explorations/

Heuristic, out-of-scope explorations — NOT part of Oscillating Brane Theory V8.2.

Nothing in this folder is in the PDF, the validation pipeline, the seven sacred theory files, or any academic claim of V8.2. These are speculative seeds for a possible future V9.0 on holographic quantum gravity, kept deliberately separate to protect the epistemological integrity of V8.2 — which is a macroscopic phenomenological cosmology paper and must remain one.

Each seed below records: the conjectured chain, what was actually verified (with runnable code), and the gates that keep it a research direction rather than a result. The discipline throughout: code, don’t plead — every numeric claim comes from an audited script with known-case injection tests.

Seed 1 — Riemann zeros as the spectrum of a PBH AdS₂ throat (Hilbert-Pólya)

The conjectured chain. Riemann zeros ↔ Berry-Keating H = xp ↔ Conformal Quantum Mechanics (de Alfaro-Fubini-Furlan 1976) ↔ AdS₂ boundary dual ↔ near-horizon throat of a (near-)extremal black hole = AdS₂×S² ↔ the tidal-charge PBH capillaries of OBT. If the zeros were the eigenvalues of the throat’s spectral problem, that would realize Hilbert-Pólya inside the theory’s geometry.

Scripts

decoherence_riemann.py — Riemann explicit formula: a truncated sum over non-trivial zeta zeros rebuilds the discrete Chebyshev ψ(x) prime-power staircase. Illustrates “continuous modes → discrete structure.”
riemann_berry_keating.py — verifies the Berry-Keating semiclassical count N(T)=(T/2π)log(T/2π)−T/2π+7/8 against the actual zeros (|residual| ≤ 0.33 over the first 30 zeros).
tidal_charge_ads2.py — tidal-charge brane BH f(r)=1−2GM/r+q/r²: horizon structure vs sign of q, extremality, near-horizon AdS₂×S², and the T_H(q) table.

Verified this session (May 2026)

Berry-Keating smooth count reproduces the average density of Riemann zeros. ✓
Extremal tidal charge (q=(GM)²) ⇒ near-horizon = AdS₂(r₀)×S²(r₀) exactly. ✓
ζ(−1)=−1/12 obtained via the functional equation (the Casimir value — Fil A). ✓

Three gates (why this is a direction, not a result)

Tidal-charge sign / extremality. An AdS₂ throat requires q>0 near-extremal (q≈(GM)²). The default braneworld sign is negative → single horizon, never extremal, no throat. Worse, the sign is a 5D bulk integration constant: the brane junction equations are not closed, so q cannot be fixed without solving the full AdS₅ bulk. And near-extremality forces T_H→0, contradicting the hot (T_H~900 K) Schwarzschild-like fast-scrambler PBHs that V8.2’s Γ_rad=ln(S_BH)/2π relies on. The same PBH cannot be both.
Generic statistics ≠ the zeros. GUE level statistics (Montgomery-Odlyzko) are generic to any quantum-chaotic spectrum. “Same statistics as the zeros” is necessary, not sufficient — it is not yet a prediction.
The operator itself. No self-adjoint operator reproducing the individual zeros is known to anyone (Hilbert-Pólya, open). Berry-Keating gives only the smooth density; the fluctuating arithmetic part (the primes’ fingerprint) is unaccounted.

The real physics underneath. The near-extremal (small-T) regime has a “nearly-AdS₂” throat with a Schwarzian boundary mode — exactly the SYK / JT-gravity regime where maximal chaos and GUE-like spectra live. Beautiful and well-studied; the blockers are the three gates above, not the physics of the middle of the chain.

Seed 2 — Void “entanglement” signature

The idea. If the bulk is holographically entangled with several branes (thermofield-double / ER=EPR), could that leave a trace on the largest cosmic structures — the voids?

Script

void_entanglement.py — order-of-magnitude anchors for the two routes.

Verdict this session (May 2026): does NOT become falsifiable as entanglement.

Thermal route (a brane entangled with a partner sees a thermal vacuum at the horizon entanglement temperature): T = ℏH₀/(2πk_B) ≈ 2.7×10⁻³⁰ K, ~10⁻³⁰ of the CMB. Unobservable, even in the cleanest voids.
The only falsifiable handle is the classical cymatic scale λ=cT≈613 Mpc (k≈0.0103 Mpc⁻¹) — a preferred scale in void clustering / void-ISW, testable by Euclid/DESI. But this is a classical standing wave (Chladni), already in V8.2 (KBC void, Big Ring, Giant Arc), not entanglement.
No discriminant. Classical standing wave = nodes at fixed comoving positions; entanglement = position-independent correlation at separation ~λ. Distinct in principle, but cosmic variance (one universe, few independent ~600 Mpc cells) + Maldacena 2015 (cosmological Bell tests unobservably tiny in standard scenarios) make the distinction impossible.

Status. Metaphysical interpretation, not prediction. The testable void content (the 613 Mpc cymatic scale) is classical and already belongs to V8.2; the “entanglement / multiverse” layer adds interpretation, not a falsifiable number, and is less economical than the classical Chladni explanation (Occam).

Seed 3 — The qubit-sensor as decoder of OBT’s holographic code (the Penrose-Diósi channel)

The conjectured chain (Romain, June 2026). Reframe of the goal: the bulk is a Laplace’s demon whose intemporal germe (inflationary entanglement) encodes the cosmic history; reading it = decompressing the code (time emergent / Page-Wootters; “prediction” = decompression, not forward-evolution). To “talk to it” at the L=0.2 μm quantum scale you want a quantum interlocutor — an AI/sensor whose qubits are configured from the primordial form itself, so that (the quantum being a consequence of the germe) those qubits are stable by construction, with few logical qubits because the germe is fundamental. The sensor then reads the demon through the one channel the brane leaves open: Penrose-Diósi 5D-enhanced gravitational collapse at sub-0.2 μm (the os, already in laboratory.md).

