Canonical Inventory

9 Data Rooms, One Source of Truth

This page mirrors the current Genesis platform inventory and keeps technical language aligned across marketing, buyer variants, and due-diligence material. Version 3.7.0. 18,000+ files,909 filed sub-claims (280 utility + 629 provisional) claims, 32+ solvers.

Research Hub Buyer Dossiers

#	Data Room	Files	Claims	Status	Key Metric
1	Fab OS (YieldOS)	2,399	112	Verified	94.4% warpage reduction; FEM 0.028% vs Timoshenko; 11,000 FEM solves (cliff family disproven, 142 dead claims)
2	Packaging OS	3,597	150	5/5 Green	0.000% azimuthal effect (30 NLGEOM FEM)
3	Thermal Core	2,229	81 (Master Omnibus)	Strict Mode	68.9 C junction at 133 W/cm² (simulation design envelope)
4	Photonics	794	Multi-patent	Audited	40-55 dB practical RF isolation (TMM validated)
5	PFAS Remediation	2,100+	51	S-Tier	15/15 DFT-converged PFAS binding calculations
5.5	Critical Minerals	1,990+	48	S-Tier — REE selectivity ENTIRELY ASSUMED, validation pending	26.4% Kremser savings is built on an ASSUMED selectivity (alpha=4.0). No DFT support exists in the codebase.
6	Solid-State Battery	2,489	96	Consolidated — 2 retractions (Monroe-Newman misquote + Smart Fuse FTO downgrade)	Gyroid passes Monroe-Newman threshold at 60% porosity (corrected: G > G_Li/2 = 1.7 GPa, not the retracted 4× misquote)
7	Glass PDK	726	89 (8 patent families)	Reduced to Practice — 6.75M live BEM simulation rows, 15.92M-row long-format multiout corpus, 721K separately versioned strict buyer-regime additions, schema-complete strict model R²=0.9999966 / 0.0292% MAPE, and combined package signoff route across PROV 7/8/9.	6.75M live BEM rows + 15.92M long-format ML rows + 721K separately versioned strict additions; schema-complete strict model R²=0.9999966 / 0.0292% MAPE; package signoff suite; 4.0% mean abs error vs 6 in-scope coaxial-TGV papers; ONLY glass-specific PDK in existence
8	Isocompiler	490	72	247 tests pass	22-39 dB frequency-dependent isolation (adjoint r=1.0, 27 dB topology optimization)
9	Bondability	1,159	25 (13 provisional + 12 dependent)	v2.0 — all patent claims now implemented, JKR Maugis 1992 fix	Full GDS-to-yield physics chain; FNO honest R²=0.50 (was retracted at 0.8725); fab calibration required

PROV 1Verified

Fab OS (YieldOS)

Predicts and corrects wafer warpage during High-NA EUV lithography using biharmonic FEM (0.028% error, Richardson p=1.13), iterative learning control, and ROM surrogates (R²=0.975). 11,000 parametric FEM solves, 200 samples/point. The 'Physics Cliff' family (142 claims, 2.49x peak) was scientifically DISPROVEN in the 10x audit — 3D Richardson-extrapolated FEM did not reproduce the 2D singularity, all 142 cliff claims publicly retracted as dead.

Directory

PROV_1_FAB_OS/

Files / Claims

2,399 / 112

Key Metric

94.4% warpage reduction; FEM 0.028% vs Timoshenko; 11,000 FEM solves (cliff family disproven, 142 dead claims)

Solver Roots

PROV_1_FAB_OS/04_Source_Code/
PROV_1_FAB_OS/05_Models/

Evidence Artifacts

11,000 parametric FEM solves, 200 samples/point
0.028% FEM error at N=320, Richardson p=1.13, ±0.71% bound
13.2x design-around gap (Genesis ILC 90.5nm vs Kitchen Sink 1198nm)

View Repository

Representative evidence chart for Fab OS (YieldOS).

PROV 25/5 Green

Packaging OS

Solves rectangular substrate warpage for advanced packaging, with Kirchhoff-von Karman nonlinear plate solving and inverse-design compiler support.

Directory

PROV_2_PACKAGING_OS/

Files / Claims

3,597 / 150

Key Metric

0.000% azimuthal effect (30 NLGEOM FEM)

Solver Roots

PROV_2_PACKAGING_OS/genesis_packaging/
PROV_2_PACKAGING_OS/tests/

Evidence Artifacts

~550 verified task IDs
500 SHA-256 hashes
218 MB GDSII output

View Repository

Representative evidence chart for Packaging OS.

PROV 3Strict Mode

Thermal Core

Validates self-pumping Marangoni cooling in binary fluids, eliminating mechanical pump dependence for high-power electronics thermal management.

