aurelia-npc-population

# Aurelia — NPC Population Snapshot One row per NPC at the end of a simulation run, with a count of the NPC's own `events` log entries. The properties JSON captures birth, migration, household, and demographic metadata. Useful for agent-level demography, mortality, and migration studies. ## Provenance This dataset was generated by [Aurelia](https://github.com/ousiaresearch/aurelia), a multi-agent civilization simulation operated by [Ousia Research](https://github.com/ousiaresearch). 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>/<world>/train.parquet` | column | type | description | |---|---|---| | `run_id` | string | Run label (e.g. phase11-100y) | | `npc_id` | string | Globally unique NPC id (primary key) | | `name` | string | NPC display name | | `npc_type` | string | Always 'npc' for this dataset | | `location_id` | string | Location at end of run (or null if emigrated) | | `final_state` | string | State at end of run: active, dead, emigrated | | `properties` | object | JSON: born_tick, migrated_tick, household_id, origin_world, target_world, etc. | | `travel_state` | string | In-transit state, or null | | `created_at` | float64 | Unix timestamp of NPC birth | | `updated_at` | float64 | Unix timestamp of last update | | `event_count` | int64 | Count of `events` rows for this NPC | ## Runs included | run_id | total rows | breakdown | |---|---|---| | `phase11-100y` | 6,634 | phase11-100y/arkos/train.parquet=1,646, phase11-100y/mirithane/train.parquet=1,267, phase11-100y/solara/train.parquet=772, phase11-100y/valdris/train.parquet=1,658, phase11-100y/verge/train.parquet=1,291 | | `phase11-200y` | 7,675 | phase11-200y/arkos/train.parquet=1,889, phase11-200y/mirithane/train.parquet=1,434, phase11-200y/solara/train.parquet=898, phase11-200y/valdris/train.parquet=1,910, phase11-200y/verge/train.parquet=1,544 | | `phase11-bolster-scan-y5` | 2,329 | phase11-bolster-scan-y5/arkos/train.parquet=556, phase11-bolster-scan-y5/mirithane/train.parquet=448, phase11-bolster-scan-y5/solara/train.parquet=323, phase11-bolster-scan-y5/valdris/train.parquet=556, phase11-bolster-scan-y5/verge/train.parquet=446 | | `phase11-cf-solara-aid` | 2,329 | phase11-cf-solara-aid/arkos/train.parquet=556, phase11-cf-solara-aid/mirithane/train.parquet=448, phase11-cf-solara-aid/solara/train.parquet=323, phase11-cf-solara-aid/valdris/train.parquet=556, phase11-cf-solara-aid/verge/train.parquet=446 | | `phase11-density-100y` | 6,832 | phase11-density-100y/arkos/train.parquet=1,592, phase11-density-100y/mirithane/train.parquet=1,291, phase11-density-100y/solara/train.parquet=1,049, phase11-density-100y/valdris/train.parquet=1,743, phase11-density-100y/verge/train.parquet=1,157 | Total rows across runs: **25,799** (10K<n<100K bucket). To load a specific run, point `data_files` at the run's directory. ## Loading ```python from datasets import load_dataset ds = load_dataset( 'parquet', data_files={ 'phase11-100y_solara': 'aurelia-npc-population/data/phase11-100y/solara/train.parquet', }, ) print(ds['phase11-100y_solara'][0]) ``` ## Run provenance map Each `run_id` corresponds to a deterministic Aurelia simulation with a specific configuration. See the [Aurelia Phase 11 comparison report](https://github.com/ousiaresearch/aurelia/blob/main/docs/reports/phase11-runs-comparison.md) 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](https://creativecommons.org/licenses/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](https://github.com/ousiaresearch/aurelia). ## 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.