINE PRODUCT, by the evaluation's own criterion:**
+
+- **v1.41** Modbus unified — registry entry IS the real client; disciplined
+  minimal refusal; static source discloses the runtime upgrade.
+- **v1.42** The well assembler — WellAssembly per site family; strict
+  measured-coverage lookups; overburden from the site's own density;
+  engine bridge with Rule-7 method labels.
+- **v1.43** Measured wells as engine defaults — `run --well ktb_hb` on the
+  1994 log; Volve F-12 measured downhole P as the reconciler live leg
+  (drawdown → UNEXPLAINED_TREND: not drift).
+- **v1.44** The operator surface — headless gate-tested OperatorSession +
+  Qt6 view; rating check BLOCKS the run on measured conditions; permanent
+  citations pane (suppression labeled DERIVED_HYBRID in the UI).
+- **v1.45** The acceptance suite — in-package product gate (`accept`),
+  grown to 55 checks; independence from the physics corpus enforced by
+  construction (subprocess + static verification).
+
+**Field tier:**
+
+- **v1.46** (6a) Gamma/lithology from the catalogue's own curves — six
+  gamma-bearing entries under unit discipline; KTB pilot 80 intervals;
+  Volve sand/shale split; every number labeled.
+- **v1.47** (6b) Mixed toolstrings — honest per-station legs; aggregates
+  over twin stations only; rating check inside the engine constructor.
+- **v1.48** (8) Bench-test protocol — BENCH_TEST_PROTOCOL.md in-package;
+  falsifiable R = 1.0324 both directions; four-verdict analysis
+  self-verified as SIMULATION_SELF_TEST.
+
+Mid-arc audit (Daniel): twin track verified by measurement at every layer;
+three small gaps found and fixed in-arc (gui extra + matplotlib,
+LAS-ingest button, twin-leg drift tab). Remaining: step 7 only (one live
+site path — needs a real host + cited register map).
+
+Gate 5,904 → 5,924, 0 failures. Acceptance 55/55. 23-file pass.
+
+**REMANUFACTURED 2026-08-27 (same version; tag preserved, PyPI never
+received the first push):** CI failed on all runners at the FINISH-SEQ 2
+pin — `AttributeError: numpy has no attribute 'trapz'` (numpy ≥ 2.0
+removed the alias; CI resolves newer numpy than the authoring sandbox's
+proxy-pinned 2.2.6, which still carried it — every local repro was green
+for exactly that reason). Fix: the assembler's overburden integral now
+prefers `np.trapezoid` with a 1.x fallback; a portability-guard pin bans
+direct `np.trapz` from the assembler and canonizes the standing rule
+(local dependency versions are NOT the ship's — removed-alias sweeps are
+ship prep). Also found in the same audit: the tracked
+`star_magic_program-0.402.0/` sdist staging tree (129 stale files left
+by a failed in-repo build on the mounted FS, committed by `add -A` at
+v0.402.0) — `.gitignore` now blocks `star_magic_program-*/`; Daniel
+removes the tree with `git rm -r star_magic_program-0.402.0` on Windows.
+Acceptance A2's stochastic band widened; the gate's acceptance-subprocess
+pin now embeds subprocess stderr in its failure message. Gate 5,925/0.
+
+---
 ## v0.402.0 — 2026-08-27 — THE POLE-TO-POLE SHIP
 
 The downhole catalogue grows **TWENTY → THIRTY entries** across ten

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/CITATION.cff

@@ -3,7 +3,7 @@ message: "If you use Star-Magic-Program (UQFF systematic rebuild) in your resear
 type: software
 title: "Star-Magic-Program: UQFF Systematic Rebuild"
 abbreviation: "UQFF"
-version: "0.402.0"
+version: "0.403.0"
 date-released: "2026-08-27"
 authors:
   - family-names: "Murphy"
@@ -53,7 +53,7 @@ preferred-citation:
     - family-names: "Murphy"
       given-names: "Daniel T."
   year: 2026
-  version: "0.402.0"
+  version: "0.403.0"
   url: "https://github.com/Daniel8Murphy0007/Star-Magic-Program"
   notes: >-
     Cite both the software and the relevant whitepapers by PAPER_N

{star_magic_program-0.402.0/star_magic_program.egg-info → star_magic_program-0.403.0}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: star-magic-program
-Version: 0.402.0
-Summary: Star-Magic UQFF calculator v0.402.0: THE POLE-TO-POLE SHIP. 2,253 dispatches (2,308 keys), gate 5,906/0. Downhole catalogue 20->30 entries (v1.31-v1.40): live heat-flow closure q=k*dT/dz; 504B four-dataset family with cross-entry impedance join; Antarctica, Nankai, SW Indian Ridge, Mid-Atlantic, Chicxulub and central Arctic (87.9N) join; crust ladder complete from seafloor mud to mantle peridotite; ACEX age model = first time axis. 18 regions / 19 kinds. Next: PAPER_2258.
+Version: 0.403.0
+Summary: Star-Magic UQFF calculator v0.403.0: THE PRODUCT SHIP. 2,253 dispatches (2,308 keys), gate 5,924/0. The downhole simulator completes the adopted independent-evaluation plan (v1.41-v1.48): finish sequence 1-5 - modbus unified, well assembler, measured wells as engine defaults, operator surface, in-package 55-check acceptance suite = FINISHED OFFLINE PRODUCT; field tier 6a gamma/lithology from catalogue curves, 6b mixed toolstrings + in-engine rating block, 8 bench-test protocol. Next: PAPER_2258.
 Author-email: "Daniel T. Murphy" <daniel.murphy00@enrgyone.com>
 License: AGPL-3.0-or-later OR LicenseRef-StarMagic-Commercial
 Project-URL: Homepage, https://github.com/Daniel8Murphy0007/Star-Magic-Program
@@ -34,21 +34,22 @@ Provides-Extra: modbus
 Requires-Dist: pymodbus>=3.15; extra == "modbus"
 Provides-Extra: gui
 Requires-Dist: PyQt6; extra == "gui"
+Requires-Dist: matplotlib; extra == "gui"
 Dynamic: license-file
 
 # Star-Magic-Program
 
-[![PyPI version](https://img.shields.io/pypi/v/star-magic-program.svg?cacheBust=0.402.0)](https://pypi.org/project/star-magic-program/)
-[![Python versions](https://img.shields.io/pypi/pyversions/star-magic-program.svg?cacheBust=0.402.0)](https://pypi.org/project/star-magic-program/)
+[![PyPI version](https://img.shields.io/pypi/v/star-magic-program.svg?cacheBust=0.403.0)](https://pypi.org/project/star-magic-program/)
+[![Python versions](https://img.shields.io/pypi/pyversions/star-magic-program.svg?cacheBust=0.403.0)](https://pypi.org/project/star-magic-program/)
 [![Documentation Status](https://readthedocs.org/projects/star-magic-program/badge/?version=latest)](https://star-magic-program.readthedocs.io/en/latest/?badge=latest)
 [![License: AGPL-3.0 + Commercial](https://img.shields.io/badge/License-AGPL--3.0%20%2B%20Commercial-blue.svg)](LICENSE)
-[![Fidelity gate](https://img.shields.io/badge/fidelity_gate-5906%2F0-brightgreen)](uqff_fidelity_tests.py)
+[![Fidelity gate](https://img.shields.io/badge/fidelity_gate-5925%2F0-brightgreen)](uqff_fidelity_tests.py)
 [![Public surfaces](https://img.shields.io/badge/public_surfaces-2308-blue)](uqff_calculator.py)
 [![Whitepapers](https://img.shields.io/badge/whitepapers-2292-orange)](whitepapers/)
 
