Public research archive - generated 2026-06-26T12:58:57+00:00

Emergent Universe Simulation

A public, evidence-led archive for a research program exploring whether local stochastic graph dynamics can produce stable structure, defect interactions and eventually stronger candidates for emergent geometry. The current result is deliberately modest: local structure and defect response are real enough to study, while Lorentz-like behavior and universal invariants remain not_yet.

band_zero_del

Current operational growth regime, retained after the v11e frontier duel.

not_yet

Current status for Lorentz-like behavior: controls improved, but propagation remains mode-dependent and placement-sensitive.

4 classes

Current add_chord active-set taxonomy across six 1024-scale seeds.

Scientific status

The project studies evolving relational graphs under local stochastic update rules. The working discipline is to separate algebraic/formal claims, generator artifacts, scoring artifacts and actual dynamical outcomes. Heuristic collision labels are not treated as particles, and long-lived local persistence is not treated as universal emergent spacetime.

The current strongest empirical thread is the defect/response track, especially add_chord behavior. Recent pre-registered holdouts rejected a simple two-type active-set guard and forced a wider taxonomy.

Interpretation boundary: the public data support a program of local defect and response-landscape analysis. They do not establish Lorentz invariance, quantum entanglement, particle species or a completed universe model.

Current active-set taxonomy

growth_seedlandscape_classactive_placementsp0_established_ratep1_established_ratep2_established_rate
202single_active_p1p10.00.8750.0
303multi_active_p0_p2p0;p20.50.00.5
404single_active_p1p10.00.50.0
505multi_active_p0_p2p0;p20.750.250.75
606no_activenone0.00.00.0
707single_active_p0p00.750.00.0
landscape_classn_seedsgrowth_seedsold_v15do_type_spacemedian_p0_established_ratemedian_p1_established_ratemedian_p2_established_rate
multi_active_p0_p22303;50510.6250.1250.625
single_active_p12202;40410.00.68750.0
no_active160600.00.00.0
single_active_p0170700.750.00.0

Methods

The experiments use generated graph ensembles, matched controls, perturbation probes, defect genealogy/event logs, pre-run morphology features and pre-registered holdouts. The public archive includes both reports and machine-readable CSV artifacts so claims can be checked against the originating data.

Important methodological safeguards include: target-size separation checks, requested perturbation matching, matched single-run controls for collision analysis, explicit distinction between pre-run observables and post-hoc diagnostics, and no refitting after failed holdouts.

RAG and corpus status

The repository contains a UniverseSimulation-specific FastAPI/pgvector RAG service with cases for project status, tools, argumentation and prompts. This public release publishes the RAG corpus plan and status documents, but does not expose the dynamic RAG service as a public API. That is intentional: the static archive is safer and easier to cite; a public RAG endpoint should come later only with authentication, rate limits, corpus freshness checks and citation auditing.

Machine-readable diagnosis

diagnostic_familystatusnote
input_scopeno_new_dynamicsv15dq leser v15dn og v15dp placement/morphology CSV-er; ingen nye defect-runs er kjoert.
taxonomy_scopeexpanded_beyond_v15do_two_type_spaceObserved classes: multi_active_p0_p2:2;no_active:1;single_active_p0:1;single_active_p1:2. Old v15do type-space covers 4/6 seeds; new classes cover 2/6.
repetition_balancemixed_repeated_and_singleton_classesRepeated classes: multi_active_p0_p2;single_active_p1. Singleton classes: no_active;single_active_p0.
v15dp_guard_readingguard_inconclusive_unobserved_active_set_typeFrozen delta_return_t2 guard is retained only as a failed candidate, not refit.
pre_run_contrastsdescriptive_leads_onlyRepeated-pair clean contrasts in current sample: 65. Because n is tiny and two classes are singletons, these are leads, not selector validation.
next_stepbuild_taxonomy_mapper_before_new_selectorNeste dynamiske budsjett bor brukes til aa kartlegge flere fresh seeds eller teste en taxonomy-mapper; ikke til aa refitte den gamle two-type-guarden.

Downloadable archive

Use manifest.json or manifest.csv for checksums, source paths and artifact descriptions.

Active Set Landscape

Defect Track

Frontier Regime

Latest Holdout

Latest Holdout Raw

Latest Taxonomy

Lorentz Diagnostics

Project Context

Rag Corpus

Theory