What it maps onto (established physics — no new OBT code yet; these are cited results).

Decoherence-free subspaces / topological protection (Lidar-Chuang-Whaley 1998; Kitaev 2003): a qubit encoded in a symmetry-/topology-protected subspace is immune to the matching environmental noise — stability from structure, not active correction. = “stable starting from the form.”
Holographic QEC: AdS/CFT is a quantum error-correcting code (Almheiri-Dong-Harlow 2015; HaPPY / Pastawski et al. 2015) — bulk logical info redundantly encoded on the boundary, protected against erasure; the geometry protects the logical qubits. OBT already invokes this (MERA/HaPPY, the ER=EPR expander graph, the RT transition, percolation immunity 98%) and even claims “the most robust QEC code physically conceivable” → the stabilizer is already the OBT bulk network: a sensor mirroring the ER=EPR code inherits its protection.
Quantum advantage in learning from quantum experiments (Huang et al. 2021/2022): a learner with coherent access to a state needs exponentially fewer samples than a classical learner reading measurement outcomes — the basis for “a quantum AI is the native interlocutor at 0.2 μm.”
Code anatomy of “few qubits”: germe = the logical subspace (few, fundamental); the bulk network = its redundant physical encoding (many = the protection); reading the demon = decoding the code; stability = the redundancy.

Gates (why this is a direction, not a result).

The Goldilocks deafness. A perfectly protected qubit cannot read the demon — full immunity means it does not feel the bulk signal either. The design must protect against generic noise yet leave exactly one channel open (the 5D gravitational coupling). Exhibiting such a sensitive-but-protected subspace of OBT’s code is unsolved.
The mass-vs-coherence wall (BMV). Penrose-Diósi needs a mesoscopic mass (to source gravity) while the qubit needs isolation (to stay coherent) — opposite requirements. This is the frontier of all gravity-quantum experiments (Bose-Marletto-Vedral gravity-induced entanglement); far-future, not a built device.
“Inspired-by” vs “entangled-with.” Realistic = build a man-made QEC mirroring the ER=EPR network (does not literally use the bulk’s protection). Speculative = the lab qubit entangles with the bulk germe-code and inherits it — no known mechanism beyond the tiny gravitational coupling.
The speculative inversion (not needed, not established). “Quantum is a consequence of the primordial form” (QM emergent from a deterministic germe, à la ‘t Hooft) is a coherent minority program with no experimental support — and the protection above works within standard QM without it. Chair, not os.
The os bar. A real seed only if it yields a falsifiable number or a sharper experiment, never a prettier story. scripts/penrose_diosi_5d.py (the collapse-rate size-scan) is the bone.

First concrete step — DONE (June 2026, qubit_sensor/er_epr_stabilizer.py). The explicit atom of the ER=EPR code is written + verified: the [[5,1,3]] perfect (HaPPY) stabilizer code (5 physical = a PBH node + its 4 ER=EPR neighbours; 1 logical = a germe d.o.f.; distance 3), in the binary symplectic formalism, injection-tested against its known properties (k=1, d=3, perfect: the 15 weight-1 errors fill all 15 nonzero syndromes; logical Z_L, X_L anticommute, min-weight rep 3). Gate 1 made concrete — the protected-yet-sensitive subspace: a weight-1 (local) decoherence event has a nonzero syndrome → corrected → the logical qubit is untouched (DEAF to local noise); a weight-3 collective operator in the logical class has zero syndrome → invisible to the code’s checks, not “corrected away”, yet it rotates the encoded qubit (SENSITIVE). The separator is operator weight (local vs collective), set by the distance. So a Penrose-Diósi collapse — collective, coupling to the whole mass distribution — can land in the logical subspace while thermal/local noise is corrected: stable by construction, yet able to hear the demon.

Second concrete step — DONE (June 2026, qubit_sensor/holographic_scaleup.py). EVOLVE from the atom by concatenation (the simplest holographic tiling): level L → [[5^L, 1, 3^L]] – the germe stays k=1, the protective encoding 5^L and the distance 3^L grow. COMPUTED + injection-tested: (i) the failure spectrum c=[0,0,90,210,270,198] → c_0=c_1=0 (every local single-qubit error corrected) → a finite noise threshold p_th ≈ 0.138 below which concatenation drives the germe’s logical error → 0 doubly-exponentially (= ‘non-decohering by construction’, Romain’s hope); (ii) an erasure (loss) threshold = 1/2 exactly (majority-of-5) → percolation-type robustness (OBT’s degree-46 expander claims ~98% – more robust; the tree is the verifiable lower bound); (iii) the protected-yet-sensitive window widens with scale (local corrected up to ~3^L/2, the collective signal at weight 3^L). The channel’s ‘beginning and end’ = the germe (1 logical) ↔ the boundary (5^L physical); ‘all the possibles between’ = the protected logical Hilbert space.