-**UQFF systematic rebuild — v0.402.0 complete-compile campaign live**
+**UQFF systematic rebuild — v0.403.0 complete-compile campaign live**
 
-**This release (v0.402.0): THE POLE-TO-POLE SHIP — THE CATALOGUE GROWS FROM TWENTY TO THIRTY ENTRIES, AND THE SPAN IS MEASURED.** Ten catalogue packages (`uqff_downhole_simulator` v1.31.0–v1.40.0), each a complete public dataset transcribed verbatim with a provenance sidecar, each verified live by the gate. The arc opens with **the heat-flow closure** (entry 21: thermal conductivity of ODP 1027B pairs with entry 20's CORK temperatures — q = k·dT/dz ≈ 130 mW/m² computed from two catalogued datasets of one site, not quoted) and closes at the North Pole. **DSDP/ODP Hole 504B becomes a four-dataset family** — 1979 sound velocity, 1979 physical properties (the WBD identity closing on all 61 rows), 1986 sheeted-dike elastic moduli (four identities re-derived on every run), 1991 borehole fluids — and the two 1979 tables archive the same core suite, so the gate joins them sample-by-sample and computes a **live acoustic-impedance profile** (Z = ρ·Vp = 13.9–18.5 ×10⁶ kg/m²s) from independently archived datasets. **Six regions join**: Antarctica (entry 23, Prydz Bay — whose replicate audit caught a dropped-digit typo IN the source archive), the Nankai Trough (26, first subduction zone and first Chikyu data, with Ikari et al.'s slope-stability conclusion re-derived: FS 2.38–15.31, all stations stable), the Southwest Indian Ridge (27, Atlantis Bank gabbros — completing the **oceanic-crust ladder**: sediments → pillow basalt → sheeted dikes → layer-3 gabbro), the Mid-Atlantic Ridge (28, serpentinized **mantle peridotite** — the ladder goes below the crust, the five-identity lock closing at 0.0016 g/cm³), the **Chicxulub impact crater** (29, the K-Pg peak ring: 717 rows where no sample reaches 5,400 m/s against intact granite's 5,500–6,000 — the deficit IS the shock damage — in a dataset that checksums its own transcription), and the **central Arctic Ocean** (30, the ACEX age model at 87.89°N, 235 km from the North Pole: the catalogue's first time axis, 56 Myr of history, with a 26.2-Myr hiatus encoded as a duplicate-depth pair and all six sedimentation rates re-derived hiatus-aware). The ship pin measures the claim: the latitude span between catalogue endpoints — computed live from the entries' own in-file coordinates — exceeds 152 degrees. **THIRTY entries / EIGHTEEN regions / NINETEEN kinds: pole to pole, seafloor mud to mantle, 66-million-year-old impact shock to day-indexed production data — every dataset verbatim, every flaw kept, every derivable number re-earned by the gate on every run.** **Totals: 2,253 wired (2,308 DISPATCH keys) / gate 5,906 green / 4,180 defs. Next paper: PAPER_2258.**
+**This release (v0.403.0): THE PRODUCT SHIP — THE INDEPENDENT EVALUATION'S FINISH SEQUENCE, EXECUTED.** An independent component-inventory and gap analysis of the v1.30.0 simulator (supplied 2026-08-27, verified claim-by-claim against the code, adopted verbatim as the standing plan) drew the line: after steps 1–5 it is a finished offline product. Eight package versions later (`uqff_downhole_simulator` v1.41.0–v1.48.0), the line is crossed and machine-verified. **The finish sequence:** (1) Modbus unified — the registry entry IS the real pymodbus client, refusal disciplined and minimal (refuses ONLY when host/register map are missing, naming which), the static source now disclosing what the runtime does; (2) the **well assembler** un-strands the catalogue — one WellAssembly per site family (KTB-HB, Site 1027, U1324, 504B), strict measured-coverage lookups that refuse rather than clamp, overburden integrated from the site's own density; (3) **measured wells are the engine defaults** — `run --well ktb_hb` simulates on the 1994 German deep-hole log, and the catalogued Volve F-12 measured downhole pressure becomes the reconciler's live leg, its −1,096 psi/yr drawdown classified UNEXPLAINED_TREND (reservoir physics correctly refused as gauge drift); (4) the **operator surface** — a headless, fully gate-tested OperatorSession under a Qt6 view: well picker, toolstring builder whose rating check BLOCKS the run against measured conditions (the 1994 KTB log is hotter than the 177 °C tool class at depth — no contrived kick needed), live P/T + twin-leg drift tabs, undervalued-stream alerts, and a citations pane that never disappears; (5) the **in-package acceptance suite** — `python -m uqff_downhole_simulator accept`, 55 checks across nine sections, independence from the physics corpus enforced by construction (the fidelity gate runs it as a subprocess AND statically verifies the module never references the calculator or any paper). **The field tier opens:** (6a) **lithology from the catalogue's own gamma curves** — six entries qualify under unit-disciplined detection (the catalogue's own trap cases pinned: KTB GRAV/mGals is gravimetry), the KTB pilot's 1,082-point log yielding 80 alternating intervals and Volve SR the textbook sand/shale split, every number labeled (INDUSTRY_STANDARD / STATISTICAL_PICKS / CONVENTION / no invented detector); (6b) **mixed toolstrings** — per-station tool models with honest legs (twin / reference-only / labeled class envelope / refused-but-still-streaming-well-P/T), aggregates computed over twin stations only with counts disclosed, and the rating check enforced INSIDE the engine constructor; (8) the **bench-test protocol** — BENCH_TEST_PROTOCOL.md ships in-package with the falsifiable prediction (paired GEOQ-class gauges, drift ratio R = 1.0324 at unity trims, ≥90 days honoring the reconciler's own 18-day floor), both-outcome labeling rules (confirmation → MEASURED_ON_BENCH with the full record attached; refutation → DERIVED_HYBRID with the refutation on record; no silent trim retuning), and `uqff_bench.py`'s four-verdict analysis self-verified on synthetic legs in output labeled SIMULATION_SELF_TEST — arithmetic verified, physics honestly unclaimed. **Remaining: step 7 only** — one live site path, blocked on what no code can supply (a real host and a cited register map from a real site). DERIVED_HYBRID stayed DERIVED_HYBRID throughout; U_i stays loaded-but-unused pending the Rule-10 derivation path; the twin solution track was audited by measurement at every layer mid-arc at Daniel's direction, and three small gaps that audit found were fixed in-arc, not deferred. **Totals: 2,253 wired (2,308 DISPATCH keys) / gate 5,925 green / 4,180 defs / acceptance 55 green. Next paper: PAPER_2258.**
 
 Author: Daniel T. Murphy · Star-Magic Research Program
 License: AGPL-3.0-or-later OR Commercial
@@ -106,7 +107,7 @@ Mizuno, Rossi).
 Full framework physics lives in the whitepaper corpus — the physics is the
 whitepapers; the calculator computes what the whitepapers derive.
 
