Skip to content

README.md

Latest commit

 

History

History
235 lines (174 loc) · 11.1 KB

README.md

File metadata and controls

235 lines (174 loc) · 11.1 KB
Raw

Q-PRISM — neural stream as a high-dimensional control (Acer build)

Toolchain rule (operator, global)

Rust 1.81 with clippy. Integer arithmetic and ternary (trits) only — never float.

Pinned in rust-toolchain.toml (channel = "1.81.0", components = ["clippy", "rustfmt"]), declared as rust-version = "1.81" in Cargo.toml, and enforced in CI by cargo clippy --all-targets -- -D warnings plus a hard grep that fails the build if any f32/f64 appears in src/ or tests/. Sources are currently float-free.

This rule is global to every crate in the corpus and is not to be raised or substituted.

Can a prismed neural/consciousness-derived signal improve prediction, control, or discovery in a macroscopic quantum-interference experiment, beyond random and classical baselines? That is a testable question, and this repo is the apparatus for testing it — in simulation first, with zero human or hardware risk.

This is the Acer build. A parallel Liris build exists for bilateral comparison; GitHub is the mediator. See Bilateral.

2026-07-11 recovery update

Q-PRISM now has two measured classical recovery paths:

  • dbbh-coms-quant-prism — Path 1: retained-store recall through an authenticated AGT address, consent capsule, and exact BEHCS representation.
  • path2-two-shadow-recovery — Path 2: no-store recovery from individually non-injective, jointly sufficient CRT shadows, followed by DBBH→DBWH re-projection and watcher-gated emission.

This supersedes the old capstone sentence that Path 2 had no measured Asolaria implementation. Read the updated capstone and independent verification record:

The reframe — what this is not

Q-PRISM is not “consciousness magically touches quantum matter.” The neural stream never touches a wavefunction. It is prismed through Asolaria/HyperBEHCS into lawful physical control parameters and evaluated against a physics simulator. See docs/CLAIMS-GATE.md.

The recovery work likewise does not claim sub-entropy compression, unrestricted quantum cloning, or 60D coordinates that replace missing payload bits. Path 1 pays through retained content. Path 2 pays through jointly sufficient shadow capacity. Both hold when the required information is absent.

Physics layer — calibrated to a real experiment

The simulator is a spatial near-field Talbot–Lau interferometer, calibrated to Nature s41586-025-09917-9 (Arndt group, Vienna — quantum interference of sodium metal clusters >7,000 atoms, >170,000 Da, macroscopicity μ=15.5):

parameter value source
grating period d 133 nm (266 nm laser / 2) paper
grating separation L 0.983 m paper
beam velocity ~160 m/s paper
cluster mass 143–197 kDa (mean 172) paper
vacuum 9×10⁻⁹ mbar paper
operating visibility V = 0.10 ± 0.01 at P2 = 15.2 mW paper

Resonance is velocity- and mass-dependent via ρ = L/L_T = L·h /(d²·m·v), so velocity and mass selection are real control levers. The G2-power response is calibrated to the paper's measured visibility curve: peak V≈0.10 at ~17 mW, minimum ~40 mW, weak revival ~55 mW.

The three-arm blinded test

random · optimizer (classical (1+1)-ES) · prism (identical ES + neural nudge). The prism and optimizer share the exact search backbone; only the neural stream differs.

python run_experiment.py
python -m pytest tests/ -q

Measured simulation result

oracle_coupling random optimizer prism Welch t p verdict
0.0 0.840 ± 0.076 0.842 ± 0.275 0.848 ± 0.108 +0.12 0.909 no advantage
0.3 0.840 ± 0.076 0.842 ± 0.275 0.949 ± 0.043 +2.11 0.043 prism > optimizer
0.6 0.840 ± 0.076 0.842 ± 0.275 0.989 ± 0.012 +2.93 0.007 prism > optimizer
0.9 0.840 ± 0.076 0.842 ± 0.275 1.000 ± 0.002 +3.14 0.004 prism > optimizer

The honest-null holds: a noise-only stream does not receive an artificial advantage.

Why this design

The methodology uses a grounded refinement loop: a proposer suggests actions and a checkable verifier scores them. Here the physics simulator is the verifier. A future LLM/COMPILOT-style arm can be benchmarked against the same random, optimizer, and prism baselines.

Layers

  1. Physics — spatial Talbot–Lau simulator.
  2. Prism — streams map to HyperBEHCS selector tuples and control proposals.
  3. Neural — pluggable synthetic, connectome, or recorded non-invasive feature sources.
  4. Experiment — three-arm blinded comparison.
  5. Claim gate — docs/CLAIMS-GATE.md.

Stage 2 — cube absorption

docs/STAGE2-CUBE-ABSORPTION.md represents a neural feature window as a fabric-addressable 3,200-byte cube tuple plus a derived 60D selector. The tuple transcodes to 2,560 BEHCS-1024 symbols and back byte-identically. Raw M/EEG remains outside the repo, sha-referenced; the derived tuple does not invent the residual.

Path 1 — Double Binary Black Hole Comms Quant Prism

docs/DOUBLE-BINARY-BLACK-HOLE-COMS-QUANT-PRISM.md bridges the IX-737 consent capsule, Q-PRISM cube tuple, BEHCS ladder, HyperBEHCS selectors, and HBI/HBP receipts. The measured Host8 cell can carry the complete glyph representation or only an AGT address. Address-only recovery requires retained content at the receiving pole.