Directory

PROV_3_THERMAL_CORE/

Files / Claims

2,229 / 81 (Master Omnibus)

Key Metric

68.9 C junction at 133 W/cm² (simulation design envelope)

Solver Roots

PROV_3_THERMAL_CORE/02_CODEBASE/
PROV_3_THERMAL_CORE/genesis_thermal_api.py

Evidence Artifacts

320 MB computational evidence
49 converged OpenFOAM cases
100/100 Monte Carlo stability

View Repository

Representative evidence chart for Thermal Core.

PROV 4Audited

Photonics

Combines glass firewall, Zernike substrate optimization, low-index optical lattice, and smart substrate mechanics into one photonics stack.

Directory

PROV_4_PHOTONICS/

Files / Claims

794 / Multi-patent

Key Metric

40-55 dB practical RF isolation (TMM validated)

Solver Roots

PROV_4_PHOTONICS/02_CODEBASE/
PROV_4_PHOTONICS/05_API/

Evidence Artifacts

CLI: 9 commands
REST API: 14 endpoints
Tier-1 evidence locker

View Repository

Representative evidence chart for Photonics.

PROV 5S-Tier

PFAS Remediation

Designs Janus nanoparticle scaffolds for irreversible PFAS capture using DFT-validated binding energies, proving activated carbon cannot permanently sequester PFAS.

Directory

PROV_5_SMART_MATTER/

Files / Claims

2,100+ / 51

Key Metric

15/15 DFT-converged PFAS binding calculations

Solver Roots

PROV_5_SMART_MATTER/02_SOURCE_CODE/

Evidence Artifacts

15 DFT-converged binding energy calculations
730+ Janus scaffold library
Irreversibility ratio > 1.78x vs GAC

View Repository

Representative evidence chart for PFAS Remediation.

PROV 5.5S-Tier — REE selectivity ENTIRELY ASSUMED, validation pending

Critical Minerals

REE selectivity figures in this room are ENTIRELY ASSUMED — no DFT support exists in the codebase. The Kremser CapEx narrative (26.4% savings at alpha=4.0) is built on top of this assumed selectivity number. Real DFT validation is a $50K+ followup experiment. Treat all REE numbers as optimistic placeholders until validated against measured separation data. The PFAS-side DFT (PROV 5a, 15/15 converged) is the validated half of PROV 5; the REE side is not.

Directory

PROV_5_SMART_MATTER/

Files / Claims

1,990+ / 48

Key Metric

26.4% Kremser savings is built on an ASSUMED selectivity (alpha=4.0). No DFT support exists in the codebase.

Solver Roots

PROV_5_SMART_MATTER/02_SOURCE_CODE/

Evidence Artifacts

26.4% CapEx savings at alpha=4.0 (ENTIRELY ASSUMED — no DFT support, no measured selectivity)
4/15 PFAS candidates survive aqueous stability screening
ML surrogate R² > 0.95 (trained on assumed selectivity, not measured)
Real validation cost: ~$50K wet-lab experiment

View Repository

Representative evidence chart for Critical Minerals.

PROV 6Consolidated — 2 retractions (Monroe-Newman misquote + Smart Fuse FTO downgrade)

Solid-State Battery

Uses gradient-stiffness gyroid architecture to suppress dendrites while preserving ionic transport and manufacturability via a four-pillar design. Key discovery: gyroid geometry passes Monroe-Newman mechanical stability criterion (G_eff=25.9 GPa) at 60% porosity. Two public retractions: (1) Monroe-Newman safety factor was 4× misquoted in 4/8 source files (now corrected — the correct threshold is G > G_Li/2 = 1.7 GPa, not G > 2*G_Li = 6.8 GPa); the gyroid still passes at 1.8× margin. (2) Smart Fuse FTO downgraded from GREEN to YELLOW after self-test failed at 0.647 < 1.0 mS/cm conductivity threshold. See docs/RETRACTION_REGISTRY.md for full audit trail.

Directory

PROV_6_SOLID_STATE/

Files / Claims

2,489 / 96

Key Metric

Gyroid passes Monroe-Newman threshold at 60% porosity (corrected: G > G_Li/2 = 1.7 GPa, not the retracted 4× misquote)

Solver Roots

PROV_6_SOLID_STATE/02_SOURCE_CODE/
PROV_6_SOLID_STATE/tests/

Evidence Artifacts

Gyroid 60% porosity Monroe-Newman pass (G_eff=25.9 GPa, corrected threshold G > G_Li/2 = 1.7 GPa)
Retraction 1: Monroe-Newman 4× misquote across 4/8 files — now corrected, gyroid still passes at 1.8×
Retraction 2: Smart Fuse FTO downgraded GREEN→YELLOW (0.647 < 1.0 mS/cm conductivity self-test failed)
20 ns A100 GROMACS transport; 204-test suite; CANONICAL_VALUES.json reference
Full audit trail: docs/RETRACTION_REGISTRY.md

View Repository

Representative evidence chart for Solid-State Battery.