-## What is currently shipped (v0.402.0)
+## What is currently shipped (v0.403.0)
 
 ### Wiring campaign — LIVE (see `CLAUDE.md` charter)
 

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/README.md

@@ -1,16 +1,16 @@
 # Star-Magic-Program
 
-[![PyPI version](https://img.shields.io/pypi/v/star-magic-program.svg?cacheBust=0.402.0)](https://pypi.org/project/star-magic-program/)
-[![Python versions](https://img.shields.io/pypi/pyversions/star-magic-program.svg?cacheBust=0.402.0)](https://pypi.org/project/star-magic-program/)
+[![PyPI version](https://img.shields.io/pypi/v/star-magic-program.svg?cacheBust=0.403.0)](https://pypi.org/project/star-magic-program/)
+[![Python versions](https://img.shields.io/pypi/pyversions/star-magic-program.svg?cacheBust=0.403.0)](https://pypi.org/project/star-magic-program/)
 [![Documentation Status](https://readthedocs.org/projects/star-magic-program/badge/?version=latest)](https://star-magic-program.readthedocs.io/en/latest/?badge=latest)
 [![License: AGPL-3.0 + Commercial](https://img.shields.io/badge/License-AGPL--3.0%20%2B%20Commercial-blue.svg)](LICENSE)
-[![Fidelity gate](https://img.shields.io/badge/fidelity_gate-5906%2F0-brightgreen)](uqff_fidelity_tests.py)
+[![Fidelity gate](https://img.shields.io/badge/fidelity_gate-5925%2F0-brightgreen)](uqff_fidelity_tests.py)
 [![Public surfaces](https://img.shields.io/badge/public_surfaces-2308-blue)](uqff_calculator.py)
 [![Whitepapers](https://img.shields.io/badge/whitepapers-2292-orange)](whitepapers/)
 
-**UQFF systematic rebuild — v0.402.0 complete-compile campaign live**
+**UQFF systematic rebuild — v0.403.0 complete-compile campaign live**
 
-**This release (v0.402.0): THE POLE-TO-POLE SHIP — THE CATALOGUE GROWS FROM TWENTY TO THIRTY ENTRIES, AND THE SPAN IS MEASURED.** Ten catalogue packages (`uqff_downhole_simulator` v1.31.0–v1.40.0), each a complete public dataset transcribed verbatim with a provenance sidecar, each verified live by the gate. The arc opens with **the heat-flow closure** (entry 21: thermal conductivity of ODP 1027B pairs with entry 20's CORK temperatures — q = k·dT/dz ≈ 130 mW/m² computed from two catalogued datasets of one site, not quoted) and closes at the North Pole. **DSDP/ODP Hole 504B becomes a four-dataset family** — 1979 sound velocity, 1979 physical properties (the WBD identity closing on all 61 rows), 1986 sheeted-dike elastic moduli (four identities re-derived on every run), 1991 borehole fluids — and the two 1979 tables archive the same core suite, so the gate joins them sample-by-sample and computes a **live acoustic-impedance profile** (Z = ρ·Vp = 13.9–18.5 ×10⁶ kg/m²s) from independently archived datasets. **Six regions join**: Antarctica (entry 23, Prydz Bay — whose replicate audit caught a dropped-digit typo IN the source archive), the Nankai Trough (26, first subduction zone and first Chikyu data, with Ikari et al.'s slope-stability conclusion re-derived: FS 2.38–15.31, all stations stable), the Southwest Indian Ridge (27, Atlantis Bank gabbros — completing the **oceanic-crust ladder**: sediments → pillow basalt → sheeted dikes → layer-3 gabbro), the Mid-Atlantic Ridge (28, serpentinized **mantle peridotite** — the ladder goes below the crust, the five-identity lock closing at 0.0016 g/cm³), the **Chicxulub impact crater** (29, the K-Pg peak ring: 717 rows where no sample reaches 5,400 m/s against intact granite's 5,500–6,000 — the deficit IS the shock damage — in a dataset that checksums its own transcription), and the **central Arctic Ocean** (30, the ACEX age model at 87.89°N, 235 km from the North Pole: the catalogue's first time axis, 56 Myr of history, with a 26.2-Myr hiatus encoded as a duplicate-depth pair and all six sedimentation rates re-derived hiatus-aware). The ship pin measures the claim: the latitude span between catalogue endpoints — computed live from the entries' own in-file coordinates — exceeds 152 degrees. **THIRTY entries / EIGHTEEN regions / NINETEEN kinds: pole to pole, seafloor mud to mantle, 66-million-year-old impact shock to day-indexed production data — every dataset verbatim, every flaw kept, every derivable number re-earned by the gate on every run.** **Totals: 2,253 wired (2,308 DISPATCH keys) / gate 5,906 green / 4,180 defs. Next paper: PAPER_2258.**
+**This release (v0.403.0): THE PRODUCT SHIP — THE INDEPENDENT EVALUATION'S FINISH SEQUENCE, EXECUTED.** An independent component-inventory and gap analysis of the v1.30.0 simulator (supplied 2026-08-27, verified claim-by-claim against the code, adopted verbatim as the standing plan) drew the line: after steps 1–5 it is a finished offline product. Eight package versions later (`uqff_downhole_simulator` v1.41.0–v1.48.0), the line is crossed and machine-verified. **The finish sequence:** (1) Modbus unified — the registry entry IS the real pymodbus client, refusal disciplined and minimal (refuses ONLY when host/register map are missing, naming which), the static source now disclosing what the runtime does; (2) the **well assembler** un-strands the catalogue — one WellAssembly per site family (KTB-HB, Site 1027, U1324, 504B), strict measured-coverage lookups that refuse rather than clamp, overburden integrated from the site's own density; (3) **measured wells are the engine defaults** — `run --well ktb_hb` simulates on the 1994 German deep-hole log, and the catalogued Volve F-12 measured downhole pressure becomes the reconciler's live leg, its −1,096 psi/yr drawdown classified UNEXPLAINED_TREND (reservoir physics correctly refused as gauge drift); (4) the **operator surface** — a headless, fully gate-tested OperatorSession under a Qt6 view: well picker, toolstring builder whose rating check BLOCKS the run against measured conditions (the 1994 KTB log is hotter than the 177 °C tool class at depth — no contrived kick needed), live P/T + twin-leg drift tabs, undervalued-stream alerts, and a citations pane that never disappears; (5) the **in-package acceptance suite** — `python -m uqff_downhole_simulator accept`, 55 checks across nine sections, independence from the physics corpus enforced by construction (the fidelity gate runs it as a subprocess AND statically verifies the module never references the calculator or any paper). **The field tier opens:** (6a) **lithology from the catalogue's own gamma curves** — six entries qualify under unit-disciplined detection (the catalogue's own trap cases pinned: KTB GRAV/mGals is gravimetry), the KTB pilot's 1,082-point log yielding 80 alternating intervals and Volve SR the textbook sand/shale split, every number labeled (INDUSTRY_STANDARD / STATISTICAL_PICKS / CONVENTION / no invented detector); (6b) **mixed toolstrings** — per-station tool models with honest legs (twin / reference-only / labeled class envelope / refused-but-still-streaming-well-P/T), aggregates computed over twin stations only with counts disclosed, and the rating check enforced INSIDE the engine constructor; (8) the **bench-test protocol** — BENCH_TEST_PROTOCOL.md ships in-package with the falsifiable prediction (paired GEOQ-class gauges, drift ratio R = 1.0324 at unity trims, ≥90 days honoring the reconciler's own 18-day floor), both-outcome labeling rules (confirmation → MEASURED_ON_BENCH with the full record attached; refutation → DERIVED_HYBRID with the refutation on record; no silent trim retuning), and `uqff_bench.py`'s four-verdict analysis self-verified on synthetic legs in output labeled SIMULATION_SELF_TEST — arithmetic verified, physics honestly unclaimed. **Remaining: step 7 only** — one live site path, blocked on what no code can supply (a real host and a cited register map from a real site). DERIVED_HYBRID stayed DERIVED_HYBRID throughout; U_i stays loaded-but-unused pending the Rule-10 derivation path; the twin solution track was audited by measurement at every layer mid-arc at Daniel's direction, and three small gaps that audit found were fixed in-arc, not deferred. **Totals: 2,253 wired (2,308 DISPATCH keys) / gate 5,925 green / 4,180 defs / acceptance 55 green. Next paper: PAPER_2258.**
 
