Verified Finite Contracts for Grassmannian EBR, All 80 Crystallographic Layer Groups, PSD Slavnov–Gram DPPs, Multi-Reference NOCI, Tequila-Compiled Grid DSLs, and Finite-Graph Ihara–Selberg Zeta Functions in Lean 4, Rust/PyO3 (maturin), and Python/SymPy


Abstract

Translating speculative research dialogues into trustworthy scientific software requires separating machine-checked finite mathematics and deterministic reference implementations from unmeasured physical, benchmark, or analytic claims. Following the audit of 64 fenced code blocks in data/chats/Gemini_Quantum-Causal-Grassmannian-Monograph-Review_20261118-1629.md (Porting Plan), this paper reports the tri-language execution of all nine workstreams (WS-1 through WS-9) in Lean 4 (3,631 Lake jobs, zero sorry), Rust + PyO3 (maturin) (122 unit tests compiled into the editable Python extension signals_numerical_adapter), and Python/SymPy (15 combined SymPy/PyO3 + Tequila tests passing).

Crucially, rather than duplicating existing infrastructure, the implementation directly reuses our companion Signals and Tequila stack:

  1. WS-1 (Dispersion Physics): Connects Gr24DispersionParameters to Signals.Proca.FourVector (+,-,-,-) and SI Signals.Units, proving Hermitian Pauli spinor determinant preservation \(\det K(k) = k_0^2 - k_x^2 - k_y^2 - k_z^2\) and exact \(z\)-Lorentz boost invariance in Lean 4 and SymPy while keeping physical Maxwell/Proca/evanescent identifications gated in Signals.Pending.

  2. WS-2 (Closed-Cycle Conditional Landauer & QFIM–Bures): Quarantines speculative vacuum-extraction prose in Signals.QuantumCooling and proves in Signals.Pending.InferenceThermodynamicCycleContract that even when single-shot conditional erasure is negative (\(S(A\mid M) < 0\)), closing the cycle via memory reset and dimensioned QFIM–Bures action dissipation \(\hbar\mathcal{L}_{\mathrm{QFIM}}^2 / (4\tau)\) obeys the Second Law (\(Q_{\mathrm{closed}} \ge k_B T \ln 2 \cdot N_{\mathrm{bits}} > 0\)).

  3. WS-3 (All 80 Crystallographic Layer Groups): Formalizes exact integer/half-lattice affine composition modulo \(\mathbb{Z}^2 \times \{0\}\) in Signals.LayerGroups and exhaustively verifies identity, closure, inverses, exact group orders, and 17-wallpaper projections across all 80 International Tables Vol. E layer groups (\(\mathcal{L}_1\text{–}\mathcal{L}_{80}\)) in Rust/PyO3, keeping G-MGS transport/QAHE claims decoupled.

  4. WS-4 (PSD-Verified Slavnov–Gram DPP): Proves \(2\times 2\) regularized Gram positivity, duplicate collapse, and inverse identities in Lean 4; implements exact Cholesky \(\log\det(X^\top X + \varepsilon I)\) and analytical gradients \(2 X G_\varepsilon^{-1}\) verified against SymPy symbolic differentiation (\(<10^{-11}\)); and empirically demonstrates on 8-Queens that raw-coordinate Gram-DPP spans only 5/12 \(D_4\) orbits without explicit orbifold quotienting (vs. 12/12 after \(D_4\) quotienting).

  5. WS-5 (QCI-GXC Multi-Reference NOCI & VRC-3): Proves VRC-3 smooth degeneracy evaluation at \(\Delta\varepsilon = 0\), the \(1/4\) QNG step from \(F_{\mu\nu} = 4 g_{\mu\nu}^{\mathrm{FS}}\), and exact 2-state NOCI variational lowering in Lean 4; implements Löwdin canonical orthogonalization and Jacobi eigensolving in Rust/PyO3; and cross-checks stretched \(\mathrm{H}_2\) (2-site Hubbard) 2-state PUHF vs. 3-state NOCI against exact FCI and tequila_simulator::expectation_pauli_hamiltonian_msb (\(|\Delta E| < 10^{-12}\), \(\langle\hat{S}^2\rangle = 0\)).

