Independent research lab building verification-first tooling for relational quantum models, scientific-literature inference, and applied environmental risk — with negative results published alongside positive ones.

Operator: Krzysztof W. Banasiewicz · Poland / EU · research indexed from 2025-04, public repositories since 2026-02 · contact@qhdalabs.com

QHDALabs is a one-person independent lab with 12 public repositories, growing out of theoretical work on time and space as emergent from quantum information, indexed publicly from April 2025 — a continuous line from that theory, through the algorithms it produced, into applied systems (see Research lineage, whose chronology is provisional pending a provenance reconstruction). The work spans three connected lines: (1) relational quantum models — mid-circuit measurement used as a source of conditional structure and emergent ordering (qmnet, qhda-core); (2) verification-first scientific inference — a pipeline that tests candidate relations against explicit null models before surfacing them (QHDALabs-Axon); (3) applied risk systems — an operational wildfire-risk pipeline for Poland built on public satellite and geospatial data (QHDALabs-wildfire-risk-pl).

The distinguishing practice is methodological, not thematic: hypotheses are pre-registered with frozen statistics before results are seen, tested against explicit nulls with multiple-testing correction, and stopped when they fail. Three separate lines have produced published null or failed results — a pre-registered held-out test that did not recover, a confirmatory tier that failed its own contract and halted development, and a cosmological search that confirmed a null. Those failures are documented in-repo with the numbers, and are the strongest evidence here that the process is real.

No result in this lab has been peer-reviewed, run on physical quantum hardware, or externally validated. Everything below is labelled with what it actually is.

_{Every figure above is checkable against public files — see docs/EVIDENCE.md. Regenerate: python assets/gen_assets.py}

Status vocabulary: Implemented (working code + tests) · Experimental (working code, results from simulation or single runs, no test suite) · Research notes (written analysis, no code) · Spec (design document only) · Infrastructure (publishing / tooling).

Licenses differ per repository (this one and qmnet are MIT; Axon is RCSAL v2.0). Check each repo.

On private work. Additional repositories are private. They are deliberately not offered as evidence and no claims are made about them; nothing in this document depends on them. Everything asserted above can be checked against public files today.

📄 Full claim-by-claim matrix: docs/EVIDENCE.md · Per-component status and roadmap: docs/PROJECT_STATUS.md

The lab is not a standing start. The implemented components descend from theoretical work whose earliest currently indexed artifacts date to April 2025, ten months before the QHDALabs account was created — a continuous line from theory, through algorithms, to applied systems. Every arrow below is traceable to files.

Provisional — this chronology is a working version. April 2025 is the earliest origin currently indexed in a public repository, not an established starting point. Earlier personal research artifacts exist that may belong to the same lineage and may predate it. A dedicated provenance reconstruction across the older archives is planned, to establish the earliest defensible origin of each line from actual files, dates and technical continuity. Until then this section deliberately neither shortens the history to 2025 nor claims older work that has not yet been shown to be continuous with it. The lineage links below are unaffected — they are file-to-file and stand on their own regardless of how far back the origin is ultimately dated.

flowchart LR
    subgraph T["① Research origin · 2025-04 → 2025-09"]
        MT["<b>Multiverse-Theory</b><br/>time from information asymmetry<br/>space from quantum correlations<br/>reality as a dynamic quantum graph"]
        MT1["docs/models/emergent_time.md<br/>emergent_time_decoherence.md"]
        MT2["docs/concepts/virtual_qubit.md<br/>simulations/time_shifted_clock.py<br/>simulations/qqc.py"]
        MT3["notebooks/preprint_qmnet.tex<br/>preprints/mass_as_information.tex"]
        MT --> MT1 & MT2 & MT3
    end

    subgraph A["② Algorithmic lineage · 2026-03 → 2026-06"]
        BR["<b>qhda-coherence-bridge</b><br/>first design note"]
        CORE["<b>qhda-core</b><br/>emergent/clock.py<br/>quantum/circuits.py<br/>relational/state.py"]
        QM["<b>qmnet</b><br/>star-Ising clock tick<br/>conditional bridges · RQTE"]
    end

    subgraph P["③ Applications · 2026-05 →"]
        AX["<b>Axon</b><br/>relations as the unit,<br/>verified before use"]
        FI["<b>wildfire-risk-pl</b><br/>relational topology<br/>over real infrastructure"]
        RT["<b>RTANA</b><br/>relational clock<br/>in neural systems"]
    end