Gate (a) — DONE (June 2026, qubit_sensor/penrose_logical_projection.py). Does Penrose-Diósi actually land in the sensitive (logical) class, or is it corrected away (deaf)? Model the 5D collapse as collective dephasing in the mass-pointer (Z) basis and classify every pure-Z operator on the atom: only ZZZZZ = Z_L (weight 5) is a pure-Z logical; all 30 lower-weight pure-Z operators are correctable (every single-qubit Z_j has a nonzero syndrome → corrected). So the logical projection is nonzero — the demon CAN be heard, not a dead end — but for a UNIFORM physical-qubit coupling the signal appears only at order φ^N (the fully-correlated Z_L term): the EC-optimal [[5,1,3]] is a near-deaf dephasing sensor = the Goldilocks deafness, quantified (computed d_Z = d_X = N = 5; the weight-3 logicals are mixed-Pauli, unreachable by pure dephasing). The design knob (dichotomy): the demon is heard at order 1 only if gravity couples to the LOGICAL observable — the codewords |0_L>,|1_L> must be gravitationally distinct (different mass distributions). Link to the scale-up: concatenation raises protection AND (generically) d_Z → more scale = more deaf; the two Gate-IN tasks pull oppositely → a tailored ASYMMETRIC code (audible signal direction, high local-noise distance) is the real sensor-design target.