 Author: Daniel T. Murphy · Star-Magic Research Program
 License: AGPL-3.0-or-later OR Commercial
@@ -68,7 +68,7 @@ Mizuno, Rossi).
 Full framework physics lives in the whitepaper corpus — the physics is the
 whitepapers; the calculator computes what the whitepapers derive.
 
-## What is currently shipped (v0.402.0)
+## What is currently shipped (v0.403.0)
 
 ### Wiring campaign — LIVE (see `CLAUDE.md` charter)
 

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/SESSION_LOG.md

@@ -12081,3 +12081,80 @@ release paragraph rewritten + badges (fidelity 5906, cacheBust ×2);
 CHANGELOG + _BUILD_LOG + SHIP_MESSAGE + RULINGS trail; 10 POLE_TO_POLE_ARC
 trail rows + WHITEPAPER_INDEX ship note. Gate 5,904 → 5,906, 0 failures.
 Awaiting Daniel: `.\ship.ps1` (clear .git/index.lock first).
+
+## Entry 275 — 2026-08-27 — v0.403.0 THE PRODUCT SHIP (downhole v1.41–v1.48: the finish sequence)
+
+Daniel supplied an independent evaluation of the v1.30.0 simulator and
+ruled "I want to follow the plan if it looks right." Verified
+claim-by-claim against the code (all four spot-checks held; two stale
+catalogue counts corrected), adopted verbatim, executed step-by-step on
+GO-cadence:
+
+1 (v1.41) modbus unification — split was already unified AT RUNTIME (the
+evaluator's static read could not see it); fixed both real halves:
+source-honesty disclosure + disciplined config refusal. 2 (v1.42) well
+assembler — catalogue un-stranded; strict coverage; KTB overburden 190.9
+MPa under 253-MPa UCS. 3 (v1.43) measured defaults — demo_config +
+CLI --well/wells; production_live_stream (bar→psi labeled; 9-day stuck
+fault survives; station MD caller-supplied); Volve drawdown −1,096 psi/yr
+= UNEXPLAINED_TREND. 4 (v1.44) operator surface — headless session fully
+gate-tested + thin Qt view; blocking rating check demonstrated on REAL
+data (KTB window 184 C > 177 C tool class); ServiceLifeSimulator API
+mismatch found+fixed. 5 (v1.45) acceptance suite — in-package, grown to
+55 checks; independence by construction; caught its own docstring
+violating the static check, a case-study --well gap, and its own A2
+string-luck flakiness. 6a (v1.46) gamma — unit-disciplined (trap cases
+GRAV/Density-grain pinned); KTB 80 intervals; labels everywhere.
+6b (v1.47) mixed toolstrings — honest legs; in-engine rating block;
+first demo steered to the CORK column by the product's own rules.
+8 (v1.48) bench protocol + four-verdict analysis (selftest R=1.0263
+±0.0118 CONFIRMS; refute/span/SNR paths earned; SIMULATION_SELF_TEST
+self-labeled).
+
+Mid-arc audit (Daniel's three questions): twin track verified by
+measurement at every layer; UQFF-catalogue coordination honestly
+inventoried (composition evaluated at measured conditions + envelope
+confronted with field data; no derivation-match yet — step 8 territory by
+design); sweep found 3 real gaps (gui extra, LAS button, drift tab),
+fixed in-arc; two sweep greps were substring false-positives, re-verified
+before certifying.
+
+Ship prep: version sync 0.403.0 (desc 500 chars) across pyproject /
+calculator / gate / CITATION ×2 / registry version; SHIP GUARD marker
+POLE_TO_POLE_ARC→PRODUCT_ARC (+ GAPS id product_arc_rule7); +2 ship pins
+(live product-surface census + standing record); README release
+paragraph + campaign/shipped lines + badges; CHANGELOG + _BUILD_LOG +
+SHIP_MESSAGE + RULINGS trail; 10 PRODUCT_ARC trail rows + WHITEPAPER_INDEX
+note. Gate 5,922 → 5,924, 0 failures. Remaining: step 7 only (site
+details). Awaiting Daniel: `.\ship.ps1` (clear .git/index.lock first).
+
+## Entry 276 — 2026-08-27 — v0.403.0 REMANUFACTURE (numpy-2 trapz + stale staging tree)
+
+CI failure diagnosed from Daniel's log screenshots: fidelity-gate (3.12)
+died at gate line 13695 (FINISH-SEQ 2 pin) -> uqff_well_assembler.py:136
+`np.trapz` -> AttributeError: numpy >= 2.0 removed the alias. Root cause
+of the blind-repro failure: the sandbox's proxy-pinned numpy 2.2.6 STILL
+carries trapz, so every local reproduction (build under setuptools 82/84,
+pip install ., gate without pymodbus/matplotlib, 36 s green) passed while
+every CI runner failed. Fix: module-level _TRAPEZOID = np.trapezoid with
+1.x fallback; overburden call switched; PORTABILITY GUARD 2 pin bans
+np.trapz in the assembler and re-runs the 190.9-MPa KTB integral on the
+current interpreter's numpy. STANDING RULE canonized: local dependency
+versions are NOT the ship's dependency versions - removed-alias sweeps
+(trapz/in1d/alltrue/product/row_stack/NaN: repo-wide sweep found exactly
+the one hit) are part of ship prep.
+
+Same audit found Daniel's 'stale files': the TRACKED
+star_magic_program-0.402.0/ sdist staging tree (129 files, ~5 MB - the
+failed in-repo v0.402-prep build could not delete its staging dir on the
+mounted FS and add -A committed it; it rode through v0.402.0's green CI
+as dead weight). Sandbox cannot delete it (same mount permission);
+.gitignore now carries star_magic_program-*/; Daniel removes it on
+Windows: git rm -r star_magic_program-0.402.0. Hardening kept from the
+diagnosis pass: acceptance A2 stochastic band widened, acceptance-
+subprocess pin embeds stderr tail + timeout.
+
+Per Daniel: tag v0.403.0 NOT burned (PyPI never received it) - same
+version remanufactured. Gate 5,924 -> 5,925, 0 failures; acceptance
+55/55; sdist/wheel rebuilt clean. Daniel: delete the stale tree, commit,
+re-tag v0.403.0 (move tag to the new commit), push.

