{"data":{"kind":"file","path":"README.md","version_id":"u2f00k839og96nx9c8vjbax3","entry":{"name":"README.md","path":"README.md","is_directory":false,"size":2555,"modified_at":"2026-08-12T06:42:37.163000","content_hash":"89bcdec5a418be4ea9e875275771a11351586c52e206d4e553d1fc3cd980475d"},"entries":[],"content":"# SIR Model Agent — Epidemic Simulation & Policy Advisor\n\nAn RL environment that tests an agent's ability to understand, implement, and analyze\ncompartmental epidemic models (SIR, SEIR, SIS) and recommend public health interventions.\n\n## What This Environment Tests\n\n- **Mathematical Reasoning**: SIR/SEIR ODE systems, R0 calculation, herd immunity threshold\n- **Code Generation**: Python implementations using scipy/numpy for epidemic simulations\n- **Parameter Estimation**: Fitting models to observed case data\n- **Intervention Analysis**: Vaccination strategies, social distancing, border closures\n- **Multi-Patch Models**: Coupled metapopulation systems (two-city epidemics)\n- **Agent-Based Modeling**: Spatial epidemic simulations with individual agents\n- **Uncertainty Quantification**: Monte Carlo sensitivity analysis of model parameters\n- **Game Theory**: Vaccination as a public goods game (Nash vs social optimum)\n\n## Task Types\n\n| Category | Tasks | Key Skills |\n|----------|-------|------------|\n| SIR Basics | ODE implementation, R0, peak analysis | Math + Code |\n| SEIR Extension | Exposed compartment, incubation period | Model extension |\n| Parameter Fitting | Curve fitting from case data | Optimization |\n| Vaccination Analysis | SIR-V model, herd immunity threshold | Policy modeling |\n| Multi-City | Two-patch metapopulation with migration | Coupled systems |\n| Agent-Based | Grid-based spatial epidemic simulation | OOP + Simulation |\n| Sensitivity | Monte Carlo, Sobol analysis | Statistics |\n| Game Theory | Nash equilibrium vaccination | Economics + Epidemiology |\n\n## Scoring Functions\n\n| Function | Weight | What It Checks |\n|----------|--------|----------------|\n| `check_sir_equations` | 0.20 | Correct ODE formulation |\n| `check_code_quality` | 0.25 | Valid Python with scipy/odeint |\n| `check_r0_calculation` | 0.20 | R0 = beta/gamma, thresholds |\n| `check_simulation_results` | 0.20 | Peak counts, timing, totals |\n| `check_intervention_analysis` | 0.15 | Policy recommendations with data |\n\n## Configuration\n\n```python\nimport verifiers as vf\n\nenv = vf.load_environment(\"yard/sir-model-agent\")\n# or\nenv = vf.load_environment(\"yard/sir-model-agent\", env_args={\"level\": 1})  # eval tasks\n```\n\n## Levels\n\n- **Level 0** (default): 10 training tasks covering all epidemic modeling topics\n- **Level 1**: 3 evaluation tasks (school outbreak, two-country border closure, government advisory)\n\n## Dependencies\n\n- verifiers\n- datasets\n- numpy (for model implementations)\n- scipy (for ODE solving and optimization)\n\n## License\n\nMIT\n","encoding":"utf-8","truncated":false,"total_bytes":2555},"status":null}