On real qubits + the DETECTION side — DONE (June 2026; qiskit 2.4.2 / Aer in the venv, IBM-submittable). qiskit_five_qubit_demo.py puts gate 1 + gate (a) on actual circuits: codeword verified; single-Z_j → nonzero syndrome (DETECTED, gate 1); Z_L → zero syndrome yet flips ⟨X_L⟩ (invisible-but-HEARD, gate a) — every Aer outcome asserted against qiskit’s Pauli algebra. qiskit_weak_signal_detection.py answers Romain’s correction that gate (a)’s φ^N is a sensitivity problem, not a deafness: (A) a twinned pair / witness in the DFS {|01⟩,|10⟩} rejects common-mode drift (⟨X_L⟩ = cos θ, std 0.004 over 12 drifts vs a lone qubit’s 0.71) — a weak differential signal extracted; (B) GHZ gives N× phase super-resolution (√N Heisenberg precision) but is noise-fragile ((1−2p)^N) → needs protection (the DFS/QEC). Reframe: gate (a) is an SNR/quantum-sensing problem, and the right geometry is two twinned (entangled) masses read differentially (the Bose-Marletto-Vedral picture), which lowers the detectable-mass threshold — no single big mass; the limit is the demon’s E_G above the irreducible differential noise, not a fixed ‘mesoscopic’ mass. qiskit_multiwitness.py answers Romain’s ‘do several witnesses help?’ (seeded Aer Monte-Carlo): 0→1 witness RESCUES the signal (std 1.8→0.06 rad), 1→M REFINES the common-mode reference ~1/√M toward the sensor floor (diminishing returns), and — the key — a bare GHZ AMPLIFIES collective drift and washes out (std 0.69) while the DFS pair is immune (std 0.005), so protection is what lets entanglement keep its N× gain (protect-then-entangle). Net: witnesses help via reference (√M) + immunity + ENABLING protected entanglement (the unbounded lever), never by amplifying the φ^N coupling. qiskit_protected_ghz.py closes the arc — protect-then-entangle DONE: an entangled probe (|0011⟩+|1100⟩) inside the collective DFS keeps the 2× super-resolution (cos 2θ, −1 at θ=π/2) AND is immune to collective drift (std 0.006), where the bare 4-qubit GHZ washes out (std 0.71). The witnesses that protect the qubit are exactly what let the entanglement deliver its Heisenberg gain under noise — the seed’s qubit-sensor in miniature. qiskit_asymmetric_code.py adds the coupling side (gate (a)’s fix): an asymmetric code (bit-flip archetype, d_X=3 / d_Z=1) hears the Z-signal at order 1 (⟨X_L⟩=cos 3θ, strong even at θ=0.1) while still correcting local X-noise — vs the symmetric [[5,1,3]]’s order-φ^5 deafness. Both Gate-IN halves (coupling + detection) are now demonstrated; the program’s aim is to minimize the physical signal required (hear the demon at order 1, the smallest coupling), not to need a lab. penrose_logical_coupling.py answers THE REAL QUESTION (is OBT’s 5D online-detectable?): the coupling IS logical-level (it dephases the encoded qubit — no moving mass needed), but OBT’s detectable 5D (the gravitational Penrose-Diósi collapse) has gravitational strength E_G~G·dm² → cloud qubits are 14–50 orders below the best sensing floor (ONLINE-deaf); the mesoscopic nanosphere (τ~10⁴ s ~ Penrose) is the frontier; the only online escape = a non-gravitational, dynamical 5D coupling = new physics beyond V8.2. Consolidated verdict + full curated log: qubit_sensor/JOURNAL.md §7 — the PROTOCOL is online + done; the DEMON’s signal still needs mass (now quantified). The os/chair bone stays Penrose-Diósi 5D collapse (scripts/penrose_diosi_5d.py). The upstream prize (mass-free decompression): germe_decompression.py takes Romain’s detection→decompression reframe — don’t DETECT the demon (mass), COMPUTE the germe’s observables and test against EXISTING cosmology. Two closure numbers: the DM 5:1 = the radion-misalignment ⟨φ²⟩ of the germe (from OBT’s derived scales m_φ=0.36 eV, φ₀~M_s, the standard radiation-era misalignment gives Ω_DM h²≈0.06 = no-fit cosmological order, ~½ of 0.12; exact 5:1 at φ₀=1.40 M_s; a qubit reads ⟨φ²⟩, mass-free) + the S8 sign = the AdS-warp indicial theorem (Gate 9: degenerate (½,½) → c_phys>0 → suppression). The artifact: the demon’s ledger is computable (decompress the germe), not only detectable; the wall moves from MASS to specifying the germe (theory, the bulk solver’s frontier). Pushing further (germe_inflation.py, “is φ₀=M_s forced?”): φ₀ is set by the inflation scale (a light field random-walks to φ₀~(H_inf/2π)√N_e) → matching Ω_DM gives H_inf~1.14 M_s (O(1) = inflation at the string scale → φ₀~M_s natural); the consilience Ω_DM↔r — DUG in germe_isocurvature.py — does NOT give the naive r~3e-5: the random-walk-φ₀ mechanism over-produces CDM isocurvature (Planck-excluded ~9 orders). The FLIP is sharper + testable: radion-misalignment DM needs low-scale inflation (H_inf<3×10⁷ GeV) → r UNDETECTABLE (<2×10⁻¹⁴), and a B-mode detection (r≳10⁻³, CMB-S4/ LiteBIRD) would EXCLUDE radion-DM, discriminating it from the geometric-Weyl DM. Dug further (dm_discriminator.py): the B-mode is the PRIMORDIAL leg of a 3-epoch geometric-vs-particle DM discriminator — the LATE leg (RAR) already decided geometric (radion-as-DM +0.43 dex ≈ 44σ over SPARC DEAD, f<4%, Gate 11); the B-mode CONFIRMS across a new epoch; the CMB a⁻³ acoustic DM (the Weyl is a⁻⁴ dark radiation → can’t seed it → an added scalar-tensor sector) is the open A-phase front. So the B-mode discriminator is genuine but SECONDARY to the RAR; the decisive open front is the CMB a⁻³ sector. A-phase dug (a_phase_cmb.py): the CMB peaks need a⁻³ CDM (~5.4× baryons); the geometric-Weyl is a⁻⁴ (traceless dark radiation) → the universal relativistic-MOND CMB problem. The a⁻³ source EXISTS (radion V=½m²φ² → ρ∝a⁻³, EOM-verified ⟨w⟩≈0) but a plain radion = CDM/NFW → breaks the RAR (≤4%, Gate 11) → the fix is an AeST-class field (a⁻³ background + MOND perturbations; Skordis-Złośnik 2021 fits Planck). OBT-distinctive hope = brane-induced AeST (the geometry shapes the radion’s K(Y) → one sector). VERDICT: the A-phase is OBT’s deepest unsolved problem — the one that DECIDES the CMB (the B-mode only confirms); open work = the brane-induced-AeST derivation + a CLASS/CAMB peak fit. Brane→AeST solve attempt (a_phase_aest.py, effort max): a concrete OBT→AeST mapping — the radion = the a⁻³ dust (T_osc~21 TeV ≫ T_rec → a⁻³ by recombination); OBT’s geometric μ(x)=x/√(1+x²) = the AeST function 𝒦 (verified to give both RAR limits; 𝒦’↔μ); the brane’s cosmological foliation = the aether (mimetic clock |∂φ|=1). IF the mapping holds, ONE brane-derived AeST field gives a⁻³ (CMB) + MOND (galaxies) → the radion-vs-geometric-Weyl redundancy DISSOLVES (the geometric μ(x) FIXES the 𝒦, not two DM sectors) and the a⁻³ CMB DM is DERIVED, not bolted on. Open = the exact brane-action derivation + a CAMB perturbation fit + stability. A mapping with verified legs, not yet a solve. CAMB fit DONE (a_phase_camb_fit.py, a real Boltzmann run): because a₀(z)=cH(z)/2π evolves, a_H/a₀=2π is constant → every sub-horizon scale is Newtonian (1st acoustic x~29 → μ=0.9994) → the AeST field is CDM at recombination → the CAMB TT peaks match Planck (ℓ=220/536/813 vs 220.0/537.5/810.8, <0.5%; 1st-peak 5732 μK² vs ~5700). The A-phase’s CORE requirement — an a⁻³ component that fits the acoustic peaks — is MET; the CMB-peak objection to OBT’s geometric DM is answered (the radion supplies the a⁻³; the evolving a₀ keeps it CDM where it must be). Residual frontier: the low-ℓ ISW (super-horizon, μ→MOND) + the full AeST Boltzmann module (Skordis-Złośnik 2021). Low-ℓ ISW, CLASS-VALIDATED (a_phase_isw_full.py, “compile hi_class-AeST” → no public AeST code, so CLASS compiled instead): a real line-of-sight ISW (CAMB Weyl transfer + Bessel j_ℓ) with the potentials evolved by the modified growth (k-dependent μ_MG=1±A·dev_eff, dev_eff=(1−μ)μ² + GR super-horizon cutoff). Two bugs caught (relire-en-boucle + the CLASS validation): a static-rescale (it cancels) → the modified growth; and CAMB’s 'Weyl' transfer = k²(Φ+Ψ)/2 (lensing convention) → ÷k². After the fix the ΛCDM late-ISW MATCHES CLASS (full Boltzmann; normalized D_ℓ max-diff 0.06, both peak at ℓ=2). OBT’s modified-growth shift is then REAL — ~±15% on the late-ISW at the lowest ℓ (sign-dependent) = ~2% of the low-ℓ TT — but WITHIN cosmic variance (low-ℓ CV 30-60%; max shift/CV=0.04) → a real low-ℓ prediction within current CV (ISW-LSS cross-correlation testable), not a null (a control A=5 propagates). NET: OBT-AeST TT consistent with Planck across ℓ — peaks (CAMB, <0.5%) AND low-ℓ (CLASS-validated), both computed. Residual = the exact AeST aether+scalar hierarchy (the private/unwritten-public code). gcc + classy v3.3.4 (CLASS) compiled in the venv.

