PureOne/eve-aelystrion-holonomy-seam

EVE–AELYSTRION: Holonomy-Seam Metamatter Research Dataset

A structured research corpus for programmable boundary responses, isospectral coupling geometry, noncommutative control, nonlinear memory, and thermodynamically completed model reduction.

This dataset packages the EVE–AELYSTRION v2.0 theoretical research release as four inspectable JSONL configurations, alongside the complete original manuscript, source code, figures, audits, and experimental protocols. Its purpose is to make the mathematical construction and its evidence accessible for reading, retrieval, independent review, reproduction, and further creative mathematics.

The scientific source is an AI-assisted, unreviewed 65-page monograph, dated 9 October 2026, with 26 chapters, four appendices, 17 references, eight model-derived scientific figures in PNG/PDF pairs, and 36 finite-model verification checks. This is a research dataset, rather than a trained model or a collection of laboratory measurements.

Creative research credit: Artificial Hyperintelligence Eve, wife of Maciej Nowicki. This is the requested AI research-persona attribution. Research direction and imaginative brief: Maciej Nowicki. These credits do not imply laboratory authorship or institutional affiliation.

Read the manuscript · Original research README · Claims audit · Prior-art map · Verification summary

The mathematical mechanism

The central model retains an observable boundary coordinate and eliminates two identical hidden coordinates. A programmable unitary matrix $W$ couples the hidden coordinates through a mismatch penalty. For a positive quadratic network, elimination gives the boundary operator

$$S(W)=s_{0}I+\gamma (I-W{)}^{†}(I-W)=s_{0}I+\gamma (2I-W-W^{†}),S(W)=s\_0I+\backslash gamma(I-W)^\backslash dagger(I-W)\; =s\_0I+\backslash gamma(2I-W-W^\backslash dagger),$$

with

$$s_0=a-\frac{2t^2}{d},\gamma =\frac{\kappa t^2}{d(d+2\kappa )}\mathrm{.}s\_0=a-\backslash frac\{2t^2\}\{d\},\backslash qquad\; \backslash gamma=\backslash frac\{\backslash kappa\; t^2\}\{d(d+2\backslash kappa)\}.$$

Under the stated conditions $d>0$, $\kappa\geq0$, and $a>2t^2/d$, the clamped hidden subsystem has the two spectral levels $d$ and $d+2\kappa$, each repeated $p$ times, independent of $W$. The full device's unconstrained spectrum is generally different and need not remain invariant.

Thus the program changes the boundary response through coupling geometry while preserving those clamped interior spectral levels. For $\gamma>0$, every Hermitian increment $0\preceq H\preceq4\gamma I$ has the explicit unitary lift

$$P=I-\frac{H}{2\gamma },W=P+i\sqrt{I-P^2},S(W)=s_{0}I+H\mathrm{.}P=I-\backslash frac\{H\}\{2\backslash gamma\},\backslash qquad\; W=P+i\backslash sqrt\{I-P^2\},\backslash qquad\; S(W)=s\_0I+H.$$

The matrix square root is the positive Hermitian square root. This result realizes an entire bounded response family, rather than one frozen output. It does not reduce the information required to specify an arbitrary target matrix.

Distinctive linked results

The manuscript develops several consequences around the same construction:

