Body
Country: SE (Sweden)
Project root: /home/synthestat
Target geography: finest available official geography
Release mode: internal research review
Parent orchestrator: t_0c611b3b
Mission:
Generate a complete Synthestat population review bundle for the best currently possible 1:1 synthetic population for SE: persons in households, households in dwellings, dwellings in real houses/buildings where available. Include uncertainty-aware, separately flagged overlays where evidence supports them for homelessness, refugees/asylum seekers, Ukrainian displaced people, Syrian refugees, undocumented/seasonal populations, students, and institutional populations. Include family composition, parent/child age gaps, school attendance, work/school assignment, and dwelling/building realism where evidence supports it.
Mandatory context to read before work:
- /home/synthestat/docs/SOUL.md (if stale/misaligned, use the injected Synthestat SOUL rules from manager profile: uncertainty first-class, HARD/FIRM/SOFT/GUIDE precedence, no silent degradation, provenance required)
- /home/synthestat/docs/contracts/population_review_bundle.md
- /home/synthestat/docs/specs/research_knowledge_base.md
- Existing SE source/registry/output artifacts under /home/synthestat/config, /home/synthestat/output, /home/synthestat/docs/wiki, and /home/synthestat/workspace/manager_handoffs as relevant.
Required output:
Create a complete review bundle under a deterministic run directory matching docs/contracts/population_review_bundle.md, preferably:
/home/synthestat/output/runs/SE/<run_id>/
where <run_id> is deterministic and declared in build_manifest.json.
The bundle must include, or explicitly mark unavailable with reason, all contract files:
- synthetic_persons.parquet|csv
- synthetic_households.parquet|csv
- synthetic_dwellings.parquet|csv or unavailable.json
- synthetic_building_assignments.parquet|csv or unavailable.json
- hidden_population_overlays.parquet|csv or unavailable.json
- work_school_assignments.parquet|csv or unavailable.json
- build_manifest.json
- constraint_residuals.json
- distribution_diagnostics.json
- household_diagnostics.json
- dwelling_building_diagnostics.json
- assignment_diagnostics.json
- geography_quality_tiers.json
- uncertainty_summary.json
- source_provenance.json
- model_notes.md
Non-negotiables:
- HARD constraints must not break. If exact hard constraints cannot be satisfied, stop and produce a valid diagnostic bundle rather than faking outputs.
- FIRM/SOFT/GUIDE precedence must be explicit in diagnostics and residuals.
- Model-based estimates without uncertainty bounds are invalid.
- Hidden-population overlays must not silently rewrite de jure population constraints; keep them separate and flagged.
- Fine-geography occupation/industry/hidden-population estimates are model-driven unless directly measured and must be flagged.
- Every missing source, relaxed constraint, degraded zone, unavailable artifact, and modelled estimate must be explicit.
- Reuse existing Synthestat modules and generators; do not duplicate country-specific synthesis logic.
Definitions of done:
1. Bundle directory exists and satisfies the population review bundle contract.
2. build_manifest.json names country=SE, run_id, created_at, project_root, git/dirtiness marker, random_seed, source catalogue/geography/classification versions, constraints_used, constraints_relaxed, zones_degraded, hidden_population_scope, assignment_scope, and known_limitations.
3. source_provenance.json includes source IDs, retrieval timestamps when known, geography levels, reference periods, and quality/status flags.
4. uncertainty_summary.json covers all modelled or weakly measured components.
5. model_notes.md states what is measured vs constrained vs modelled vs unavailable, and gives an explicit candidate reviewer verdict suggestion only as a modeller self-assessment, not the final verdict.
6. Complete with metadata containing bundle_path, run_id, files_written, tests_or_checks_run, and known_limitations.
Do not create reviewer or follow-up tasks yourself; the manager has already routed review as a dependent card.