  6. WS-6 (Seeded EBR Harness): Separates continuous Sinkhorn + McWeeny + DPP relaxation from exact backtracking, showing transparently that unlearned continuous relaxation achieves Pass@1 = 0.20 on 4-Queens and 0.00 on 6- and 8-Queens (seed = 42), refuting unbenchmarked 100% Pass@1 claims without an upstream trained EBR energy network (arXiv:2611.08977).

  7. WS-7 (Compositional Grid DSL Reusing Tequila & multi_array_imaging): Implements typed GridOp primitives (rotations, flips, transpose, connected-component object extraction, cropping, recoloring, tiling, mirroring, bounded BFS synthesis) that reuse multi_array_imaging rotation-invariant PCA descriptors (cov_det_2d, aspect_ratio, bbox_fill_density, principal_angle_rad), layer_groups \(D_4 \subset \mathcal{L}_{55}\) matrices, and exact reversible TequilaGateOp (X, SWAP) quantum-circuit execution via tequila_simulator::simulate_circuit_msb.

  8. WS-8 & WS-9 (Textile Gate & Isolated Finite-Graph Ihara–Selberg): Encodes TextileObservationContract in Signals.Pending with Units.Length/Units.Force/Units.Duration, verifies 3D knitted centerline Klein residual \(= 0\) in SymPy, and proves the established finite-graph Ramanujan critical-circle equivalence (\(\lambda^2 \le 4q \implies |u|^2 = 1/q\)) verified symbolically on the 10-vertex 3-regular Petersen graph (\(q=2\)) without making any Riemann Hypothesis claim.


1. Cross-Module Reuse Architecture

        flowchart TB
    subgraph Existing["Existing Verified Owners Reused (No Duplication)"]
        E1["Signals.Proca & Signals.Units<br/>(FourVector, minkowskiDot, SI types)"]
        E2["Signals.QISM & Signals.Pending<br/>(slavnovDet2_pauli_exclusion, Pending gates)"]
        E3["tequila_simulator.rs & src/tequila<br/>(TequilaGateOp, simulate_circuit_msb,<br/>expectation_pauli_hamiltonian_msb)"]
        E4["multi_array_imaging.rs (Paper 17)<br/>(2D PCA cov_det_2d, aspect_ratio,<br/>bbox_fill_density, principal_angle_rad)"]
    end

    subgraph NewModules["New Workstreams WS-1..WS-9 (Lean 4 + Rust/PyO3 + SymPy)"]
        N1["WS-1 & WS-2: GrassmannianEBR.lean,<br/>QuantumCooling.lean, Pending.lean<br/>(Pauli spinor boost + Closed-Cycle Second Law)"]
        N2["WS-3 & WS-4: LayerGroups.lean,<br/>layer_groups.rs (80 groups), slavnov_dpp.rs"]
        N3["WS-5 & WS-6: multi_reference_noci.rs,<br/>ebr_benchmark.rs (seeded harness)"]
        N4["WS-7, WS-8, WS-9: arc_dsl.rs,<br/>TextileObservationContract, Ihara–Selberg"]
    end

    subgraph Bridge["PyO3 0.24 + maturin Editable Extension & SymPy Suite"]
        P1["signals_numerical_adapter (19 PyO3 functions + 2 pyclasses)<br/>+ tests/test_gebr_advancements_sympy_pyo3.py"]
    end

    E1 --> N1
    E2 --> N2
    E3 --> N3
    E3 --> N4
    E4 --> N4
    N2 --> N4
    N1 --> P1
    N2 --> P1
    N3 --> P1
    N4 --> P1
    

2. Phase A: Dimensioned Dispersion (WS-1) & Closed-Cycle Thermodynamics (WS-2)

2.1 Hermitian Pauli Spinor Map and Lorentz-Boost Invariance (WS-1)

In Signals.GrassmannianEBR (src/signals/Signals/GrassmannianEBR.lean), DimensionedWavevector lifts SI Frequency and Speed (Signals.Units) into Signals.Proca.FourVector \(k^\mu = (f/c, k_x, k_y, k_z)\) in fixed signature \((+,-,-,-)\) and maps it to the \(2\times 2\) Hermitian Pauli matrix:

\[\begin{split} K(k) = \begin{pmatrix} k_0 + k_z & k_x - i k_y \\ k_x + i k_y & k_0 - k_z \end{pmatrix}, \qquad \det K(k) = (k_0 + k_z)(k_0 - k_z) - (k_x^2 + k_y^2) = k_0^2 - k_x^2 - k_y^2 - k_z^2. \end{split}\]