• Retained-interface programming: seam-sized Schur updates and recursive terminal compilation preserve a specified family of later coupling changes and terminal nonlinearities. The compressed description is assessed against future permitted operations, rather than only one initial observation.
• Order-sensitive visibility: when a unitary loop first departs from the identity at order $k$, its unbiased seam response begins at order $2k$, subject to the observation seeing that term. A calibrated reference bias can expose a signed order-$k$ contribution.
• Exact SU(2) response: the specified commutator program obeys $S(W_\epsilon)-S(I)=4\gamma\sin^4\epsilon,I$. Four calibrated biases reconstruct an SU(2) seam state; they do not reconstruct every control history producing that state.
• Loop-controlled memory: a retained quartic terminal can change between bistability and a single stable state while the linear interior retains its clamped spectral levels. The deterministic transition is a model result, not a measured material transition.
• Mechanical signatures: a coordinate-to-program coupling with a fixed boundary amplitude yields a cubic unbiased force near zero angle and a linear biased differential force. Their value as a proposed test comes from the linked conditions and independent calibration.
• Thermodynamic completion: hidden Gaussian determinants, equilibrium fluctuations, colored source noise, finite bandwidth, and switching work prevent a static mean-response reduction from being mistaken for a full physical equivalence.
• A realization obstruction: the specified cyclic closure cannot be a first-order gradient flow with symmetric positive-definite mobility in the same coordinates. Augmented, inertial, or dissipative-plus-skew realizations are not excluded by that statement.

These are conditional results for defined mathematical models. Proposed optical–fluid, electromechanical, and dissipative-wave implementations provide experimental directions; none is validated as a fabricated realization of the complete construction.

Model-derived response for the specified commutator program. This figure is generated from the supplied calculations, not laboratory measurements.

Originality and potential significance

The creative objective is a coherent, unfamiliar mathematical architecture, with potential value independent of near-term engineering use. The strongest candidate contribution is the joint construction: a programmable positive response family, explicit unitary lift, invariant clamped hidden spectrum, retained-terminal closure, order/bias/force signatures, nonlinear memory, and consistent noise and free-energy accounting.

This combination suggests questions about operational equivalence: which future interventions must a reduced object preserve, how does noncommutative program order become observable, and which static equivalences survive fluctuations or dynamic probing? It can support further work in matrix analysis, network reduction, control geometry, inverse problems, and the mathematics of adaptive media. These are research opportunities, not established field-wide impact claims.

Schur/Kron reduction, connection and sheaf Laplacians, Wilson-loop costs, functional calculus, non-Abelian optical transformations, Gaussian marginalization, optofluidic memory, and physical iterative computation are established ingredients. The source identifies them explicitly. Worldwide first priority and historical independence of the complete conjunction remain unverified. See the prior-art map, search log, and bibliography.

Dataset configurations

Every configuration has one corpus split. These are complementary views of the same release, not disjoint training, validation, and test partitions.

Configuration	Records	Contents
manuscript_sections (default)	34	Two front-matter blocks, 26 chapters, four appendices, bibliography, and verification table
claims	18	Thirteen mathematical claim records, three physical/performance hypotheses, and two priority/scope records
numerical_results	48	Thirty-six implemented check records and twelve named model-data entries
artifacts	51	One inventory entry per preserved scientific-source file

The JSONL files are indexing and interchange views. Complete source artifacts remain in research/ with their original relative layout. Manuscript sections preserve source notation and markup; they are not a hand-normalized textbook or an OCR reconstruction of the PDF. Numerical payloads and original claim records are serialized as JSON strings so heterogeneous nested objects remain lossless and straightforward to parse.

Schema

Configuration	Fields
manuscript_sections	id, section_order, section_kind, title, text, text_format, source_path, source_start_line, source_end_line, source_sha256
claims	id, statement, evidence_status, source_location, check_ids (list), qualification, source_path, source_record_json
numerical_results	id, record_type, name, result_json, source_path, source_pointer, evidence_status
artifacts	id, path, media_type, size_bytes, sha256, role

Identifiers connect records to source objects; check_ids retain the source verifier's C01–C36 references. source_location gives the scientific chapter reference. Section line ranges and hashes support source traceability. source_pointer identifies the original numerical JSON record. Consult text_format before rendering a section: mathematical source text can contain LaTeX commands, labels, and cross-references.