star_magic_program-0.403.0/UNIFIED_REGISTRY_VERSION.txt

@@ -0,0 +1 @@
+v0.403.0

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/UNIFIED_REGISTRY_XGEO_QUEUE.csv

@@ -1729,3 +1729,4 @@ DEEPDATA_ARC,n/a,n/a,NOT_APPLICABLE,application real-data catalogue single-geome
 TWENTYWELLS_ARC,n/a,n/a,NOT_APPLICABLE,application real-data catalogue single-geometry,v0.400.0
 INCORPORATION_ARC,n/a,n/a,NOT_APPLICABLE,corporate record ship - no physics content,v0.401.0
 POLE_TO_POLE_ARC,n/a,n/a,NOT_APPLICABLE,real-data catalogue ship - no cross-geometry content,v0.402.0
+PRODUCT_ARC,n/a,n/a,NOT_APPLICABLE,product-engineering ship - no cross-geometry content,v0.403.0

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/UNIFIED_REGISTRY_XGEO_ROUTES.csv

@@ -1730,3 +1730,4 @@ DEEPDATA_ARC,n/a,XGEO_NOT_REQUIRED,none,v0.399.0
 TWENTYWELLS_ARC,n/a,XGEO_NOT_REQUIRED,none,v0.400.0
 INCORPORATION_ARC,n/a,XGEO_NOT_REQUIRED,none,v0.401.0
 POLE_TO_POLE_ARC,n/a,XGEO_NOT_REQUIRED,none,v0.402.0
+PRODUCT_ARC,n/a,XGEO_NOT_REQUIRED,none,v0.403.0

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/WHITEPAPER_INDEX.md

@@ -99,7 +99,7 @@ Every previously non-numeric key now carries a canonical PAPER number (2179-2212
 | 2233 | PAPER_1203b |
 
 - **Verification arc 2026-08-16 (v0.387.0):** paradox-corpus audit ZERO recalcs; grammar dragnet 110 family sites; origin term verified; AP text layer EXHAUSTED (458 docs). Runnable queue empty — remaining items Daniel-owned.
-- **Distinct wired papers: 2,253 (+55 canonical alias numbers 2179-2233; 2,308 DISPATCH keys). PAPER_2257 two-stream architecture landmark, 2026-08-24; v0.398.0 CATALOGUE SHIP (downhole v1.14-v1.20, ten real-data entries), 2026-08-25; v0.399.0 DEEP DATA SHIP (v1.21-v1.25, entries 11-15, KTB suite + live overburden integral), 2026-08-25; v0.400.0 TWENTY WELLS SHIP (v1.26-v1.30, entries 16-20, quantity ledger closed), 2026-08-25; v0.401.0 INCORPORATION SHIP (six ENRGYONE founding PDFs in pdf/), 2026-08-25; v0.402.0 POLE-TO-POLE SHIP (v1.31-v1.40, entries 21-30: heat-flow closure, 504B four-dataset family + impedance join, crust ladder complete to mantle, Chicxulub, ACEX at 87.89N - thirty entries pole to pole, latitude span >152 deg measured live), 2026-08-27.** = `wired_count()` = `len(DISPATCH)` (live at band 2121-2130, 2026-08-16; earlier eras' 1,417 figure superseded).
+- **Distinct wired papers: 2,253 (+55 canonical alias numbers 2179-2233; 2,308 DISPATCH keys). PAPER_2257 two-stream architecture landmark, 2026-08-24; v0.398.0 CATALOGUE SHIP (downhole v1.14-v1.20, ten real-data entries), 2026-08-25; v0.399.0 DEEP DATA SHIP (v1.21-v1.25, entries 11-15, KTB suite + live overburden integral), 2026-08-25; v0.400.0 TWENTY WELLS SHIP (v1.26-v1.30, entries 16-20, quantity ledger closed), 2026-08-25; v0.401.0 INCORPORATION SHIP (six ENRGYONE founding PDFs in pdf/), 2026-08-25; v0.402.0 POLE-TO-POLE SHIP (v1.31-v1.40, entries 21-30: heat-flow closure, 504B four-dataset family + impedance join, crust ladder complete to mantle, Chicxulub, ACEX at 87.89N - thirty entries pole to pole, latitude span >152 deg measured live), 2026-08-27; v0.403.0 PRODUCT SHIP (v1.41-v1.48: evaluation finish sequence 1-5 executed = finished offline product + field tier 6a/6b/8; acceptance suite 55 checks in-package; step 7 alone remains, site-blocked), 2026-08-27.** = `wired_count()` = `len(DISPATCH)` (live at band 2121-2130, 2026-08-16; earlier eras' 1,417 figure superseded).
 - **Index table file-row marks:** **934 ✓ CLEAN**, **245 ⚠ OPEN_RULING**, **1076 ⬜** not-touched (934 + 245 + 1076 = 2255 ✓). Wired file-rows (934 + 245 = 1,179) sit below `wired_count()` = 1,417 because variant files share a base dispatch.
 - **ORPHAN-PHYSICS (v0.367.1 audit):** no paper numbers are missing, but 71 non-numbered `.md` files in the predecessor hold **6,615 equation blocks** outside the corpus. Queued as Q-ORPHAN-PHYSICS.
 - **OPEN targets:** 0

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/_BUILD_LOG.md

@@ -327,3 +327,9 @@ THE INCORPORATION SHIP: six ENRGYONE, Inc. founding documents (Articles, Bylaws,
 
 ## v0.402.0 — 2026-08-27
 THE POLE-TO-POLE SHIP: downhole v1.31-v1.40 — catalogue 20→30 entries / 18 regions / 19 kinds. Heat-flow closure computed live; 504B four-dataset family + cross-entry impedance join; Antarctica/Nankai/SW-Indian/Mid-Atlantic/Chicxulub/central-Arctic join; crust ladder complete seafloor-mud→mantle; ACEX age model = first time axis; two source-archive typos and two archival row-slips detected and disclosed, never repaired. Ship pin measures the latitude span (>152 deg) live. Gate 5,906/0. 23-file pass verified.
+
+## v0.403.0 — 2026-08-27
+THE PRODUCT SHIP: downhole v1.41-v1.48 — the independent evaluation's finish sequence executed 1-5 (modbus unified, well assembler, measured-well engine defaults, operator surface, 55-check in-package acceptance suite) = FINISHED OFFLINE PRODUCT by the evaluation's own criterion; field tier 6a gamma/lithology, 6b mixed toolstrings + in-engine rating block, 8 bench-test protocol with four-verdict self-labeled analysis. Remaining: step 7 only (site details). Gate 5,924/0. Acceptance 55/55. 23-file pass verified.
+
+## v0.403.0-remanufacture — 2026-08-27
+CI red on all runners: numpy>=2.0 removed np.trapz; assembler overburden now uses np.trapezoid (1.x fallback); PORTABILITY GUARD 2 pin + standing removed-alias-sweep rule. Stale TRACKED sdist staging tree star_magic_program-0.402.0/ identified (mounted-FS delete block; gitignored; Daniel removes on Windows). Same version per Daniel (tag preserved; PyPI untouched). Gate 5,925/0. Acceptance 55/55.