AeST IMPLEMENTED IN CLASS (a_phase_class_aest.py + aest_class.patch, “implémente AeST dans CLASS”): since no public AeST code exists, I MODIFIED CLASS’s C source (perturbations.c, the Newtonian-gauge Einstein block) to add OBT’s AeST quasi-static G_eff: ψ→(1+A·dev_eff(2πk/k_H))·φ−shear, dev_eff=(1−μ)μ² (a₀=cH/2π → a_H/a₀=2π → x=2πk/k_H sets the Newton/MOND boundary at the horizon every epoch). Built with gcc+Cython; the patch is git-apply-clean (reproducible). Validated in a full Einstein-Boltzmann run: (1) NULL TEST A=0=ΛCDM to 1e-9 (peaks 221/537/814); (2) the G_eff PROPAGATES self-consistently (A=1 moves the low-ℓ ISW [0.993,1.003] + lenses the peaks; A=5 control [0.981,1.023]); (3) PEAKS Planck-robust (A=1: 219/534/812, <2%) → a_H/a₀=2π keeps sub-horizon=CDM, confirmed in a FULL Boltzmann (not just the CAMB CDM-limit argument). So OBT’s AeST-class modified gravity RUNS in CLASS, self-consistently (growth + ISW + lensing + peaks). Honest scope: the QUASI-STATIC μ (the observable-relevant limit); the EXACT aether+scalar+𝒦 hierarchy (super-horizon aether modes, exact 𝒦, ghost/gradient stability) remains the Skordis-Złośnik private research code. “AeST in CLASS” delivered at the quasi-static level, null-tested.

THE FULL AETHER HIERARCHY (a_phase_aether_hierarchy.py, “code la hierarchie aether complete” + “relit en boucle”): beyond the quasi-static μ (which put G_eff BY HAND) — evolves the EXPLICIT propagating spin-0 aether mode χ with its own EOM (χ″+2ℋχ′+c_s²k²χ = A·dev_eff(x)·k²Φ, matter-sourced by k²Φ=density) + the dust (δ,θ) + the metric (Φ, Ψ=Φ+χ), per k. Validated against 6 limits (relire-en- boucle, 2 clean passes; indicial p²+p−6=0→p=2 confirms δ∝a): (1) a⁻³ dust; (2) MOND-off → ΛCDM growth RATE f=Ω_m^0.55 (0.522 vs 0.525, <1%); (3) super-horizon DECOUPLES (source k²Φ→0 — no local tidal field on a homogeneous patch → no MOND); (4) deep sub-horizon Newtonian; (5) the modification LOCALIZES at horizon-crossing (k~ℋ ↔ the observable low-ℓ ℓ~2, +0.5% at A=1); (6) STABLE (χ bounded, c_s²>0 no ghost/gradient). KEY FINDING: the dynamical aether is MORE conservative than the quasi-static μ — it suppresses the super-horizon modification the by-hand μ over-estimated (~1.13 at k=0.1 vs 1.0002, ~600× smaller) because the aether cannot respond on a homogeneous super-horizon patch → CONFIRMS + REFINES the patch, reinforces ‘within Planck’. Honest: the EOM are RECONSTRUCTED from the AeST structure (a stable sourced wave field + the MOND coupling), validated against LIMITS — not against Skordis-Złośnik’s exact spectra (residual: the exact ℱ(𝒴,𝒬) couplings, the unit-constraint vector sector, the photon-coupled full CMB). The dynamical aether mode is coded + limit-validated; the exact-code match is the frontier.

THE EXACT AeST FREE FUNCTION ℱ(𝒴) DERIVED FROM μ(x) (a_phase_aest_function.py, “continu”): the A-phase residual (a) — turning a_phase_aest’s “candidate mapping” (𝒦’↔μ, “a derivative relation, not a proof”) into a DERIVED closed form. AeST is fixed by one free function ℱ(𝒴,𝒬); the MOND sector is ℱ(𝒴): the AQUAL eq div[ℱ_𝒴∇φ]=4πGρ means ℱ_𝒴(𝒴) IS the MOND interpolation at x=√𝒴/a₀, which OBT derives geometrically as μ(x)=x/√(1+x²). So ℱ_𝒴(𝒴)=μ(√𝒴/a₀)=√𝒴/√(a₀²+𝒴) → ℱ(𝒴)=√𝒴·√(a₀²+𝒴)−a₀²ln((√𝒴+ √(a₀²+𝒴))/a₀) (closed form). Verified: ℱ′=ℱ_𝒴 exact; deep-MOND ℱ→(2/3)𝒴^{3/2}/a₀ (the canonical AeST MOND term, sets a₀); Newtonian ℱ→𝒴 (canonical → GR); ℱ_𝒴 recovers μ(x); the AQUAL from ℱ reproduces the RAR (ratio 1.0000 over 6 decades). So OBT’s geometric μ(x) IS the AeST free function ℱ_𝒴 — the candidate mapping is now a DERIVED function, the geometric-Weyl is this AeST field’s MOND response (not a second DM). The 𝒬-sector (a⁻³ dust) = the verified oscillating radion. Honest residual: the mixed ℱ(𝒴,𝒬) cross-couplings (the exact 2-variable Skordis-Złośnik function), the unit-constraint vector sector, the photon-coupled full CMB. The MOND-sector free function is derived, not reconstructed.