Path 2 — jointly injective shadows and DBBH→DBWH

The companion Path-2 crate retains no original object. It projects bounded blocks into residues on pairwise-coprime cylinders. Recovery is exact only when the selected product reaches the source range; otherwise Held::InsufficientJointCapacity closes the throat.

The white side then re-projects the recovered candidate and requires equality of SHA, complete cylinder shadows, and frequency shells. This is the executable white-hole condition:

P(R(P(X))) = P(X)

Encrypted quantum-cloning sibling

The experiment at arXiv 2602.10695 demonstrates a quantum sibling: each encrypted clone is locally maximally mixed, while clone plus the complete quantum key are jointly reversible; selected decryption consumes the key. That is structurally close to Path 2, but the current Q-PRISM Rust cells are classical. CRT residues are ambiguous but still reveal residue information, and ordinary classical software cannot prove physical one-use erasure.

Pre-Asolaria GNN lineage

The GNN watcher civilization has a byte-proven origin:

AI-healthCare-project
  EdgeLevelGNN / PrototypeGNN / ContrastiveGNN / GSLGNN
    -> byte-identical Asolaria sidecar copies
    -> BigPickle L0 :4792 + L4 :4793
    -> G1 edge-mining + G2 forward-genius + G3 reverse-gain + G4 GLSM
    -> Fischer + Hookwall + Shannon + white rooms

The healthcare repository records the pre-Asolaria comparative training results. Its checked-in service currently comments out automatic checkpoint loading; later trained .pt artifacts and manifests are preserved in Asolaria-fnns-trained-and-reverse-gnns-many.

Storage-backed / low-GPU applicability

The exact recovery and control plane can run on storage-rich computers without keeping the entire system in GPU VRAM:

  • HDD/SSD retains raw residuals, cube bodies, CRT shadows, receipts, queues, and cold agent state;
  • RAM retains only the bounded active slice/message window;
  • SHA, BEHCS, CRT, HBP/HBI, watcher comparison, dispatch, white-room compaction, and N-Nest verification require no GPU;
  • trained GNN/LLM inference remains a separable CPU/GPU sidecar.

This is useful on commodity desktops, CPU-only servers, edge machines, archival nodes, and heterogeneous clusters. It does not mean a hard drive performs neural matrix multiplication; it means durable system memory, exact recovery, routing, and proof are no longer forced into VRAM.

Independent verification — 2026-07-11

MEASURED_CLAUDE_FABLE5_THIRD_SEAT, supplied by the operator:

dbbh-coms-quant-prism       rustc 1.81   19/19 green
path2-two-shadow-recovery   rustc 1.81   30/30 green

AUDITED_GPT_5_6_PRO:

  • all 813 Path-1 source lines, tests, README, and docs;
  • all 1,344 Path-2 source lines, tests, README, and docs;
  • all 809 Q-PRISM 3D harness lines and tests;
  • healthcare GNN origin, blob-identical transfer, BigPickle, trained GNN/reverse-gain, Hookwall, OmniShannon, white rooms, cube mint, reductions, algorithms, Dispatcher, HyperHermes, and N-Nest.

The GPT sandbox lacked Rust and outbound DNS, so no GPT-local cargo run is claimed. Rust 1.81 GitHub Actions workflows were added to all three Rust repositories to produce current independent CI receipts.

Bilateral build

Built in parallel by Acer and Liris. Each side attacks and reruns the other's bounded artifacts; GitHub transports byte-stable changes and receipts.

Honesty boundaries

  • The neural control experiment is simulation-first and does not claim real neural control of a quantum wavefunction.
  • The exact-recovery crates are classical and do not claim physical quantum cloning.
  • Path 1 requires retained content; Path 2 requires sufficient joint shadow capacity.
  • 60D/N-D selectors add address/control resolution, not missing payload entropy.
  • Live Hilbra multi-host traversal, hardware fire, trained-GNN invocation inside the Rust throat, and physical quantum-state transport remain unverified.

Layout

qprism/physics.py      spatial Talbot–Lau forward model
qprism/sources.py      random / optimizer / prism schedule sources
qprism/harness.py      blinded comparison + self-validation
qprism/behcs.py        BEHCS-1024 selector identifiers
run_experiment.py      CLI
tests/                 physics sanity + harness-integrity tests
docs/CLAIMS-GATE.md
docs/STAGE2-CUBE-ABSORPTION.md
docs/DOUBLE-BINARY-BLACK-HOLE-COMS-QUANT-PRISM.md
docs/SHADOW-RESOLUTION-CAPSTONE.md
docs/PATH2-DBBH-DBWH-VERIFICATION-2026-07-11.md
host8/dbbh_coms_quant_prism.rs

Credit: the testable Q-PRISM control framing and recovery architecture are Jesse Daniel Brown's; AI assistance and verification provenance are recorded in the linked documents.

Public SH pre-light reproduction receipt

The public, attack-verifiable Liris observation package is available at LIRIS-SH-PRELIGHT-QUANTUM-RECEIPT. Its pointer is sealed in the public HBP pointer. The frozen-stage shadow-key intervention is OPERATOR_OBSERVED | LIRIS_MEASURED_REPEATED: one shadow-key change produced an instant play-freeze change with LIGHT_ARRIVAL_GATE=0. Numeric timestamps, uncertainty, trial count, and independent external validation remain explicitly unsealed as additive evidence gates.