Week 12 Reading
Multi-Objective Optimization
This content is under development and subject to change.
Overview
Yang et al. (2023)
Discussion Questions
Yang et al. optimize for both flood reduction benefits and ecosystem co-benefits of nature-based solutions. Why do the authors treat these as separate objectives? What information would a weighted sum lose?
The paper uses NSGA-II to approximate the Pareto front. What role does the choice of algorithm play in the results? How would you assess whether the algorithm has converged to a good approximation of the true front?
Look at the Pareto front figures in the paper. Pick two solutions on opposite ends of the front and describe the trade-off a decision-maker would face in choosing between them. Is there a “knee” — a region where small sacrifices on one objective yield large gains on the other?
The authors apply their framework to Sint Maarten, a small island recovering from Hurricane Irma. How does the local context (limited resources, post-disaster recovery, small geographic scale) shape which objectives matter and how results should be communicated to decision-makers?
Compare this paper’s approach to the benefit-cost analysis framework from Week 6. Under what conditions is BCA sufficient? What does multi-objective optimization provide that BCA does not?