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/pyproject.toml

@@ -4,8 +4,8 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "star-magic-program"
-version = "0.402.0"
-description = "Star-Magic UQFF calculator v0.402.0: THE POLE-TO-POLE SHIP. 2,253 dispatches (2,308 keys), gate 5,906/0. Downhole catalogue 20->30 entries (v1.31-v1.40): live heat-flow closure q=k*dT/dz; 504B four-dataset family with cross-entry impedance join; Antarctica, Nankai, SW Indian Ridge, Mid-Atlantic, Chicxulub and central Arctic (87.9N) join; crust ladder complete from seafloor mud to mantle peridotite; ACEX age model = first time axis. 18 regions / 19 kinds. Next: PAPER_2258."
+version = "0.403.0"
+description = "Star-Magic UQFF calculator v0.403.0: THE PRODUCT SHIP. 2,253 dispatches (2,308 keys), gate 5,924/0. The downhole simulator completes the adopted independent-evaluation plan (v1.41-v1.48): finish sequence 1-5 - modbus unified, well assembler, measured wells as engine defaults, operator surface, in-package 55-check acceptance suite = FINISHED OFFLINE PRODUCT; field tier 6a gamma/lithology from catalogue curves, 6b mixed toolstrings + in-engine rating block, 8 bench-test protocol. Next: PAPER_2258."
 readme = "README.md"
 authors = [
     { name = "Daniel T. Murphy", email = "daniel.murphy00@enrgyone.com" }
@@ -54,7 +54,7 @@ Issues = "https://github.com/Daniel8Murphy0007/Star-Magic-Program/issues"
 
 [project.optional-dependencies]
 modbus = ["pymodbus>=3.15"]
-gui = ["PyQt6"]
+gui = ["PyQt6", "matplotlib"]
 
 [tool.setuptools]
 packages = ["uqff_downhole_simulator"]
@@ -163,6 +163,7 @@ py-modules = ["uqff_session_closures", "uqff_fubii_variants", "uqff_ngc_catalog"
     "uqff_downhole_simulator/catalog/odp_1274a_mantle_peridotite_mad.provenance.json",
     "uqff_downhole_simulator/catalog/chicxulub_m0077a_pwave_velocity.txt",
     "uqff_downhole_simulator/catalog/chicxulub_m0077a_pwave_velocity.provenance.json",
+    "uqff_downhole_simulator/BENCH_TEST_PROTOCOL.md",
     "uqff_downhole_simulator/catalog/acex_lomonosov_age_depth_model.txt",
     "uqff_downhole_simulator/catalog/acex_lomonosov_age_depth_model.provenance.json",
     "uqff_downhole_simulator/catalog/volve_15_9_19a_core_excerpt.provenance.json",

{star_magic_program-0.402.0 → star_magic_program-0.403.0/star_magic_program.egg-info}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: star-magic-program
-Version: 0.402.0
-Summary: Star-Magic UQFF calculator v0.402.0: THE POLE-TO-POLE SHIP. 2,253 dispatches (2,308 keys), gate 5,906/0. Downhole catalogue 20->30 entries (v1.31-v1.40): live heat-flow closure q=k*dT/dz; 504B four-dataset family with cross-entry impedance join; Antarctica, Nankai, SW Indian Ridge, Mid-Atlantic, Chicxulub and central Arctic (87.9N) join; crust ladder complete from seafloor mud to mantle peridotite; ACEX age model = first time axis. 18 regions / 19 kinds. Next: PAPER_2258.
+Version: 0.403.0
+Summary: Star-Magic UQFF calculator v0.403.0: THE PRODUCT SHIP. 2,253 dispatches (2,308 keys), gate 5,924/0. The downhole simulator completes the adopted independent-evaluation plan (v1.41-v1.48): finish sequence 1-5 - modbus unified, well assembler, measured wells as engine defaults, operator surface, in-package 55-check acceptance suite = FINISHED OFFLINE PRODUCT; field tier 6a gamma/lithology from catalogue curves, 6b mixed toolstrings + in-engine rating block, 8 bench-test protocol. Next: PAPER_2258.
 Author-email: "Daniel T. Murphy" <daniel.murphy00@enrgyone.com>
 License: AGPL-3.0-or-later OR LicenseRef-StarMagic-Commercial
 Project-URL: Homepage, https://github.com/Daniel8Murphy0007/Star-Magic-Program
@@ -34,21 +34,22 @@ Provides-Extra: modbus
 Requires-Dist: pymodbus>=3.15; extra == "modbus"
 Provides-Extra: gui
 Requires-Dist: PyQt6; extra == "gui"
+Requires-Dist: matplotlib; extra == "gui"
 Dynamic: license-file
 
 # Star-Magic-Program
 
-[![PyPI version](https://img.shields.io/pypi/v/star-magic-program.svg?cacheBust=0.402.0)](https://pypi.org/project/star-magic-program/)
-[![Python versions](https://img.shields.io/pypi/pyversions/star-magic-program.svg?cacheBust=0.402.0)](https://pypi.org/project/star-magic-program/)
+[![PyPI version](https://img.shields.io/pypi/v/star-magic-program.svg?cacheBust=0.403.0)](https://pypi.org/project/star-magic-program/)
+[![Python versions](https://img.shields.io/pypi/pyversions/star-magic-program.svg?cacheBust=0.403.0)](https://pypi.org/project/star-magic-program/)
 [![Documentation Status](https://readthedocs.org/projects/star-magic-program/badge/?version=latest)](https://star-magic-program.readthedocs.io/en/latest/?badge=latest)
 [![License: AGPL-3.0 + Commercial](https://img.shields.io/badge/License-AGPL--3.0%20%2B%20Commercial-blue.svg)](LICENSE)
-[![Fidelity gate](https://img.shields.io/badge/fidelity_gate-5906%2F0-brightgreen)](uqff_fidelity_tests.py)
+[![Fidelity gate](https://img.shields.io/badge/fidelity_gate-5925%2F0-brightgreen)](uqff_fidelity_tests.py)
 [![Public surfaces](https://img.shields.io/badge/public_surfaces-2308-blue)](uqff_calculator.py)
 [![Whitepapers](https://img.shields.io/badge/whitepapers-2292-orange)](whitepapers/)
 
-**UQFF systematic rebuild — v0.402.0 complete-compile campaign live**
+**UQFF systematic rebuild — v0.403.0 complete-compile campaign live**
 