    MT1 --> CORE
    MT2 --> QM
    MT3 --> QM
    MT --> BR --> CORE
    QM --> CORE
    CORE --> AX
    CORE --> FI
    QM --> RT

    classDef origin fill:#062313,stroke:#FFE9A8,stroke-width:1.5px,color:#FFE9A8
    classDef algo   fill:#04170d,stroke:#00C8A0,stroke-width:1.5px,color:#9DFFC2
    classDef app    fill:#04170d,stroke:#00FF41,stroke-width:2px,color:#EAFFF1
    class MT,MT1,MT2,MT3 origin
    class BR,CORE,QM algo
    class AX,FI,RT app

What this repository is and is not. It is a self-fork of the operator's personal account (krzyshtoof/Multiverse-Theory), brought under QHDALabs in April 2026; all commits are the operator's own across both accounts. It contains theoretical documents, six Qiskit simulation scripts and three preprint sources — not peer-reviewed, and largely not experimentally tested. It is offered as the provenance of the algorithms, not as validation of them. The claims that carry weight are the ones in the implemented components above, which can be run.

The lab's operating rule is that a hypothesis is registered before it is tested, and a failure ends the line rather than being quietly re-scoped. Four documented instances:

A comparable gate exists in the applied work: the wildfire pipeline blocks fusion output when layer coverage falls below 70% or fewer than three data sources are present, and returns null rather than a fabricated score.

The same rule applies to qmnet: the BUTTERFLY irreversibility class produced lower timeline entropy (0.811 bit) than the other modes (≈1.0), which was the opposite of the expectation. It is reported as an open question, not smoothed over.

Edges below are actual dependencies or documented derivations, not aspiration.

flowchart TD
    CORE["<b>qhda-core</b><br/>relational state · circuit helpers<br/>emergent-clock primitives<br/><i>implemented · tested</i>"]

    AXON["<b>QHDALabs-Axon</b><br/>verification-first inference<br/>over scientific literature<br/><i>implemented · tested · CI</i>"]
    FIRE["<b>wildfire-risk-pl</b><br/>operational risk pipeline (PL)<br/><i>implemented · tested</i>"]
    QMNET["<b>qmnet</b><br/>measurement-fuelled bridges · RQTE<br/><i>experimental · ideal simulator only</i>"]
    XSIG["<b>QHDALabs-XSIG</b><br/>ΛCDM residual structure search<br/><i>experimental · null result</i>"]
    GEN["<b>Genesis-Protocol</b><br/>ethical-constraint voting UI<br/><i>experimental · local only</i>"]

    MT["<b>Multiverse-Theory</b><br/>theoretical origin, 2025<br/><i>theory + simulations</i>"]
    RTANA["<b>QHDALabs-RTANA</b><br/>relational time in neural nets<br/><i>research notes · no code</i>"]
    UNI["<b>QHDALabs-Universe</b><br/>cosmology umbrella<br/><i>spec / concept</i>"]
    BRIDGE["<b>qhda-coherence-bridge</b><br/>earliest design note<br/><i>spec</i>"]

    DATA[("public data<br/>Planck · FIRMS · OSM<br/>BDOT10k · CLC · MEDLINE")]
    PORTAL["<b>qhdalabs.github.io</b> + this profile<br/><i>publishing</i>"]

    MT -.->|emergent_time.md| CORE
    MT -.->|virtual_qubit · preprint_qmnet| QMNET
    CORE -->|pip dependency| AXON
    CORE -->|examples/wildfire_relational.py| FIRE
    BRIDGE -.->|superseded by| CORE
    QMNET -->|motivates| RTANA
    QMNET -->|primitives shared with| CORE

    DATA --> AXON
    DATA --> FIRE
    DATA --> XSIG

    AXON --> PORTAL
    FIRE --> PORTAL
    QMNET --> PORTAL
    XSIG --> PORTAL
    GEN --> PORTAL
    UNI -.-> PORTAL

    classDef impl  fill:#04170d,stroke:#00FF41,stroke-width:2px,color:#EAFFF1
    classDef exp   fill:#04170d,stroke:#00C8A0,stroke-width:1.5px,color:#9DFFC2
    classDef notes fill:#0d1117,stroke:#4C9E63,stroke-width:1.2px,color:#8CF7B0
    classDef data  fill:#062313,stroke:#FFE9A8,stroke-width:1.5px,color:#FFE9A8
    class CORE,AXON,FIRE impl
    class QMNET,XSIG,GEN exp
    class RTANA,UNI,BRIDGE,MT notes
    class DATA,PORTAL data