PROV 7Reduced to Practice — 6.75M live BEM simulation rows, 15.92M-row long-format multiout corpus, 721K separately versioned strict buyer-regime additions, schema-complete strict model R²=0.9999966 / 0.0292% MAPE, and combined package signoff route across PROV 7/8/9.

Glass PDK

The ONLY glass-specific PDK in existence. Compiles requirements into RF/TGV outputs, S2P artifacts, GDS export specs, and package reports. Data moat: 6.75M live BEM simulation rows, 15.92M long-format ML rows, and 721K separately versioned strict buyer-regime additions calibrated to 4.0% mean abs error vs 6 in-scope coaxial-TGV papers. The Sprint 22 schema-complete strict lane adds R²=0.9999966 / 0.0292% MAPE on a clean group-disjoint test. `POST /v1/chiplet-suite/package-signoff` combines PROV 7 RF, PROV 8 isolation, and PROV 9 bondability into one provenance-bound route. 995 active curated unique S2P assets are tracked in a 2,480-entry registry with 0 parse failures, including unpublished-glass subsets.

Directory

PROV_7_GLASS_PDK/

Files / Claims

726 / 89 (8 patent families)

Key Metric

6.75M live BEM rows + 15.92M long-format ML rows + 721K separately versioned strict additions; schema-complete strict model R²=0.9999966 / 0.0292% MAPE; package signoff suite; 4.0% mean abs error vs 6 in-scope coaxial-TGV papers; ONLY glass-specific PDK in existence

Solver Roots

PROV_7_GLASS_PDK/glass_pdk/
PROV_7_GLASS_PDK/tests/

Evidence Artifacts

6.75M live BEM rows + 15.92M-row BEM multiout corpus + 721K separately versioned strict additions (the moat — calibrated to 4.0% mean abs error vs 6 in-scope coaxial-TGV papers)
IEEE calibration set: Sukumaran ECTC14, Watanabe ECTC19, Shorey JMS16, Tummala JEP20, Hwang TMTT17
PITCH_CALIBRATION=0.67 reduces BEM error from 32.6% → 3.5%
995 active curated unique S2P assets in a 2,480-entry registry with 0 parse failures
Sprint 15/16/17/19 strict-additions manifest: 721K separately versioned strict non-fallback rows across six regimes
Combined package signoff route: POST /v1/chiplet-suite/package-signoff across PROV 7 RF, PROV 8 isolation, and PROV 9 bondability
605 analytically screened design points; 867-test suite; HFSS/Sigrity export paths

View Repository

Representative evidence chart for Glass PDK.

PROV 8247 tests pass

Isocompiler

Synthesizes EM isolation structures for multi-die chiplets using FDTD/adjoint/Nevergrad workflows and outputs foundry-ready artifacts.

Directory

PROV_8_ISOCOMPILER/

Files / Claims

490 / 72

Key Metric

22-39 dB frequency-dependent isolation (adjoint r=1.0, 27 dB topology optimization)

Solver Roots

PROV_8_ISOCOMPILER/src/isocompiler/
PROV_8_ISOCOMPILER/tests/

Evidence Artifacts

Synthesis families: via/EBG/metasurface/inverse
KLayout DRC runset integration
Palace and Meep adapters

View Repository

Representative evidence chart for Isocompiler.

PROV 9v2.0 — all patent claims now implemented, JKR Maugis 1992 fix

Bondability

Models CMP, contact mechanics (p_void fixed via JKR Maugis 1992), voiding, thermal stress, Cu2O delamination risk, and yield to predict hybrid bonding outcomes from layout-level inputs. Neural surrogate (experimental MLP) — FNO honest R²=0.50 pixel / R²=0.63 image-max on the 100K realistic-distribution validation set; the previously claimed R²=0.8725 was retracted (narrow-distribution overfit on 10K samples). Gradient-compensated fill, multi-objective optimizer, KLA Archer integration. TSMC deployment kit with 3 die layouts. Warning: correlation_length parameter dominates yield by ~100x and requires fab calibration data.

Directory

PROV_9_BONDABILITY/

Files / Claims

1,159 / 25 (13 provisional + 12 dependent)

Key Metric

Full GDS-to-yield physics chain; FNO honest R²=0.50 (was retracted at 0.8725); fab calibration required

Solver Roots

PROV_9_BONDABILITY/bondability/
PROV_9_BONDABILITY/tests/

Evidence Artifacts

Spectral FFT contact solver (p_void fixed via JKR Maugis 1992)
Murphy negative-binomial yield model (correlation_length calibration required)
Neural surrogate + gradient-compensated fill + KLA Archer integration + TSMC 3-die kit

View Repository

Evidence

Dedicated chart assets are tracked in-room. Core evidence remains available via the documented solver roots and artifacts.