OusiaResearch/aurelia-causal-events

Aurelia — Causal Event Stream

Per-tick causal events emitted by Aurelia world daemons. Each row is one event from one world, with a (micro|meso|macro|federation) layer, an event_type label, magnitude/valence/confidence scores, and a JSON payload for type-specific deltas. The causal graph is reconstructed from the separate federation_causal dataset.

Provenance

This dataset was generated by Aurelia, a multi-agent civilization simulation operated by Ousia Research. The data in this repository is fully synthetic — no real humans, places, or events are referenced. The simulation engine, schema, and reproducible configurations are all open-sourced; see the GitHub repository for replay instructions.

The five Aurelian worlds (arkos, mirithane, solara, valdris, verge) interact through a federation layer that schedules cross-world effects (trade shocks, migrations, cultural diffusion, diplomatic relations). Each world runs an independent agent-based simulation of NPCs with goals, factions, institutions, and macro-level state transitions.

Schema

Primary file: data/<run_id>/train.parquet

column	type	description
run_id	string	Run label (e.g. phase11-100y)
event_id	string	Globally unique event id (primary key)
tick_number	int64	Simulation tick; 6 ticks = 1 year by default
world_id	string	World this event belongs to (arkos, mirithane, solara, valdris, verge)
layer	string	Causal layer: micro
event_type	string	Event-type label (e.g. wage_dispute, sovereignty_charter)
actor_ids	list	NPC or world ids that triggered the event
target_ids	list	NPC or world ids affected
scope	string	Scope of the event (npc, region, world)
magnitude	float32	Raw magnitude of the event
valence	float32	Signed valence in [-1, 1]
confidence	float32	Engine confidence in [0, 1]
payload	object	Type-specific extra fields as JSON
created_at	float64	Unix timestamp of the simulation tick

Runs included

run_id	total rows	breakdown
phase11-100y	139,476	phase11-100y/train.parquet=139,476
phase11-200y	197,182	phase11-200y/train.parquet=197,182
phase11-bolster-scan-y5	45,056	phase11-bolster-scan-y5/train.parquet=45,056
phase11-cf-solara-aid	45,056	phase11-cf-solara-aid/train.parquet=45,056
phase11-density-100y	133,658	phase11-density-100y/train.parquet=133,658

Total rows across runs: 560,428 (100K<n<1M bucket).

To load a specific run, point data_files at the run's directory.

Loading

from datasets import load_dataset
ds = load_dataset(
    'parquet',
    data_files='aurelia-causal-events/data/phase11-100y/train.parquet',
)['train']
print(ds[0])

Run provenance map

Each run_id corresponds to a deterministic Aurelia simulation with a specific configuration. See the Aurelia Phase 11 comparison report for the demographics and population coefficient-of-variation across these runs.

run_id	config	years	density_diversification
phase11-bolster-scan-y5	baseline (5y)	5	0.0
phase11-100y	baseline (100y)	100	0.0
phase11-200y	baseline (200y)	200	0.0
phase11-density-100y	density-diversification (100y)	100	0.7
phase11-cf-solara-aid	counterfactual: solara federation aid early	5	0.0 (counterfactual intervention)

Licensing

Released under CC-BY-4.0. You may use, share, and adapt the data for any purpose, including commercial, with attribution. Attribution: "Aurelia Simulation Dataset, Ousia Research, 2026."

If you use this dataset in a publication, please cite the companion technical report and link back to the Aurelia repository.

Limitations

• The simulation models abstract civilization dynamics; it is not a forecast of any real-world society.
• All names, world ids, and event labels are fictional.
• The number of NPCs is small (≤ 200 active per world) by design — this is a micro-society scale, not a planet-scale demographic model.
• Results are sensitive to RNG seed and engine version; pin both when reporting.