Lean 4 proves pauliSpinorDet_eq_minkowskiDot, pauliSpinorDet_lorentzBoostZ (for any boost satisfying \(\cosh^2\eta - \sinh^2\eta = 1\)), dimensionedWavevector_minkowski_eq_dispersionInterval, and procaMode_pauliSpinorDet_eq_mass_sq (linking Signals.Proca.Mode.dispersion_relation to \(\det K(p) = m^2 > 0\)). Physical identification with Maxwell, Proca, or sub-nanometer evanescent laboratory modes remains gated in Signals.Pending.CWProcaEmissionHypothesis and Flat Light, Proca, and N-LIG Validation Plan.

2.2 Closed-Cycle Conditional Landauer & QFIM–Bures Accounting (WS-2)

The chat export (lines 522–695 and 1082–1221) proposed an unclamped single-shot Landauer–Bures work sum and interpreted negative values as net inference cooling. Signals.Pending.InferenceThermodynamicCycleContract closes the cycle with explicit SI energy units:

\[ Q_{\mathrm{closed}} = \underbrace{k_B T \ln 2 \cdot S(A \mid M)\,N_{\mathrm{bits}}}_{Q_{\mathrm{single\text{-}shot}}} + \underbrace{Q_{\mathrm{reset}}}_{\ge k_B T \ln 2 (1 - S(A \mid M)) N_{\mathrm{bits}}} + \underbrace{\frac{\hbar \mathcal{L}_{\mathrm{QFIM}}^2}{4 \tau_{\mathrm{step}}}}_{\Delta W_{\mathrm{Bures}} \ge 0} + W_{\mathrm{ctrl}} \ge k_B T \ln 2 \cdot N_{\mathrm{bits}} > 0. \]

Both Lean (InferenceThermodynamicCycleContract.closedCycle_secondLaw_bound, singleShot_vs_unentangled_control) and SymPy (test_ws2_closed_cycle_landauer_and_qfim_bures_dimensions_sympy) confirm that the signed conditional entropy \(S(A\mid M)\) cancels identically over the closed cycle, preserving the Second Law even when single-shot conditional erasure is negative.


3. Phase B: Exact 80 Crystallographic Layer Groups (WS-3) & Cholesky Slavnov–Gram DPP (WS-4)

3.1 Exhaustive Verification of All 80 International Tables Vol. E Layer Groups (WS-3)

Every crystallographic layer group \(\mathcal{L}_1\text{–}\mathcal{L}_{80}\) acts on \(\mathbb{R}^3 / (\mathbb{Z}^2 \times \{0\})\) via exact affine pairs \((R, \mathbf{t})\) with \(R = \operatorname{diag}(A, r_z)\), \(A \in \mathrm{GL}_2(\mathbb{Z})\), \(r_z \in \{-1, +1\}\), and half-lattice in-plane translations \(t_u, t_v \in \mathbb{Z}/2\mathbb{Z}\). Signals.LayerGroups proves associativity (LayerOp.comp_assoc), identity laws, determinant multiplicativity (LayerOp.det3_comp), exact generator relations (layerGroup_generator_relations), and the crystal-system partition \(2 + 5 + 11 + 30 + 16 + 16 = 80\).

In Rust (layer_groups.rs, exposed via py_verify_all_80_layer_groups and py_layer_group_orbit), BFS closure over exact integer/mod-2 generators verifies all 80 groups for identity, closure, inverses, exact quotient order, and projection onto all 17 2D wallpaper groups:

Layer Group

Symbol

Crystal System

Wallpaper Projection

Verified Quotient Order

Generic Orbit Size

Context

\(\mathcal{L}_1\)

p1

Triclinic / oblique

p1

1

1

Trivial baseline

\(\mathcal{L}_7\)

p112/a

Monoclinic / oblique

p2

4

4

Stockinette symmetry tag (WS-8)

\(\mathcal{L}_{19}\)

p222

Orthorhombic / rect.