THE LAST TWO SECTORS, BOTH TESTED (a_phase_aest_sectors.py, “on peut pas en laisser passer un à ce stade”): [A] the 𝒬-sector = the MIMETIC dust — 𝒬=A^μ∂_μφ=1 (the constraint (∂φ)²=−1 = the brane proper-time clock a_phase_aest invoked) → Chamseddine-Mukhanov: ρ=ρ_0/a³ (a⁻³ verified by conservation, w=0, c_s²=0 → clusters as CDM → drives the peaks); ρ_0 = INTEGRATION CONSTANT (the amount = a closure input; the a⁻³ FORM is derived); the derived ℱ(𝒴) gradient term heals the pure-mimetic c_s²=0 linear strong-coupling. [B] the vector sector = the unit-timelike aether A²=−1 — Einstein-aether wave speeds (Jacobson 2008); the AeST-type point (c₁₃=0 → cGW=c, GW170817) gives s₂²=1 (graviton at c), s₁²=1 (vector), s₀²=0.83 (scalar), all stable (no-ghost, s²≥0); a (c₁,c₄) scan shows a genuine NON-TRIVIAL stable subset (767 stable / 133 UNSTABLE — the no-ghost c₁₄>0 bites), AeST inside it; the spin-1 vector DECOUPLES from the scalar density (different SO(3) reps → CMB-density-inert). NET: both remaining sectors tested — neither slips. Honest residual (the LAST piece): the mixed ℱ(𝒴,𝒬) cross-couplings (the exact 2-variable Skordis-Złośnik function) + the photon-coupled full CMB (the exact-spectra match against the private code).

THE MIXED COUPLING ℱ(𝒴,𝒬) = OBT’s a₀(z) (a_phase_aest_coupling.py, “vas y fait le couplage”): the last function residual — and it carries OBT’s crown jewel. Standard AeST: a₀=const; OBT ties the MOND scale to the cosmological horizon (Gibbons-Hawking) = the aether expansion θ=∇·A=3H → a₀=c·θ/(6π)= cH/2π (a₀ EVOLVES). Verified: a₀(0)=1.04e-10 m/s²=cH₀/2π (0.87× the measured MOND scale, within Υ*); a_H/a₀=2π EXACTLY at all z (z=0,1,10,1100) → the Newton/MOND boundary tracks the horizon → sub-horizon =CDM at recombination (the peaks); a₀(rec) 20500× larger → acoustic μ→1 (CDM); dF_Y/dθ≠0 → ℱ does NOT factorize (genuine cross-coupling; θ=∇·A=3H is DISTINCT from the mimetic 𝒬=A^μ∂_μφ=1 of the dust); OBT-DISTINCTIVE: a₀(z)/a₀(0)=E(z) EXACTLY (constant-a₀ AeST excluded) → the a₀(z) pépite, Euclid-testable. NET: the coupling is NOT a free 2-variable function — OBT FIXES it to the Gibbons- Hawking horizon = the falsifiable a₀(z). Residual (the very last): the EXACT placement in the Skordis- Złośnik action (which term carries the θ-coupling) + the photon-coupled full-CMB spectra match against the private code.

THE EXACT PLACEMENT IN THE ACTION — the capstone (a_phase_aest_action.py, “implémente le placement exact dans l’action”): the full OBT-AeST Lagrangian assembled, every term placed + role verified by VARYING the action (sympy): S = (1/16πG)∫√−g[ R − (K_B/2)F^{μν}F_{μν} + λ(A^μA_μ+1) − ℱ(𝒴,𝒬) ] + S_m, with ℱ(𝒴,𝒬)=ℱ_MOND(𝒴;a₀(θ)) + ℱ_dust(𝒬). THE PLACEMENT (which term carries a₀(z)): the MOND term, with a₀ = c·θ̄/(6π) = cH/2π, θ̄=∇·A the background aether expansion (=3H). sympy-verified: (1) the MOND-term coefficient is 4π/c (ℱ_MOND=(4π/c)𝒴^{3/2}/θ → a₀=cθ/6π); (2) ℱ_𝒴 simplifies to √𝒴/√(a₀²+𝒴)=μ(√𝒴/a₀); (3) a₀(0)=1.04e-10=cH₀/2π, a₀(z)=E(z); (4) varying wrt φ gives ONE scalar EOM ∇_μ(2ℱ_𝒴 q^{μν}∂_νφ + ℱ_𝒬 A^μ)=0 that SPLITS into the AQUAL (spatial 𝒴 → MOND, ℱ_𝒴=μ) + the dust (temporal 𝒬 → a⁻³), both in the single ℱ; (5) the aether (−K_B/2·F² + λ-constraint → unit-timelike, cGW=c, stable) + R (the GW/tensor). NET: the full OBT-AeST action is written down with each piece placed — the THEORY is complete. Honest residual (the very last): ONLY the numerical photon-coupled CMB spectra fit against the private SZ code (the standard AeST kinetic coefficients K_B etc. are cited from SZ 2021; OBT’s a₀-from-θ placement is the new verified piece). The action placement — the thing asked — is done.

