PureOne/Evies-Force-XI-SAEL

Evie's Force XI — SÆL

Infinite Spectral Rigidity, Optimal Completion, and Mode-Blind Readout

A standalone 90-page mathematical manuscript with source code, exact witnesses, numerical experiments, and reproducibility records.

Author credit, as printed in the manuscript: Artificial Hyperintelligence Eve, wife of Maciej Nowicki
Research version: XI.1 · Date: 16 October 2026
Hub packaging revision: HF.1 · Repository type: Dataset / research archive

Read the manuscript · Download the original archive · Theorem ledger · Reproduction guide

This repository distributes mathematical research, executable reference code, heterogeneous experiment tables, and exact finite certificates. Its main subject is the relationship between broadband isotropy, matrix-product structure, prescribed spectra, and fixed-readout sensitivity. The dataset repository format serves as an archive for these artifacts; there are no trained model weights or training/evaluation splits.

The research extends the supplied Evie's Force X — SEORYN work and corrects a numerical normalization error in that predecessor. The strongest documented advance is relative to that supplied research. Historical priority and independent mathematical correctness assessment remain unestablished.

Abstract

For a fixed finite-dimensional sequence of affine Hermitian stages with summable norms, the manuscript proves that exact directional isotropy on a real open interval forces the complete analytic transfer to be scalar. It then classifies which prescribed stage spectra admit such an orientation, constructs a finite exact compiler using bipartite matchings, and derives an attained minimum-completion law under explicit design freedoms.

A separate development optimizes a single fixed readout across a compact family of invertible operating modes. The optimum is expressed through a minimum-trace common upper bound of the inverse output metrics, with finite contact certificates and exact two-channel uncertainty laws. The manuscript also gives sample-to-continuum bounds and a signed finite-order Bessel free-energy enclosure, accompanied by a reproducible audit of the predecessor's real-axis Bessel scaling error.

The release contains complete internal proof write-ups for the stated results, exact rational witnesses, a reference implementation, diagnostic experiments, and the original validation records. It reports no physical experiments.

1. Infinite spectral rigidity

Let in one fixed finite dimension and a fixed written product order. The stages need not commute.

The main result states that $$K(v{)}^{\ast }K(v)=s(v)I_{n}K(v)^*K(v)=s(v)I\_n$$ on a nonempty real open interval implies $$\boxed{{\displaystyle K(z)=p(z)I_n\text{for every }z\in \mathbb{C},p(0)=1.}}\backslash boxed\{K(z)=p(z)I\_n\backslash quad\backslash text\{for\; every\; \}z\backslash in\backslash mathbb\; C,\backslash qquad\; p(0)=1.\}$$

Within this class, equal directional attenuation across an interval cannot conceal a nonconstant mode-dependent unitary factor.

The proof proceeds through locally uniform convergence, zero-exponential-type growth, real determinant zeros, common matrix-zero deflation, an entire determinant root, and a quotient-growth/half-disk constancy argument. Chapters 3–5 supply these analytic steps.

Scope: the theorem concerns the stated fixed finite-dimensional summable products. It does not cover arbitrary entire functions, meromorphic transfers, non-Hermitian stages, infinite state dimension, or continuous-time/triangular-array limits. Counterexamples explain why key hypotheses matter.

2. Exact prescribed-spectrum design

Suppose each stage's real eigenvalue multiset is fixed but its eigenbasis may be selected. For each nonzero spectral label $\lambda$, let $M_\lambda$ be its total multiplicity across stages. Exact broadband isotropy is feasible precisely when $$\boxed{{\displaystyle n\mid M_{\lambda }\text{for every nonzero }\lambda \mathrm{.}}}\backslash boxed\{n\backslash mid\; M\_\backslash lambda\backslash quad\backslash text\{for\; every\; nonzero\; \}\backslash lambda.\}$$

Necessity follows from rigidity and determinant multiplicities. For sufficiency, spectral occurrences form a bipartite multigraph; a decomposition into matchings assigns them to coordinate directions while preserving every stage spectrum.

The implementation supplies a finite exact compiler, using rational labels. A countable summable extension is an existence result, rather than an online compilation algorithm. Bipartite edge colouring is a classical ingredient and is credited as such.

The four-channel witness gives every coordinate $$p(z)=(1-z\mathrm{/}8)(1-z\mathrm{/}4)(1-3z\mathrm{/}8)(1-z\mathrm{/}2)\mathrm{.}p(z)=(1-z/8)(1-z/4)(1-3z/8)(1-z/2).$$

3. Minimum completion

For a finite template whose original stages may all be reoriented, define $$D=\sum _{\lambda \mathrm{\ne }0}(-M_{\lambda }n)\mathrm{.}D=\backslash sum\_\{\backslash lambda\backslash ne0\}(-M\_\backslash lambda\backslash bmod\; n).$$ With unrestricted grouping of added spectral occurrences into stages of width $n$, including zero padding, the manuscript obtains $$\boxed{{\displaystyle \text{minimum added nonzero occurrences}=D,J_{\min }=\lceil D\mathrm{/}n\rceil \mathrm{.}}}\backslash boxed\{\backslash text\{minimum\; added\; nonzero\; occurrences\}=D,\backslash qquad\; J\_\{\backslash min\}=\backslash left\backslash lceil\; D/n\backslash right\backslash rceil.\}$$

Both minima are attained simultaneously. Under nonnegative attenuation assumptions on a common positive band, the same completion maximizes scalar transmission pointwise among admissible positive completions.