-**This release (v0.402.0): THE POLE-TO-POLE SHIP — THE CATALOGUE GROWS FROM TWENTY TO THIRTY ENTRIES, AND THE SPAN IS MEASURED.** Ten catalogue packages (`uqff_downhole_simulator` v1.31.0–v1.40.0), each a complete public dataset transcribed verbatim with a provenance sidecar, each verified live by the gate. The arc opens with **the heat-flow closure** (entry 21: thermal conductivity of ODP 1027B pairs with entry 20's CORK temperatures — q = k·dT/dz ≈ 130 mW/m² computed from two catalogued datasets of one site, not quoted) and closes at the North Pole. **DSDP/ODP Hole 504B becomes a four-dataset family** — 1979 sound velocity, 1979 physical properties (the WBD identity closing on all 61 rows), 1986 sheeted-dike elastic moduli (four identities re-derived on every run), 1991 borehole fluids — and the two 1979 tables archive the same core suite, so the gate joins them sample-by-sample and computes a **live acoustic-impedance profile** (Z = ρ·Vp = 13.9–18.5 ×10⁶ kg/m²s) from independently archived datasets. **Six regions join**: Antarctica (entry 23, Prydz Bay — whose replicate audit caught a dropped-digit typo IN the source archive), the Nankai Trough (26, first subduction zone and first Chikyu data, with Ikari et al.'s slope-stability conclusion re-derived: FS 2.38–15.31, all stations stable), the Southwest Indian Ridge (27, Atlantis Bank gabbros — completing the **oceanic-crust ladder**: sediments → pillow basalt → sheeted dikes → layer-3 gabbro), the Mid-Atlantic Ridge (28, serpentinized **mantle peridotite** — the ladder goes below the crust, the five-identity lock closing at 0.0016 g/cm³), the **Chicxulub impact crater** (29, the K-Pg peak ring: 717 rows where no sample reaches 5,400 m/s against intact granite's 5,500–6,000 — the deficit IS the shock damage — in a dataset that checksums its own transcription), and the **central Arctic Ocean** (30, the ACEX age model at 87.89°N, 235 km from the North Pole: the catalogue's first time axis, 56 Myr of history, with a 26.2-Myr hiatus encoded as a duplicate-depth pair and all six sedimentation rates re-derived hiatus-aware). The ship pin measures the claim: the latitude span between catalogue endpoints — computed live from the entries' own in-file coordinates — exceeds 152 degrees. **THIRTY entries / EIGHTEEN regions / NINETEEN kinds: pole to pole, seafloor mud to mantle, 66-million-year-old impact shock to day-indexed production data — every dataset verbatim, every flaw kept, every derivable number re-earned by the gate on every run.** **Totals: 2,253 wired (2,308 DISPATCH keys) / gate 5,906 green / 4,180 defs. Next paper: PAPER_2258.**
+**This release (v0.403.0): THE PRODUCT SHIP — THE INDEPENDENT EVALUATION'S FINISH SEQUENCE, EXECUTED.** An independent component-inventory and gap analysis of the v1.30.0 simulator (supplied 2026-08-27, verified claim-by-claim against the code, adopted verbatim as the standing plan) drew the line: after steps 1–5 it is a finished offline product. Eight package versions later (`uqff_downhole_simulator` v1.41.0–v1.48.0), the line is crossed and machine-verified. **The finish sequence:** (1) Modbus unified — the registry entry IS the real pymodbus client, refusal disciplined and minimal (refuses ONLY when host/register map are missing, naming which), the static source now disclosing what the runtime does; (2) the **well assembler** un-strands the catalogue — one WellAssembly per site family (KTB-HB, Site 1027, U1324, 504B), strict measured-coverage lookups that refuse rather than clamp, overburden integrated from the site's own density; (3) **measured wells are the engine defaults** — `run --well ktb_hb` simulates on the 1994 German deep-hole log, and the catalogued Volve F-12 measured downhole pressure becomes the reconciler's live leg, its −1,096 psi/yr drawdown classified UNEXPLAINED_TREND (reservoir physics correctly refused as gauge drift); (4) the **operator surface** — a headless, fully gate-tested OperatorSession under a Qt6 view: well picker, toolstring builder whose rating check BLOCKS the run against measured conditions (the 1994 KTB log is hotter than the 177 °C tool class at depth — no contrived kick needed), live P/T + twin-leg drift tabs, undervalued-stream alerts, and a citations pane that never disappears; (5) the **in-package acceptance suite** — `python -m uqff_downhole_simulator accept`, 55 checks across nine sections, independence from the physics corpus enforced by construction (the fidelity gate runs it as a subprocess AND statically verifies the module never references the calculator or any paper). **The field tier opens:** (6a) **lithology from the catalogue's own gamma curves** — six entries qualify under unit-disciplined detection (the catalogue's own trap cases pinned: KTB GRAV/mGals is gravimetry), the KTB pilot's 1,082-point log yielding 80 alternating intervals and Volve SR the textbook sand/shale split, every number labeled (INDUSTRY_STANDARD / STATISTICAL_PICKS / CONVENTION / no invented detector); (6b) **mixed toolstrings** — per-station tool models with honest legs (twin / reference-only / labeled class envelope / refused-but-still-streaming-well-P/T), aggregates computed over twin stations only with counts disclosed, and the rating check enforced INSIDE the engine constructor; (8) the **bench-test protocol** — BENCH_TEST_PROTOCOL.md ships in-package with the falsifiable prediction (paired GEOQ-class gauges, drift ratio R = 1.0324 at unity trims, ≥90 days honoring the reconciler's own 18-day floor), both-outcome labeling rules (confirmation → MEASURED_ON_BENCH with the full record attached; refutation → DERIVED_HYBRID with the refutation on record; no silent trim retuning), and `uqff_bench.py`'s four-verdict analysis self-verified on synthetic legs in output labeled SIMULATION_SELF_TEST — arithmetic verified, physics honestly unclaimed. **Remaining: step 7 only** — one live site path, blocked on what no code can supply (a real host and a cited register map from a real site). DERIVED_HYBRID stayed DERIVED_HYBRID throughout; U_i stays loaded-but-unused pending the Rule-10 derivation path; the twin solution track was audited by measurement at every layer mid-arc at Daniel's direction, and three small gaps that audit found were fixed in-arc, not deferred. **Totals: 2,253 wired (2,308 DISPATCH keys) / gate 5,925 green / 4,180 defs / acceptance 55 green. Next paper: PAPER_2258.**
 
 Author: Daniel T. Murphy · Star-Magic Research Program
 License: AGPL-3.0-or-later OR Commercial
@@ -106,7 +107,7 @@ Mizuno, Rossi).
 Full framework physics lives in the whitepaper corpus — the physics is the
 whitepapers; the calculator computes what the whitepapers derive.
 
