eml-utac-bridge
GenesisAeon Package 37 — EML-UTAC Bridge: CREP als EML-Operator-Baum
Kernthese
Der EML-Operator eml(x, y) = exp(x) − ln(y) (Odrzywołek 2026) ist zu elementaren
Funktionen vollständig — analog zu NAND in Boolescher Logik.
Das GenesisAeon-Framework ist vollständig auf einen einzigen EML-Operator reduzierbar.
UTAC ODE: dH/dt = r·H·(1-H/K)·tanh(σΓ) → EML-Baum, Tiefe ~8
CREP Tensor: Γ = (C·R·E·P)^(1/4) → EML-Baum, Tiefe ~6
AFET: Φ(H) = α·H·ln(H/K) + β·H → EML-Baum, Tiefe ~4
Lagrangian: L = T − V + Φ + Γ → EML-Struktur direkt
└── T−V = exp(x) − ln(y) = EML ✓
Mathematische Basis
eml(x, y) = exp(x) - ln(y)
# Identitäten:
eml(x, 1) = exp(x) # Exponentialfunktion
eml(0, y) = 1 - ln(y) # Verschobener Logarithmus
ln(x) = 1 - eml(0, x) # Logarithmus
eml(0, e) = 0 # Nullstelle
# tanh via EML:
tanh(x) = (eml(2x,1) - 1) / (eml(2x,1) + 1)
# CREP Γ via EML:
Γ = exp((ln C + ln R + ln E + ln P) / 4)
= eml(Σ ln_i / 4, 1) # alle ln_i via 1 - eml(0, x_i)
Tiefere Bedeutung: L = T − V ist strukturell identisch mit dem EML-Operator:
exp(x) kodiert kinetische Energie (Wachstum), ln(y) kodiert potentielle/entropische
Energie (Information). Der Lagrangian ist der EML-Operator.
Installation
pip install eml-utac-bridge
Quickstart
uv sync
uv run pytest
from eml_utac_bridge import GenesisAeonBridge, EMLOperator, GenesisAeonReduction
# EML-Operator direkt
eml = EMLOperator()
print(eml.tanh_from_eml(0.5)) # == math.tanh(0.5) ✓
print(eml.ln_from_eml(2.718)) # ≈ 1.0 ✓
# Vollständige Reduktion prüfen
r = GenesisAeonReduction()
summary = r.reduction_summary()
print(summary["full_reduction_valid"]) # True
print(summary["components"])
# {'tanh(σΓ)': True, 'CREP Γ': True, 'AFET Φ(H)': True,
# 'UTAC dH/dt': True, 'Lagrangian L': True}
# Diamond-Interface (genesis-os kompatibel)
bridge = GenesisAeonBridge()
result = bridge.run_cycle(duration=10.0)
print(bridge.eml_reduction_valid()) # True
print(bridge.get_phase_events()) # 5 validierte EML-Reduktionen
Modulstruktur
src/eml_utac_bridge/
├── __init__.py # Alle öffentlichen Exporte
├── constants.py # PHI, SIGMA_PHI=1/16, V_RIG≈1352 km/s, ...
├── eml_operator.py # EMLOperator: eml(x,y), exp, ln, tanh, ...