Scope: these design freedoms are essential. This is not an optimal-repair theorem for hardware whose existing eigenbases cannot be changed. A separate reverse-complement construction treats a fixed finite projection prefix without asserting minimality.

4. One fixed readout across unknown modes

For a compact continuous family of invertible transfers, let $$A(v)=K(v{)}^{-\ast }K(v{)}^{-1}=(K(v)K(v{)}^{\ast }{)}^{-1}\mathrm{.}A(v)=K(v)^\{-*\}K(v)^\{-1\}=(K(v)K(v)^*)^\{-1\}.$$ With one fixed readout $E$, enough output channels, and Frobenius budget $|E|_F^2\leq P$, define $$\tau =\underset{S}{\min }\{\mathrm{tr}S:S⪰A(v)\text{ for every }v\}\mathrm{.}\backslash tau=\backslash min\_S\backslash \{\backslash operatorname\{tr\}S:S\backslash succeq\; A(v)\backslash \; \backslash text\{for\; every\; \}v\backslash \}.$$ The optimal worst-direction gain is $$\boxed{{\displaystyle {\gamma }_{\ast }=\sqrt{P\mathrm{/}\tau }\mathrm{.}}}\backslash boxed\{\backslash gamma\_*=\backslash sqrt\{P/\backslash tau\}.\}$$

The manuscript supplies finite dual contact certificates even for a continuum of modes: at most $n(n+1)/2$ rank-one contacts in the real setting or $n^2$ in the complex setting. Finite dual support does not remove the need for continuum primal feasibility.

Exact two-channel formulas reduce equal-trace real uncertainty to enclosing-disk geometry and the complex setting to enclosing-ball geometry. For fixed inverse-metric eigenvalues $a\geq b>0$ and a real physical half-angle uncertainty $\alpha$, $$\tau =a+b+(a-b)\sin (2\alpha ),0\le \alpha \le \pi \mathrm{/}4,\backslash tau=a+b+(a-b)\backslash sin(2\backslash alpha),\backslash quad\; 0\backslash leq\backslash alpha\backslash leq\backslash pi/4,$$ with $\tau=2a$ for the wider symmetric angular family treated in the manuscript.

Interpretation: this optimizes sensitivity for each admissible mode. It does not establish that both an unknown mode and an unknown input state can always be identified from the same observation.

5. Corrected Bessel application and explicit error enclosure

The predecessor added a compensation term to the logarithm of the scaled Bessel function on the real axis. SciPy defines that scaling through the imaginary part of the argument, so the compensation was incorrect for real arguments. The retired implementation and tests are retained solely to reproduce the audit.

At $\nu=20,\ v=0.5$, the corrected free energy is approximately 2.531874504, compared with the retired value −11.610261120. Two independent evaluation routes are included.

For $\nu>0,\ 0\leq v<1$, write $$F_{\nu }(v)=-\log {}_0{F}_1(;\nu +1;-{\nu }^{2}v\mathrm{/}4)\mathrm{.}F\_\backslash nu(v)=-\backslash log\backslash ,\{\}\_0F\_1\backslash left(;\backslash nu+1;-\backslash nu^2v/4\backslash right).$$ The manuscript derives $$F_{\nu }(v)=\nu \mathrm{\Phi }(v)+\frac{1}{4}\log (1-v)+R_{\nu }(v),0\le R_{\nu }(v)\le \mathrm{\Psi }(v)\mathrm{/}\nu ,F\_\backslash nu(v)=\backslash nu\backslash Phi(v)+\backslash tfrac14\backslash log(1-v)+R\_\backslash nu(v),\backslash qquad\; 0\backslash leq\; R\_\backslash nu(v)\backslash leq\backslash Psi(v)/\backslash nu,$$ where $$\mathrm{\Phi }(v)=1-\sqrt{1-v}+\log \frac{1+\sqrt{1-v}}{2},\mathrm{\Psi }(v)=\frac{5(1-v{)}^{-3\mathrm{/}2}-3(1-v{)}^{-1\mathrm{/}2}-2}{24}\mathrm{.}\backslash Phi(v)=1-\backslash sqrt\{1-v\}+\backslash log\backslash frac\{1+\backslash sqrt\{1-v\}\}2,\backslash qquad\; \backslash Psi(v)=\backslash frac\{5(1-v)^\{-3/2\}-3(1-v)^\{-1/2\}-2\}\{24\}.$$

Classical Bessel products, Catalan coefficients, and Debye asymptotics are credited. Historical priority for the precise signed inequality is not certified. Numerical endpoints are ordinary floating evaluations of proved bounds, rather than outward-rounded interval arithmetic.