-## What is currently shipped (v0.402.0)
+## What is currently shipped (v0.403.0)
 
 ### Wiring campaign — LIVE (see `CLAUDE.md` charter)
 

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/star_magic_program.egg-info/SOURCES.txt

@@ -52,19 +52,24 @@ star_magic_program.egg-info/SOURCES.txt
 star_magic_program.egg-info/dependency_links.txt
 star_magic_program.egg-info/requires.txt
 star_magic_program.egg-info/top_level.txt
+uqff_downhole_simulator/BENCH_TEST_PROTOCOL.md
 uqff_downhole_simulator/README.md
 uqff_downhole_simulator/__init__.py
 uqff_downhole_simulator/__main__.py
+uqff_downhole_simulator/acceptance_tests.py
 uqff_downhole_simulator/example_register_map.json
 uqff_downhole_simulator/matplotlib_demo.py
 uqff_downhole_simulator/qt6_downhole_app.py
 uqff_downhole_simulator/sample_well_profile.csv
+uqff_downhole_simulator/uqff_bench.py
 uqff_downhole_simulator/uqff_case_study.py
 uqff_downhole_simulator/uqff_deviation.py
 uqff_downhole_simulator/uqff_downhole_engine.py
 uqff_downhole_simulator/uqff_follower.py
+uqff_downhole_simulator/uqff_gamma.py
 uqff_downhole_simulator/uqff_gauge_specs.py
 uqff_downhole_simulator/uqff_modbus.py
+uqff_downhole_simulator/uqff_operator_app.py
 uqff_downhole_simulator/uqff_ports.py
 uqff_downhole_simulator/uqff_profile_catalog.py
 uqff_downhole_simulator/uqff_quartz_hpht_extension.py
@@ -72,6 +77,7 @@ uqff_downhole_simulator/uqff_reconciler.py
 uqff_downhole_simulator/uqff_service_life.py
 uqff_downhole_simulator/uqff_telemetry.py
 uqff_downhole_simulator/uqff_tool_library.py
+uqff_downhole_simulator/uqff_well_assembler.py
 uqff_downhole_simulator/catalog/acex_lomonosov_age_depth_model.provenance.json
 uqff_downhole_simulator/catalog/acex_lomonosov_age_depth_model.txt
 uqff_downhole_simulator/catalog/agassiz77_canada_temperature.csv

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/star_magic_program.egg-info/requires.txt

@@ -5,6 +5,7 @@ scipy>=1.10
 
 [gui]
 PyQt6
+matplotlib
 
 [modbus]
 pymodbus>=3.15

{star_magic_program-0.402.0 → star_magic_program-0.403.0}/uqff_calculator.py

@@ -73,7 +73,7 @@ from uqff_registry_primitives import (
     DELTA_M2_21_EV2, DELTA_M2_32_EV2,
 )
 
-VERSION = "0.402.0"
+VERSION = "0.403.0"
 
 # NAMED OBSERVED SI ANCHORS (constant drain 2026-08-16, PAPER_2141 bulk pattern + PAPER_2149 observation-headlining)
 # Bit-identical to the literals they replace; UQFF-derived counterparts live in uqff_registry_primitives.

star_magic_program-0.403.0/uqff_downhole_simulator/BENCH_TEST_PROTOCOL.md

@@ -0,0 +1,97 @@
+# Bench-Test Protocol — Measuring the 1.0324 Suppression Ratio
+
+**Field-tier step 8 (independent evaluation, adopted 2026-08-27).**
+Status of the quantity under test: `canonical_suppression()` = **1.0324 at
+unity trims** is today a **DERIVED_HYBRID** composition — industry baseline
+drift (0.215 %FS/yr class) with HPHT stress dressing, divided by a
+locked-primitive UQFF composition. It is **not** a measured constant, and the
+product labels it so everywhere it appears. This protocol is the experiment
+that would change — or refuse to change — that label.
+
+---
+
+## 1. Falsifiable prediction
+
+Two matched GEOQ-177-class quartz P/T gauges (same spec, same batch
+preferred), one operated with the UQFF-informed configuration and one as the
+conventional reference, co-located at a fixed HPHT setpoint, will exhibit a
+long-term drift-rate ratio
+
+```
+R = drift_conventional / drift_uqff = 1.0324   (at unity trims)
+```
+
+equivalently a predicted separation rate of ~2.3 psi/yr at a 30,000-psi
+full scale (the service-life module's number for these conditions). The
+prediction is falsifiable in both directions: a measured R statistically
+inconsistent with 1.0324 **refutes the composition at bench conditions**.
+
+## 2. Apparatus
+
+- 2× (minimum; 3+ pairs preferred for scatter) quartz P/T gauges,
+  177 °C / 30 kpsi class, from the same vendor lot.
+- Temperature-controlled bath or HPHT vessel holding a setpoint in the
+  150 °C / 10,000 psi class (inside every rating — the rating_check rule
+  applies on the bench too).
+- Reference pressure standard, 0.01 %-class (deadweight tester or
+  transfer standard), for baseline and periodic checks.
+- Logging at ≥ 1 reading/day per gauge into the package's historian CSV
+  format (`time_s` + one raw pressure column per gauge).
+
+## 3. Duration — the package's own rule applies
+
+The reconciler refuses trend verdicts below its minimum span
+(`ReconcilerConfig.min_trend_span_years`, the ≥18-day slope rule). The bench
+protocol requires **≥ 90 days** at setpoint for slope confidence; 180+ days
+preferred. The analysis tool enforces the 18-day floor mechanically and
+reports `INSUFFICIENT_SPAN` below it — the bench cannot be rushed past the
+product's own honesty rule.
+
+## 4. Procedure
+
+1. Baseline both gauges against the reference standard; record offsets.
+2. Install co-located at the setpoint; log continuously.
+3. Weekly: verify setpoint stability against the reference (do not adjust
+   the gauges under test).
+4. Export each leg as a historian CSV; run
+   `python -m uqff_downhole_simulator bench --uqff-csv A.csv --conv-csv B.csv`.
+5. Repeat with roles swapped on a second pair (control for unit-to-unit
+   scatter — the piezo lesson: individual units scatter around class curves).
+
+## 5. Analysis and verdict vocabulary
+
+`uqff_bench.bench_analysis()` fits each leg's drift slope (OLS, psi/yr →
+%FS/yr), propagates slope uncertainties into the ratio, and returns one of:
+
+- `MEASURED_CONFIRMS` — R within its own propagated uncertainty of 1.0324.
+- `MEASURED_REFUTES`  — R inconsistent with 1.0324 (the uncertainty is small
+  enough to exclude it). **This is a valid and useful outcome** — the
+  composition is falsified at bench conditions and the label stays
+  DERIVED_HYBRID with the refutation on record.
+- `INSUFFICIENT_SPAN` — under the 18-day floor: no verdict.
+- `INSUFFICIENT_SNR`  — the uncertainty band contains both 1.0324 and 1.0
+  (cannot distinguish suppression from no-suppression): no verdict, more
+  data required.
+
+## 6. Outcome handling (labeling rules)
+
+- **Confirmed:** `canonical_suppression()`'s label may move from
+  DERIVED_HYBRID to MEASURED_ON_BENCH **with the test ID, dates, apparatus
+  and per-pair results attached**. The composition's primitives are then a
+  measured-consistent model, not just a labeled one.
+- **Refuted:** the label stays DERIVED_HYBRID; the refutation (conditions,
+  measured R, uncertainty) is recorded beside it. No silent retuning of
+  trims to fit the bench — a post-hoc trim fit would be a new calibration
+  claim, labeled as such, never a derivation.
+- **Either way:** `U_i` remains loaded-but-unused in the drift formula until
+  a UQFF derivation path for its role is supplied (Rule 10 — the framework
+  author provides the physics; the product does not invent it).
+
+## 7. Analysis-tool verification (available today, and labeled)
+
+The analysis tool is verified NOW against the simulator's own twin legs:
+`bench --selftest` generates synthetic paired-leg data from the engine's
+models and must return `MEASURED_CONFIRMS` at R ≈ 1.0324. **This is a
+SIMULATION_SELF_TEST — it verifies the analysis arithmetic, not the
+physics.** It proves the bench pipeline is ready; it proves nothing about
+gauges. The acceptance suite runs it on every gate.