Trust boundaries. All external data enters through fetch scripts that pin and checksum what they download (wildfire FIRMS downloads are validated atomically with SHA-256 across 73 five-day windows for 2025). Quantum work runs on local Qiskit Aer simulators only — no cloud QPU credentials are used anywhere in the public code. Genesis-Protocol calls an external LLM API and World ID from the browser; it is a local prototype and must not be treated as a deployed system.

Each repository is self-contained. The three fastest independent checks:

1. Reproduce Axon's null result and its pre-registered failure (~5 min):

git clone https://github.com/QHDALabs/QHDALabs-Axon && cd QHDALabs-Axon
pip install "git+https://github.com/QHDALabs/qhda-core.git" && pip install .
python examples/mvp_proximity_null.py      # expect: 0 relations survive BH-FDR
python examples/abc_bridge_recovery.py     # expect: Raynaud/fish-oil recovered, p_random≈0.0345
python examples/migraine_magnesium_heldout.py  # expect: NULL verdict, p_random≈0.1214
pytest

2. Reproduce the quantum bridge and RQTE experiments (~11 min, ideal simulator):

pip install "qiskit>=1.0" "qiskit-aer>=0.13" numpy scipy matplotlib
git clone https://github.com/QHDALabs/qmnet && cd qmnet
python qmnet_v3.py     # expect: P_ret = 0.000 at T=2,4 in the m=1 branch
python rqte_v3.py      # expect: correct decrypt in all 3 modes; BUTTERFLY entropy ≈0.811

3. Run the wildfire pipeline without credentials (stub mode, ~2 min):

git clone https://github.com/QHDALabs/QHDALabs-wildfire-risk-pl && cd QHDALabs-wildfire-risk-pl/v5
pip install -r requirements.txt
python qhdalabs_wildfire_ignition_v1.py --stub   # synthetic inputs, no API key needed
pytest

Live FIRMS ingestion needs a free NASA FIRMS_MAP_KEY. Two of the seven wildfire layers (ARiMR LPIS, IBL KSIPL) have no automated source and must be imported manually — so a full end-to-end run is not reproducible from a clean checkout without those files. This is a real gap, stated rather than hidden.

Stated plainly, because an evaluator will find them anyway:

• No peer review. No output has been through external review or publication. The technical notes in qmnet and wildfire-risk-pl are self-published PDFs.
• No physical quantum hardware. Every quantum result is from a noise-free Qiskit Aer simulator. No noise model has been applied; no QPU run has happened. Scaling beyond 5 qubits is untested.
• RQTE is not cryptography yet. No security proof, no attacker model, no NIST SP 800-22 randomness testing. It is a construction that encrypts and decrypts correctly in simulation.
• Axon's bridge gate does not generalise. It works for closed discovery in-sample and fails on the pre-registered held-out case and on sibling-literature separation. Development is halted at that boundary by design.
• XSIG is limited by window count. 16 observational windows is acknowledged as insufficient; shared instrumental systematics across the five channels cannot be excluded; BAO enters through a parametric model rather than raw data.
• Wildfire validation is partial. Coverage-gating and EFFIS comparison exist; there is no published skill score against a held-out fire season, and the pipeline covers one voivodeship.
• Single operator, no institutional affiliation. No lab partners, no funding, no external replication of any result.
• Three repositories contain no code at all (RTANA, Universe, coherence-bridge). They are labelled as such above.

Nothing in the Next or Research rows is claimed to exist.

Working notes and long-form write-ups, published as they are written.

Published via GitHub Pages (workflow) → qhdalabs.github.io

Python (Qiskit · NumPy · SciPy · pandas · NetworkX · GeoPandas · scikit-learn/SHAP) · JavaScript/React · TeX · GitHub Actions · GitHub Pages

Open to collaboration with researchers, engineers, institutions and reviewers — particularly on external replication, noise-model and QPU access for qmnet, and validation data for the wildfire pipeline.

If you find an error in any result here, please open an issue on the relevant repository. Negative findings about this work are as welcome as positive ones.

_{MIT licensed (this repository) · © 2026 Krzysztof W. Banasiewicz}