p2mm

4

4

\(1\times 1\) Rib symmetry tag (WS-8)

\(\mathcal{L}_{37}\)

pmmm

Orthorhombic / rect.

p2mm

8

8

Garter stitch symmetry tag (WS-8)

\(\mathcal{L}_{55}\)

p4mm

Tetragonal / square

p4mm

8

8

\(D_4\) grid ops (WS-7)

\(\mathcal{L}_{61}\)

p4/mmm

Tetragonal / square

p4mm

16

16

Double-sided square layer

\(\mathcal{L}_{80}\)

p6/mmm

Hexagonal

p6mm

24

24

Hexagonal layer symmetry (decoupled from G-MGS QAHE)

3.2 PSD-Verified Cholesky Slavnov–Gram DPP and Raw vs. Orbifold Coverage (WS-4)

Signals.GrassmannianEBR proves that for \(a, b \ge 0\), \(c^2 \le ab\), and \(\varepsilon > 0\), \(\det G_\varepsilon = (a+\varepsilon)(b+\varepsilon) - c^2 > 0\) (gramDet2Reg_pos), duplicate vectors collapse \(\det G_0 = 0\) (gramDet2_duplicate_zero), and \(G_\varepsilon G_\varepsilon^{-1} = I_2\) (gram2Reg_inverse_identities). In Rust (slavnov_dpp.rs, exposed via py_evaluate_gram_dpp), Cholesky factorization computes \(\log\det(X^\top X + \varepsilon I_K)\) and exact analytical gradients:

\[ \frac{\partial}{\partial X}\,\log\det\!\left(X^\top X + \varepsilon I_K\right) = 2\,X\,(X^\top X + \varepsilon I_K)^{-1}, \]

matching both central finite differences (\(<10^{-7}\)) and exact SymPy symbolic differentiation (\(<10^{-11}\)). Against the exact 8-Queens oracle (92 boards in 12 \(D_4\) orbits), greedy MAP Gram-DPP (select_k = 12, \(\varepsilon = 10^{-4}\)) reveals a concrete geometric limitation missed by chat prose:

8-Queens Selection Pool

Exact Oracle Pool Size

Requested \(K\)

Distinct Solutions Selected

Distinct \(D_4\) Orbits Covered

Final \(\log\det G_\varepsilon\)

Raw coordinate embedding (orbifold_quotient=False)

92

12

12

5 / 12

-0.758290

\(D_4\)-orbifold quotient (orbifold_quotient=True)

12

12

12

12 / 12

-4.303787

Because rotated/reflected boards within the same \(D_4\) orbit can be nearly orthogonal in \(\mathbb{R}^{64}\), raw coordinate Gram-DPP selects orthogonal symmetry copies across only 5 of 12 fundamental orbits; full orbit diversity requires explicit orbifold quotienting prior to DPP selection.


4. Phase C: Multi-Reference NOCI (WS-5), Seeded EBR Harness (WS-6), & Tequila-Compiled Grid DSL (WS-7)

4.1 Multi-Reference NOCI vs. Exact FCI and Tequila Jordan–Wigner Pauli Expectation (WS-5)

Signals.GrassmannianEBR proves VRC-3 smooth degeneracy evaluation (vrcPairExcitationSmooth_at_degeneracy: \(V/(\Delta\varepsilon^2 + V^2)\big|_{\Delta\varepsilon=0} = 1/V\)), the \(1/4\) QNG step from \(F_{\mu\nu} = 4 g_{\mu\nu}^{\mathrm{FS}}\) (qngStepFromQfim_eq_quarter_fubiniStudy), and exact 2-state NOCI variational lowering (nociTwoState_variational_lowering).