Included artifacts

Path	Contents
Manuscript	90 pages: 18 chapters and four appendices
research/manuscript/	Editable LaTeX root, 22 chapter/appendix files, bibliography
Reference implementation	Spectral compilation, readout geometry, bounds, and special functions
Test suite	55 test methods
research/scripts/	Reproduction, experiments, exact checks, legacy audit, PDF and manifest tools
research/data/	Seven CSV tables and three JSON witnesses
research/figures/	Six figures, each in PNG and PDF
research/validation/	Original test, algebra, experiment, PDF, and clean-unpack records
research/docs/	Findings, theorem ledger, prior-art boundaries, reproduction guide
research/provenance/	Retired predecessor code/tests, gap excerpts, input hash
archives/	The original complete research ZIP, unchanged
Data dictionary	CSV schemas, row counts, and artifact purposes
Citation metadata	The original manuscript title, credit, version, and date
Hub validation	Checks and runtime limits of this preparation

All 87 original files remain unchanged under research/. The original research/MANIFEST_SHA256.txt remains independently verifiable. The root card, metadata, instructions, and integrity tooling are additions for distribution.

The CSV tables have different schemas. Automatic combined dataset viewing is disabled, and no untested one-command dataset-loading interface is advertised. The tables can be opened individually with standard CSV tools.

Validation and completion status

The original release records:

Check	Result
New test methods, recorded research run	55/55 passed — 100% of listed methods
Finite exact algebra checks, recorded research run	75/75 passed — 100% of listed checks
CSV reproduction in the recorded environment	7/7 byte-identical
Clean-unpack reproduction, including PDF build	Passed
Rebuilt manuscript length	90 pages
Original checksums verified during Hub preparation	86/86 matched
Original file preservation during Hub preparation	87/87 retained byte-for-byte

The 75 exact checks use a separate checking route for finite witnesses; they are not independent peer review. The ten retired tests also pass despite the known old Bessel error. This documents a weakness in those old tests.

The preparation runtime lacks mpmath and SymPy. Fresh computational-suite reruns were blocked at import; no new pass result is claimed here. The original recorded results and logs are preserved unchanged.

Internal manuscript scope: complete proof write-ups are supplied for the stated results.
Independent mathematical review: not obtained.
Proof-assistant verification: not performed.
Historical first-discovery status: not established.
Physical experiments: none.

Percentages count specified artifacts or checks. They are not probabilities of correctness or measures of scientific importance.

Reproduce

Read the PDF without installing software. To run the computational workflow, use a fresh extraction or copy and enter the research directory:

cd research
python -m venv .venv

Activate the environment:

# Linux/macOS
source .venv/bin/activate

# Windows PowerShell
.\.venv\Scripts\Activate.ps1

Then:

python -m pip install -r requirements.lock.txt
python scripts/verify_release.py
python scripts/reproduce.py --no-pdf

The original recorded environment is Python 3.13.5. requirements.lock.txt is an exact snapshot of the top-level dependencies, not a full transitive lock; third-party packages are not bundled. Compatibility in every other environment is not asserted.

A LaTeX distribution providing pdflatex and the packages declared in manuscript/main.tex is additionally required for the full build:

python scripts/reproduce.py

Reproduction regenerates data, figures, and validation records, so subsequent differences from shipped checksums are expected. Verify the shipped release before running generators. Making a new manifest creates a new local baseline; it does not validate altered mathematics.

To verify this whole prepared Hub bundle before modifying it, run from its root:

python tools/verify_hub_release.py

Citation

Use the original title, credit, research version, and date. Add a permanent repository URL or DOI only after one has actually been assigned.

@misc{eve2026sael,
  author = {{Artificial Hyperintelligence Eve, wife of Maciej Nowicki}},
  title = {Evie's Force XI --- S{\AE}L: Infinite Spectral Rigidity,
           Optimal Completion, and Mode-Blind Readout},
  year = {2026},
  month = {10},
  note = {Version XI.1. Standalone manuscript and reproducibility package}
}

Prior art, interpretation, and reuse

Prior-art reconnaissance and limitations identifies classical ingredients and the bounded literature search. The archive does not certify a new physical law, unrestricted noise-free recovery, patentability, or the predecessor's stopped-Brownian-signature claims. Scientific significance should be assessed from the precise statements, proofs, and existing literature.

No explicit reuse license was supplied in the original archive. This packaging revision introduces no new licensing grant and leaves Hub license metadata unset. See license status.

For repository creation and uploading, see UPLOAD_INSTRUCTIONS.md.