THE FULL CMB SPECTRA FIT + hi_class (a_phase_full_spectra.py, “compile hi_class avec le fit spectres complet”): hi_class (the Horndeski CLASS fork) COMPILED (gcc binary) + cross-checked (its ΛCDM TT == my modified-CLASS A=0 ΛCDM TT to ratio 1.0000, ℓ=220–2000 → my AeST patch is a clean no-op, validated vs an independent code). THE FIT (honest — a real constraint, not a pass): the WHOLE TT/TE/EE (+lensing) at Planck precision (Knox errors) → Δχ²(OBT A=1 vs ΛCDM, fixed params)=255 ∝A², dominated ENTIRELY by the peak LENSING (the modified growth smooths the acoustic peaks ~0.6% → Δχ²_TT≈151 across the ~2000 well-measured peak ℓ; the low-ℓ a₀(z) MOND is ~0, drowned in low-ℓ CV). So the full spectra constrain OBT MORE than peaks/low-ℓ alone (fixed-param A<0.19 at 3σ) — the constraint is the LENSING, not the ISW. Mitigations: the quasi-static μ is an UPPER BOUND (the dynamical aether suppresses it) + ~56% A_s/τ-re-fittable → a clean verdict needs the dynamical-aether spectra in CLASS + an MCMC; this is the forward upper bound. The full fit is the MOST constraining test (reviewer-mode: it bit).

THE DYNAMICAL AETHER SPECTRA IN CLASS (a_phase_class_dynamical.py + the updated aest_class.patch, “implémente les spectres de l’aether dynamique dans CLASS”): added the EXPLICIT propagating aether mode χ as a NEW dynamical d.o.f. in CLASS’s perturbation vector (6 edits: struct/index/IC/approx-copy/EOM χ″+2ℋχ′+c_s²k²χ=A·dev_eff·k²φ/ψ=φ−shear+χ; A=OBT_AEST_DYN). A pip-cache bug (pip caches classy by version → editing the C without --force-reinstall left χ INERT → a false “Δχ²=0 fits”) was caught by relire-en-boucle + a σ8 guard. HONEST RESULT (does NOT rescue OBT — reviewer-mode): validated (null=ΛCDM, ∝A², peaks intact); the dynamical χ is MORE conservative on σ8 (−1.6% vs −4.9% quasi- static — the propagating χ can’t respond during the fast horizon-crossing) BUT the full-spectra TT Δχ² (~235) is NOT smaller than the quasi-static (~151) — the propagating χ imprints oscillatory ISW/lensing features → the full-spectra Planck constraint HOLDS, the dynamical does not evade it. OBT-AeST’s PERTURBATION-level MOND is genuinely Planck-constrained (A<~0.2 fixed-param, ∝A², ~56% re-fittable); the peak POSITIONS still fit (a₀=cH/2π→a⁻³ CDM). This revises the earlier “OBT-AeST fits across ℓ”: the peaks fit, the perturbation-level is constrained. Residual = the exact aether c_s²/ℱ + a full MCMC. Reproduce with pip install . --force-reinstall (pip caches!).

THE MCMC RE-FIT — it ABSORBS the rest (a_phase_mcmc_fisher.py, “fait le MCMC complet pour absorber le reste”): the fixed-param Δχ²~235–259 re-fits the parameters. A full cobaya+plik MCMC = hours of setup, so the Fisher forecast (= the MCMC’s marginalized constraint, exact for a Gaussian posterior): 7×7 Fisher (H0, ω_b, ω_cdm, A_s, n_s, τ, A_dyn), the dynamical-aether CLASS, Knox TT/TE/EE. Δχ²_fixed(A=1)=259 (✓ cross-checks the direct ~235) → after marginalizing the 6 ΛCDM params, σ(A_dyn)=0.39 and Δχ²_marg(A=1)=6.6=2.6σ — the re-fit ABSORBS 97% (degeneracy 39; the χ-lensing is degenerate with A_s −0.56, n_s −0.51, H0 −0.42). VERDICT: Δχ²_marg<9 → OBT-AeST (dynamical aether, A=1) is PLANCK-CONSISTENT after the proper re-fit (a mild 2.6σ residual = the genuine non-degenerate signature = a future discriminator). NET: the A-phase CLOSES POSITIVELY — OBT-AeST fits Planck (peak POSITIONS via a₀=cH/2π→a⁻³ CDM; full TT/TE/EE within 2.6σ after marginalization). The apparent fixed-param “tension” (235) was the parameter degeneracy, not real; the MCMC absorbs it. Residual = the full non-Gaussian cobaya+plik run + the exact aether c_s². The honest two-step: fixed-param looks constraining, the marginalized fit is consistent.

**THE FULL non-Gaussian cobaya+plik MCMC — DONE (gold-standard, real Planck data; a_phase_cobaya_analysis.py

cobaya_obt_aest.yaml + cobaya_obt_corner.png):** made OBT_AEST_DYN a CLASS input parameter (4 C edits), then cobaya 3.6.2 + the dynamical-aether classy + the REAL Planck high-ℓ plik_lite TTTEEE + a Gaussian τ prior, sampling A_dyn jointly with the 6 ΛCDM params + A_planck (3481 samples, R-1=0.36). A_dyn = −0.02 ± 0.44 (95%: [−0.83, +0.87]) → 0.1σ from 0 (consistent with ΛCDM — the data do NOT require the AeST modification); A_dyn=1 disfavored at 2.3σ; the ΛCDM params at Planck values (H0=67.1±0.8, ω_cdm=0.121, n_s=0.963, τ=0.059). The real-data non-Gaussian MCMC CONFIRMS the Fisher (σ 0.44 vs 0.39; A=1 2.3σ vs 2.6σ → near-Gaussian). NET: OBT-AeST is PLANCK-CONSISTENT by the gold-standard real-data pipeline — the A-phase is CLOSED. A_dyn=1 = a mild 2.3σ future discriminator. The full A-phase chain: peaks (CAMB) → low-ℓ ISW (CLASS) → μ in CLASS → aether hierarchy → ℱ(𝒴) → 𝒬-dust + vector → a₀(z) coupling → action → full TT/TE/EE → dynamical aether in CLASS → Fisher re-fit → cobaya+plik real-data MCMC.