In Rust (multi_reference_noci.rs, exposed via py_solve_noci_generalized and py_benchmark_stretched_h2_noci), Löwdin canonical orthogonalization with an explicit overlap-eigenvalue cutoff solves \(H c = E S c\) on the minimal-basis 2-site 2-electron stretched-\(\mathrm{H}_2\) Hubbard Hamiltonian and cross-verifies the resulting 3-state NOCI wavefunction directly against the 4-qubit Jordan–Wigner Pauli Hamiltonian in tequila_simulator::expectation_pauli_hamiltonian_msb:

\((t, U)\)

\(E_{\mathrm{RHF}}\)

\(E_{\mathrm{UHF}}\)

\(\langle\hat{S}^2\rangle_{\mathrm{UHF}}\)

\(E_{\mathrm{NOCI\text{-}2}}\) (PUHF)

\(E_{\mathrm{NOCI\text{-}3}}\)

\(E_{\mathrm{FCI}}\) (SymPy)

\(E_{\mathrm{Tequila\text{-}JW}}\)

\(|E_{\mathrm{NOCI\text{-}3}} - E_{\mathrm{FCI}}|\)

\((1.0, 3.0)\)

-0.500000

-0.666667

0.555556

-0.923077

-1.000000

-1.000000

-1.000000

< 1e-12

\((1.0, 5.0)\)

+0.500000

-0.400000

0.840000

-0.689655

-0.701562

-0.701562

-0.701562

< 1e-12

\((0.5, 4.0)\)

+1.000000

-0.125000

0.937500

-0.235294

-0.236068

-0.236068

-0.236068

< 1e-12

Both 2-state PUHF and 3-state NOCI restore exact singlet symmetry (\(\langle\hat{S}^2\rangle = 0.000000\)), while exhibiting the strict variational hierarchy \(E_{\mathrm{FCI}} = E_{\mathrm{NOCI\text{-}3}} < E_{\mathrm{NOCI\text{-}2}} < E_{\mathrm{UHF}} < E_{\mathrm{RHF}}\).

4.2 Seeded Continuous-Relaxation EBR Benchmark Harness (WS-6)

ebr_benchmark.rs (py_run_ebr_nqueens_benchmark) keeps continuous Sinkhorn + McWeeny + DPP relaxation strictly separate from exact backtracking (finite_constraint_solvers.rs) and logs deterministic seeded runs (seed = 42, max_steps = 120, step_size = 0.15):

Problem

Trajectories

DPP Weight

Valid Trajectories

Empirical Pass@1

Distinct Valid / Exact Oracle

Mean Halting Steps

Wall Time (ms)

4-Queens

10

0.00

2

0.20

2 / 2

3.20

0.046

4-Queens

10

0.02

2

0.20

2 / 2

72.50

1.255

6-Queens

12

0.02

0

0.00

0 / 4

80.42

2.530

8-Queens

20

0.02

0

0.00

0 / 92

38.15

6.994

This negative result on \(N \ge 6\) is scientifically important: unlearned continuous Sinkhorn + McWeeny + DPP relaxation gets trapped in non-integer/conflicting stationary basins, confirming that upstream EBR performance (arXiv:2611.08977) requires a trained energy landscape rather than hand-crafted quadratic relaxations.

4.3 Compositional GridOp DSL Reusing Tequila Circuits and multi_array_imaging PCA Invariants (WS-7)

arc_dsl.rs extends the 32-element symmetry finder in finite_constraint_solvers.rs with typed compositional primitives (Rotate90/180/270, FlipHorizontal/Vertical, Transpose, CropNonZero, ExtractLargestObject, ExtractSmallestObject, Recolor, Translate, Tile, MirrorHorizontalConcat, MirrorVerticalConcat, FillBoundingBox) and bounded BFS synthesis (py_synthesize_grid_program), reusing three existing repository layers:

  1. multi_array_imaging.rs (Paper 17): Every 4-connected GridObject reuses cosmic_ray_cluster_aspect_ratio, cosmic_ray_bbox_fill_density, cosmic_ray_covariance_det_2d, and cosmic_ray_principal_axis_angle_rad to provide rotation- and flip-invariant shape descriptors (verified to \(<10^{-12}\) across all 6 non-trivial \(D_4\) operations).

  2. layer_groups.rs (WS-3): grid_op_to_layer_op maps all spatial symmetries to exact \(D_4 \subset \mathcal{L}_{55}\) (p4mm) matrices.

  3. tequila_simulator.rs (Paper 44 / src/tequila): On any \(2^m \times 2^m\) grid encoded into \(2m\) MSB address qubits (\(0\text{..}m\) = row, \(m\text{..}2m\) = col), grid_op_to_tequila_gates compiles flips into Pauli X layers and transposes into SWAP(i, m + i) layers, and apply_grid_op_via_tequila_circuit executes them through tequila_simulator::simulate_circuit_msb with exact bit-for-bit equality against classical apply_op.


