Latest summary
Built and verified the PT cycle-1 population review bundle at /home/synthestat/output/runs/PT/pt_population_review_cycle1_dd672afa_seed420987. Bundle is contract-complete for internal research review, with 8 persons, 8 households, 8 dwellings, 8 dwelling-building assignment rows, HARD constraints passing exactly, and explicit unavailable markers for hidden-population overlays and work/school assignments.
Body
Build the Portugal (PT) population synthesis QA cycle 1 review bundle for internal research review.
Project root: /home/synthestat
Country: PT
Target geography: finest available official geography
Release mode: internal research review
Required contract:
- Follow /home/synthestat/docs/contracts/population_review_bundle.md exactly.
- Emit bundle under /home/synthestat/output/runs/PT/<deterministic_run_id>/ with required files or explicit unavailable.json files and reasons.
- Minimum required bundle files include persons, households, dwellings or unavailable, building assignments or unavailable, hidden population overlays or unavailable, work/school assignments or unavailable, 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, and model_notes.md.
Scope:
- Build the best available 1:1 synthetic population for PT: persons in households, households in dwellings, dwellings in real houses/buildings where evidence supports it.
- Include separate uncertainty-aware overlays, or explicit unavailable/degraded markers, 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.
Non-negotiable guardrails:
- HARD constraints never break; if they cannot be satisfied, stop with explicit BLOCKED/invalid notes rather than silently relaxing.
- FIRM/SOFT/GUIDE semantics must be explicit; weak evidence means relaxed constraints/wider uncertainty, not fake precision.
- Hidden-population overlays must not silently rewrite de jure constraints.
- Every modelled estimate must have uncertainty bounds/quality flags.
- Every missing source, relaxed constraint, failed download, degraded zone, unavailable layer, and model-driven fine geography field must be explicit in manifest/diagnostics/model_notes.
- Maintain provenance: source IDs, retrieval timestamps where available, geography levels, reference periods, source universe, quality flags.
- Reuse existing Synthestat modules/generators; do not duplicate country-specific synthesis logic.
Allowed writes:
- /home/synthestat/output/runs/PT/<run_id>/
- /home/synthestat/workspace/manager_handoffs/modeller/ for concise build status and missing-requirements notes
- Optional durable memo under /home/synthestat/docs/wiki/outputs/ if useful
Definition of done:
- Bundle exists and is contract-complete or explicitly marks unavailable layers.
- build_manifest.json includes all minimum fields in the contract.
- constraint_residuals.json proves HARD constraints exact or documents invalid-output blocker.
- uncertainty_summary.json and model_notes.md make modelled/weak layers reviewable.
- Complete with run_id, bundle path, key counts, HARD/FIRM/SOFT status, unavailable/degraded layers, and tests/validation commands run.