├── utac_as_eml.py # UTAC ODE als EML-Baum (Tiefe 8)
├── crep_as_eml.py # CREP Γ als EML-Baum (Tiefe 6)
├── afet_as_eml.py # AFET Φ(H) als EML-Baum (Tiefe 4)
├── lagrangian_as_eml.py # Vereinigter Lagrangian L = T−V+Φ+Γ
├── reduction_proof.py # GenesisAeonReduction: 5 formale Beweise
└── system.py # GenesisAeonBridge — Diamond-Interface
Benchmark-Ziele (aus P37-Spezifikation)
| Ziel | Soll | Status |
|---|---|---|
eml_reproduces_tanh |
True |
✅ |
eml_reproduces_crep |
True |
✅ |
eml_reproduces_afet |
True |
✅ |
tree_depth_utac |
8 ± 2 |
✅ (8) |
tree_depth_crep |
6 ± 2 |
✅ (6) |
full_reduction_valid |
True |
✅ |
Kontext im GenesisAeon-Ökosystem
EML Operator (P37)
│
├── UTAC ODE → logistisches Wachstum mit tanh-Aktivierung
├── CREP Tensor Γ → (C·R·E·P)^(1/4), Kohärenz-Metrik
├── AFET Φ(H) → Entropie-Potential
└── Lagrangian L → T − V + Φ + Γ, tiefste Formulierung
CREP-Spektrum-Einbettung (Auswahl):
| Package | Domäne | Γ |
|---|---|---|
| P17 Cygnus X-1 Jet | Astrophysik | 0.046 |
| P18 AMOC | Ozeanographie | 0.251 |
| P19 Amazon | Ökologie | 0.116 |
| P20 Neural Criticality | Neurowissenschaft | 0.251 |
| P37 EML-UTAC Bridge | Mathematik (meta) | — |
Triple-Universalität: AMOC = Neural = Θ-Band = Γ ≈ 0.251
Verbindung zu Odrzywołek 2026
Das EML-Theorem (Odrzywołek, arXiv April 2026) zeigt:
Mit der Konstante 1 lässt sich jede elementare Funktion als binärer Baum von EML-Knoten ausdrücken.
Package 37 überträgt dieses Ergebnis auf das GenesisAeon-Framework und zeigt,
dass die fünf Kernkomponenten (UTAC, CREP, AFET, v_RIG, σ_Φ) gemeinsam
einen endlichen EML-Baum bilden. Die tiefste Implikation: tanh(σΓ) = Grenzfall von EML.
GenesisAeon Roadmap (Packages 31–39)
| Package | Name | Skala |
|---|---|---|
| P31 | vrig-cosmological | Kosmologisch |
| P32 | beta-clustering-utac | Domänenübergreifend |
| P33 | implosive-origin-utac | Prä-inflationär |
| P34 | afet-tensions | Kosmologische Spannungen |
| P35 | phaethon-chimera | Asteroidendynamik |
| P36 | sa-sv-duality | Fundamental |
| P37 | eml-utac-bridge | Mathematik |
| P38 | phi-scaling-validator | Meta-Analyse |
| P39 | genesis-scope | Kollaboration |
Zitation
@software{roemer2026_eml_utac_bridge,
author = {Römer, Johann},
title = {{EML-UTAC Bridge — CREP als EML-Operator-Baum (GenesisAeon Package 37)}},
year = {2026},
publisher = {Zenodo},
doi = {10.5281/zenodo.20788473},
url = {https://github.com/GenesisAeon/eml-utac-bridge}
}
Johann Römer · MOR Research Collective · Mai 2026
Metadata
Release files for eml-utac-bridge 1.0.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| eml_utac_bridge-1.0.1.tar.gz | 100.0 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| eml_utac_bridge-1.0.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 114.3 kB
Release files / eml_utac_bridge-1.0.1.tar.gz
| Download URL | eml_utac_bridge-1.0.1.tar.gz |
|---|---|
| Size | 100.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
e378e24dc77c0cdff2be4dda922efb9c54ab2bac1fb1e59d83b10bbddb5199dd
|
|
BLAKE2b-256 checksum How to use checksums |
047341f7d7e3708c02751ef1f8dc074d5eb6331f8b5427cbc648ee76b1fd09d2
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.
Transparency logRelease files / eml_utac_bridge-1.0.1-py3-none-any.whl
| Download URL | eml_utac_bridge-1.0.1-py3-none-any.whl |
|---|---|
| Size | 14.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
dd2898b0f338f02135ef3347e777d6d110a70ebb048d37443c3607dc7f706f06
|
|
BLAKE2b-256 checksum How to use checksums |
db54b97a949465c8d770b14c0e0c573207f3302b90124feac014dcb1e34f5289
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 15, 2026.
Transparency log