[gmongell]

Summary

• Package a bounded plane-wave optics alpha with PNG/CSV/JSON exports and explicit non-OLED limitations.
• Add analytic Fresnel/Airy, independent absorbing-stack matrix, power balance, input validation and buyer CLI tests.
• Preserve existing Apache-2.0 code and CC BY 4.0 documentation notices and include license texts.
• Add Windows/Linux Python 3.11–3.14 CI, build and installed-wheel checks, dependency auditing, checksums and optional draft-only release automation.
• Document proposed Gumroad copy and outstanding scientific, provenance and seller/commercial approval gates.

Validation

• Windows Python 3.14: 56 tests passed; JUnit parsed with zero failures/errors/skips.
• Wheel/sdist built; Twine checks passed; clean buyer installation and CLI output verified.
• Runtime dependency audit: no known vulnerabilities.

Not a paid launch

No seller registration, price, exclusive license, scientific certification, public release or paid listing is approved or submitted. This is not validated OLED dipole outcoupling. Remote CI results remain to be checked. Incorporates the bounded correction direction of draft PR #3 without claiming its release blockers are resolved.

[gmongell]

Verified CI evidence

Latest run: https://github.com/gmongell/Mathematica_python_TMM_OLEDPixel/actions/runs/38967043233

• All eight Windows/Linux Python 3.11, 3.12, 3.13 and 3.14 jobs passed: 56 cases per environment (448 passing case executions).
• Downloaded and parsed all eight JUnit reports: zero failures, errors or skips.
• Packaging, Twine validation, isolated buyer-wheel CLI execution and runtime dependency auditing passed.
• Downloaded distribution artifacts and independently verified their SHA-256 checksums and CycloneDX report (no known vulnerabilities).
• The initial audit blocked packaging; upgrading the clean buyer bootstrap pip resolved that gate without disabling auditing. Failed audit reports are now retained for diagnosis.
• Optional draft-release job was intentionally skipped; no release or paid storefront was published.

Requirement evidence: test_interface_matches_fresnel, test_lossless_film_matches_airy_formula, test_absorbing_stack_matches_characteristic_matrix, test_lossless_oblique_power_balance, test_cli_exports_plot_csv_and_metadata, test_invalid_scalar_inputs, test_cli_does_not_overwrite_existing_sidecar.

This remains a draft alpha plane-wave preview, not a validated OLED emission solver. Outstanding commercial/scientific approval gates remain in README. No merger, exclusive licensing, price, payout verification or production certification is implied.

[gmongell]

Added owner-review commercial and seller-onboarding drafts plus a bounded OLED dipole-solver specification in commit 0e5bfd7.

• Commercial drafts include proposed free alpha / later USD 19 educational bundle positioning, verified fee arithmetic, nonexclusive existing license rights, support/updates, refund and privacy copy, provenance matrix and explicit approval gates. These are not live or binding policies.
• Gumroad guidance follows official payout, fee and test-purchase documentation checked 2026-10-16. No self-purchase, identity/tax/bank records or agreement acceptance performed.
• Opened official payout settings in the existing Microsoft Edge installation per owner preference. Current tool connection cannot inspect/control those Edge tabs; authentication and verification remain owner-only portal actions. No credentials extracted or browser profile altered.
• Solver spec separates dipole source work, polarization/orientation, far-field channels, absorption, normalization and convergence. It defines twelve acceptance gates; no OLED solver implementation or validation is claimed.

Documentation-only change: local Markdown links, fee arithmetic and whitespace checks passed. Earlier numerical CI evidence applies to the unchanged preview source, not to the future solver.

[gmongell]

Dipole reference milestone implemented and verified

Commit: a40f407
CI: https://github.com/gmongell/Mathematica_python_TMM_OLEDPixel/actions/runs/38972288071

Added dipole_reference.py and docs/DIPOLE_FORMULATION.md: explicit exp(-i omega t), passive substrate/outgoing branch convention, homogeneous SI/solid-angle normalization and s/p patterns, vertical/parallel reflected Green-function source-work kernels, transformed propagating/infinite-evanescent integration, branch/resonance splits and refinement/error diagnostics. Unsupported lossless metal/ENZ cases and failed convergence raise errors. This is not a finite-stack OLED solver or external-efficiency calculation.

Requirement evidence

Requirement	Evidence
Homogeneous power/pattern normalization	test_homogeneous_power_si_normalization_and_scaling, test_homogeneous_sphere_and_hemisphere_normalization, test_homogeneous_directional_patterns_and_polarization
Homogeneous interface limit and orientation mixing	test_matched_interface_has_unit_rate_and_zero_spectrum, test_isotropic_pattern_is_orientation_power_average
Single-interface analytic mirror/sign benchmarks	test_perfect_mirror_rates_match_image_dipole, test_mirror_raw_spectrum_has_correct_parallel_sign
Dielectric/absorbing reference integration and near field	test_interface_matches_alternative_angular_integration, test_lossy_quasistatic_limit_and_orientation_ratio
Convergence, light lines and weak-loss resonance	test_refinement_and_diagnostics_for_lossy_interface, test_weak_dielectric_loss_approaches_lossless_branch_limit, test_weak_loss_surface_resonance_is_resolved
Explicit failures/unsupported inputs	test_underresolved_quadrature_fails_explicitly, test_invalid_interface_inputs, test_invalid_integration_controls

Local suite: 129 passed (73 new dipole + 56 preview). Remote: 129 passed in each of eight Windows/Linux Python 3.11-3.14 environments, 1,032 executions. Downloaded/parsed every JUnit report with no failures/errors/skips. Build/Twine/installed wheel smoke and runtime dependency audit passed; downloaded package hashes, SciPy inventory and source-archive formulation inclusion verified. Working tree clean.

Full finite-stack emission, far-field escape/channel accounting, measured material data and independent device-level validation remain future work. No merge, public release, paid listing or seller verification performed.

[gmongell]

Implemented and verified dispersive optical-data and orientation/resource foundations at 939b689.

• Native pinned refractiveindex.info YAML and licensed CSV ingestion, immutable provenance/checksums, offline cache and no extrapolation.
• Wavelength-dependent layer stack and actual thickness-dependent plane-wave R/T/A diagnostic, explicitly NOT EQE.
• Explicit monochromatic/broadband optical-rate EQE composition; equal-excitation independent fixed dipoles versus rapid reorientation; seeded isotropic sampling.
• Planar sheet/volume basis reuse, finite-pixel FDTD CPU/GPU workload/RAM planning, optional sustained-rate roofline lower bounds and measured existing CPU reference kernels.
• Reproducible reference at 550 nm/20 nm: synthetic lossy-interface source work 2.07018 horizontal, 7.75800 vertical, 3.96612 analytic isotropic. These are source-work ratios, not collected flux or EQE. No GPU execution/timing claim.

Validation: 215 local tests; exact-head Actions run 38975378348 passed all eight Windows/Linux Python 3.11-3.14 jobs (1,720 test executions), installed-wheel smoke and packaging/audit job. Downloaded all eight JUnit reports and checked zero failures/errors/skips, artifact SHA-256 hashes, wheel/source payload and clean CycloneDX inventories including PyYAML. Local independent buyer smoke, pip check, build and twine check also passed.

Scope remains research alpha: finite-emitting-layer source/collected-flux calculation, automatic stack-derived device EQE, full-pixel FDTD and GPU backend are NOT implemented. Source sampling/grid assumptions are NOT convergence certification. Seller verification, merge and paid publication remain unperformed; retain this PR as draft.