Coastal Sea Wall & Mangrove Resilience Investment Optimizer with PyMC Bayesian Optimization
Coastal municipalities facing sea-level rise and El Niño storm intensification must allocate limited resilience budgets between hard engineering (sea walls) and natural infrastructure (mangrove restoration). Budget allocation decisions are complicated by non-linear compound flood dynamics and uncertain future storm frequencies. In this challenge, you will formulate a Bayesian decision optimization pipeline using PyMC. You will build probabilistic risk curves under deep uncertainty, modeling storm surge reduction as a function of capital expenditure allocation. Your solution will execute Markov Chain Monte Carlo (MCMC) sampling to solve for optimal capital deployment strategies that minimize Expected Annual Damage (EAD) given fixed budget constraints.
What you are building
Optimize coastal resilience capital allocation between sea walls and mangroves under sea level uncertainty using PyMC Bayesian modeling.
How work is evaluated
Evaluates MCMC convergence and optimal allocation output given municipal constraints.
Datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
mcmc_convergence
Ensures R-hat sampler diagnostic is strictly less than 1.05
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
ead_reduction_optimality
Optimality percentage relative to analytical benchmark • target: 0.96 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Learning goals
Formulate Bayesian probabilistic models using PyMC.
Model Expected Annual Damage (EAD) functions with compound weather uncertainty.
Execute MCMC sampling to infer optimal budget allocation parameters.
Conduct sensitivity analysis across varying climate change projection scenarios.
Resources
Synthetic coastal topological profile dataset containing asset valuations, historical wave energy statistics, and unit costs for coastal protection.
Challenge setup
Evaluates MCMC convergence and optimal allocation output given municipal constraints.
Challenge input
JSON with budget_usd, coastline_km, property_value_at_risk_usd
PyMC
Leading Python framework for MCMC sampling and Bayesian inference.
Evaluated output
JSON with seawall_spend_pct, mangrove_spend_pct, ead_reduction_pct, r_hat
- Ensures R-hat sampler diagnostic is strictly less than 1.05
- Optimality percentage relative to analytical benchmark • target: 0.96 • range: 0-1
- Benchmark: ResilienceOptBench
- Ead Reduction Optimality target: 0.96
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Shows the first task, up to three tools, and up to three evidence items.
View technical recipe
Configured tools
- Lingo · Required
- Milvus · Optional
- Mistral Saba · Optional
Evaluation contract
- mcmc_convergence · Weight 1
- ead_reduction_optimality · Weight 1
Metrics
- ead_reduction_optimality · Target 0.96
Benchmarks
- ResilienceOptBench
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON with budget_usd, coastline_km, property_value_at_risk_usd
Output: JSON with seawall_spend_pct, mangrove_spend_pct, ead_reduction_pct, r_hat
Execution harness: Python
Evaluation tools
- PyMC · Probabilistic Programming
Recipe freeze marker
No recipe freeze time is recorded.