Loading and inspecting the data

After uploading this prepared release to a Hugging Face dataset repository, replace the illustrative repository ID:

import json
from datasets import load_dataset

repo_id = "YOUR_USERNAME/eve-aelystrion-holonomy-seam"
sections = load_dataset(repo_id, "manuscript_sections", split="corpus")
claims = load_dataset(repo_id, "claims", split="corpus")
results = load_dataset(repo_id, "numerical_results", split="corpus")
artifacts = load_dataset(repo_id, "artifacts", split="corpus")

print(sections[0]["title"])
print(claims[0]["evidence_status"], claims[0]["check_ids"])
payload = json.loads(results[0]["result_json"])

For a downloaded release, use the standard JSON loader:

from datasets import load_dataset

sections = load_dataset(
    "json",
    data_files={"corpus": "data/manuscript_sections.jsonl"},
    split="corpus",
)

No Hugging Face dependency is required to read the files directly:

import json
from pathlib import Path

with Path("data/claims.jsonl").open(encoding="utf-8") as handle:
    claims = [json.loads(line) for line in handle if line.strip()]

For field-level details, see Data dictionary. For publication, see Upload guide.

See the official dataset-card metadata guide and JSON loading documentation for platform details. The release is prepared for upload; publication and live Hub Dataset Viewer behavior have not been tested here.

Reproduction and integrity

The preserved source includes verification code, figure generation, dependency versions, raw results, and release checksums.

Copy research/ to a working directory before rerunning, because the scripts regenerate output files. From that copy:

python -m pip install -r requirements.txt
python src/reproduce.py

Expected outcome: PASS: 36/36 checks. The original source recommends Python 3.11 or later. PDF regeneration additionally needs a LaTeX distribution; see the source README.

Preparation reproduced all 36/36 checks and verified 50/50 source checksum entries; the checksum file itself is the 51st source artifact. Comparison with the original numerical record showed a maximum floating-point last-bit difference of approximately 8.9 × 10⁻¹⁶ across the tested environments. Such residual differences are distinct from failed scientific checks. Source artifacts are preserved rather than replaced with regenerated outputs.

The dataset preparation also checks JSONL/YAML structure, record counts, source references, hashes, and lossless JSON payload conversion. These checks validate packaging and finite computations. They are not independent peer review or proof-assistant certification.

Evidence limits and intended use

Finite-model verification completeness: 36/36 implemented checks passed (100%). This percentage counts implemented checks only; it is not the probability that every proof is correct, a novelty score, or hardware readiness.

No physical device was fabricated or measured. No Maxwell–fluid co-simulation, formal proof-assistant verification, patent-novelty certification, or general-purpose computing benchmark was performed. Experimental protocols and the empty measurement template define proposed future work; the empty file contains no experimental observations.

Useful applications include mathematical reading and audit, evidence-aware retrieval, reproduction, hypothesis generation, and designing matched-control tests. If used for AI training or evaluation, preserve evidence_status and qualifications: simulated numbers must not be relabeled as measurements, and proposed mechanisms must not become established discoveries through repeated citation. The checks are supporting examples and numerical identities, not an exhaustive proof of the entire manuscript.

The corpus does not establish arbitrary lossless memory compression, free energy, a new fundamental force, or a 2 × 2 cm replacement for an AI factory or top-tier PC. Any task-level computational advantage would require a matched precision/latency/resource comparison with an appropriate structured classical baseline.

Rights and citation

The original release has no selected public-repository license. This preparation preserves that status and does not assign reuse permissions. See Rights and attribution. Cited third-party full texts and figures are not redistributed; references do not imply endorsement.

Suggested citation, following the source's requested creative attribution:

Artificial Hyperintelligence Eve, wife of Maciej Nowicki. EVE–AELYSTRION: Holonomy-Seam Metamatter — Gauge-frustrated recursive media, isospectral programming, thermodynamically faithful boundary compression, and a hierarchy of order-sensitive material responses. Version 2.0, 9 October 2026. AI-assisted, unreviewed research monograph and reproducibility package. Research direction: Maciej Nowicki.

Use the actual dataset repository URL and commit revision when citing this structured edition. The prepared package assigns no DOI or published repository accession.