5. Phase D: Textile Gate (WS-8) & Isolated Finite-Graph Ihara–Selberg Equivalence (WS-9)

5.1 Dimensioned Textile-Observation Gate (WS-8)

In Signals.Pending, TextileObservationContract ties canonical stitch labels (stockinetteL7, rib1x1L19, garterL37) to their exact IT Vol. E layer-group numbers (\(\{7, 19, 37\}\)), SI yarn diameter/pitch (Units.Length), tensile load (Units.Force), relaxation time (Units.Duration), strain rate (Units.Frequency), minimum replicate count (\(\ge 3\)), control-fabric flag, and curl-residual uncertainty bound. SymPy (test_ws8_and_ws9_sympy_yarn_plucker_klein_and_petersen_ihara_zeta) confirms that a parametric 3D knitted loop centerline \(\mathbf{r}(s) = (s, A\sin(\omega s), B\cos(2\omega s))\) satisfies the Plücker Klein identity identically (\(p_{12}p_{34} - p_{13}p_{24} + p_{14}p_{23} = 0\)).

5.2 Isolated Finite-Graph Bass–Ihara Determinant & Ramanujan Critical Circle (WS-9)

Strictly decoupled from any claim regarding the Riemann zeta function \(\zeta(s)\), Signals.GrassmannianEBR.ihara_ramanujan_critical_circle_modulus_sq proves that for any adjacency eigenvalue \(\lambda\) of a finite \((q+1)\)-regular graph satisfying the Ramanujan bound \(\lambda^2 \le 4q\) (\(q > 0\)), the complex roots of \(1 - \lambda u + q u^2 = 0\) satisfy \(|u|^2 = 1/q\). SymPy evaluates the exact 10-vertex 3-regular Petersen graph (\(q = 2\)) determinant:

\[ \det\!\left(I_{10} - u A_{\mathrm{Petersen}} + 2 u^2 I_{10}\right) = (1 - 3u + 2u^2)(1 - u + 2u^2)^5(1 + 2u + 2u^2)^4, \]

confirming exact eigenvalues \(\{3^1, 1^5, (-2)^4\}\) and \(|u|^2 = 1/2 = 1/q\) for all 18 non-trivial roots.


6. Reproducible Tri-Language Verification Commands

# 1. Lean 4 full Signals + Pending + Tests verification (3,631 Lake jobs)
make signals_build

# 2. Rust native tests (122 passing) + PyO3/maturin editable release build
cd src/signals/numerical_adapter
TMPDIR=/workspaces/sustainablefactory/.tmp maturin develop --release
cargo test

# 3. Python + SymPy + PyO3 + Tequila integration test suites (15 passing)
pytest -q tests/test_gebr_advancements_sympy_pyo3.py tests/test_tequila_signals_backend.py

7. Provenance Metadata (BibTeX & JSON-LD)

@misc{chat_gebr_monograph_review_2026,
  author       = {Turner, Wesley and Google Gemini},
  title        = {Quantum-Causal Grassmannian Monograph Review: G-EBR, 80 Layer Groups, Slavnov-Gram DPP, Multi-Reference NOCI, ARC Grid DSL, and Ihara-Selberg Graph Zeta},
  year         = {2026},
  howpublished = {AI chat log (design provenance only; not scholarly validation)},
  note         = {Source: data/chats/Gemini_Quantum-Causal-Grassmannian-Monograph-Review_20261118-1629.md, lines 20-12155}
}
{
  "@context": "https://schema.org",
  "@type": "ScholarlyArticle",
  "headline": "Verified Finite Contracts for Grassmannian EBR, All 80 Crystallographic Layer Groups, PSD Slavnov-Gram DPPs, Multi-Reference NOCI, Tequila-Compiled Grid DSLs, and Finite-Graph Ihara-Selberg Zeta Functions in Lean 4, Rust/PyO3 (maturin), and Python/SymPy",
  "isBasedOn": [
    "data/chats/Gemini_Quantum-Causal-Grassmannian-Monograph-Review_20261118-1629.md",
    "docs/grassmannian-gebr-chat-porting-plan.md"
  ],
  "dateModified": "2026-11-18"
}