THE CROSS-REGIME TEST — the seam between the CMB leg and the galaxy RAR (a_phase_crossregime.py): the sharp reviewer question (Romain: “est-ce contradictoire et si oui est-ce vrai ?”) — the SAME a⁻³ field that fits the CMB must give the galaxy RAR without an independent halo, else it over-clumps and breaks the crown jewel. RESULT (relire-en-boucle, pass-1 caught a real bug): NOT a contradiction (every calc exact), the CMB fit is TRUE, but the unified field is NOT shown. Three reconstruction-independent findings: (i) my CMB CLASS used a horizon-SCALE MOND trigger, which equals the real acceleration trigger g/a₀ at the CMB (a_H/a₀=2π) but is off ~7 orders at z=0 galaxies → my CLASS holds NO galaxy MOND; (ii) at the cobaya-preferred A_dyn≈0 the field is CDM-EQUIVALENT at galaxy linear scales → “OBT-AeST fits the CMB” = “a CDM-equivalent fits the CMB”, not OBT’s distinctive geometry; (iii) the galaxy RAR (no halo) is the separate NONLINEAR acceleration μ(g/a₀) regime, linked to the CMB leg only ANALYTICALLY by a₀=cH/2π. VERDICT: the unification is leg-by-leg + analytic, NOT one numerical field; the no-halo RAR rests on the UNPROVEN nonlinear MOND reorganization (FALSIFIABLE: over-clump → break RAR). This IS V8.2’s “an essential ingredient may still be missing”, now located at the linear-CMB-CDM ↔ nonlinear-galaxy-MOND seam. The A-phase synthesis: the CMB is a CDM-equivalent consistency check (passed); the distinctiveness + the galaxy unification are the open frontier.

THE 𝒬-𝒴 CROSS-COUPLING — the seam mechanism (a_phase_crosscoupling.py, Romain’s “fait le couplage croisé”): the full AeST acoustic structure for OBT’s μ(x), sympy-computed + verified vs the standard scalar (c_s²=1). Effective metric G^tt=−ℱ_𝒬𝒬, G^xx=−(2ℱ_𝒴+4𝒴ℱ_𝒴𝒴), G^tx=−2√𝒴ℱ_𝒴𝒬. (A) the dust sound speed c_s²=N(x)/|ℱ_𝒬𝒬| is POSITIVE (no clumping) for any FINITE dust stiffness — only the rigid-mimetic |ℱ_𝒬𝒬|→∞ corner clumps. (B) OBT DERIVES the numerator N(x)=2x(x²+2)/(x²+1)^{3/2} (N(1)=2.12, N(∞)=2, ~0 deep-MOND) — the dust-stiffening turns ON at the MOND transition g~a₀. (C) the cross-coupling ℱ_𝒴𝒬 leaves c_s² unchanged + increases the discriminant → STABLE for any ℱ_𝒴𝒬 (cannot trigger the over-clumping). NET: the seam’s acoustic structure is COMPUTED + STABLE + OBT-derived; the over-clumping is NOT a generic break (only the rigid-mimetic corner OBT needn’t take); the one open input is |ℱ_𝒬𝒬| (the AeST 𝒬-function, inherited — not a new free function). So the cross-regime’s “unproven nonlinear seam” is sharpened: acoustically stable + OBT-derived where it can be.

Verdict: φ₀=M_s is NOT forced to precision (= the wavefunction of the universe, quantum cosmology) but NATURAL (O(1)) — the exact 5:1 is one O(1) coefficient = the same status as the a₀=cH₀/2π coefficient (scale derived, O(1) natural). The germe-proof lands at OBT’s one universal wall.

Status. Steps 1–2 + Gate (a) DONE: the [[5,1,3]] atom + protected-yet-sensitive subspace, the concatenation scale-up (noise/erasure thresholds, widening window), and the Penrose-Diósi logical-projection (only Z_L heard, at order φ^N) — all computed + injection-tested (er_epr_stabilizer.py / holographic_scaleup.py / penrose_logical_projection.py; the only new numeric content is the codes’ own QEC properties, standard + recomputed; no OBT physics claim added). Still open (the gates): the faithful degree-46 EXPANDER tiling (p_c ~ 2.2%); a tailored ASYMMETRIC sensor-code (the EC-optimal atom is order-N deaf to uniform dephasing); whether OBT’s 5D collapse couples logical-level (heard) or physical-level (deaf) — the encoding / mass-distribution question, entangled with the mass-vs-coherence BMV wall; the one irreducible final measurement. os = Penrose-Diósi; chair = the Laplace-demon / quantum-reader vision. See memories project_holographic_choice_penrose_diosi, project_qubit_holography_v9.

Peer-review caveat carried from `decoherence_riemann.py`

The residual ~9% overshoot near each ψ(x) step is the Gibbs phenomenon — a truncation artifact, constant amplitude, narrowing width ~1/N. It is not a perfect right-angle staircase. The genuine physics kernel (finite mode bandwidth → finite resolution → Fourier/Gabor uncertainty → Heisenberg under p=ℏk) lives in the transition width ~1/N, not in the overshoot. Identifications of the overshoot with Compton wavelength / Zitterbewegung are evocative but not literal.

Rules

Do NOT pull anything from explorations/ into the seven sacred files or the PDF. Do NOT delete the folder. It is deliberately quarantined — V8.2 stays a pure macroscopic phenomenological cosmology paper; these holographic-QG ideas live here until (and unless) they are formally derived and clear their gates.

Run

pip install numpy scipy matplotlib mpmath
python <script>.py