Confronting tipping points: Can multi-objective evolutionary algorithms discover pollution control tradeoffs given environmental thresholds?. (November 2015)

• Record Type: Journal Article
• Authors: Ward, Victoria L.; Singh, Riddhi; Reed, Patrick M.; Keller, Klaus
• Is Part Of: Environmental modelling & software. Volume 73(2015:Nov.)
• Page Start: 27
• Page End: 43
• Publication Date: 2015-11
• Subjects: Risk management -- Environmental thresholds -- Tipping points -- Multi-objective decision making -- Algorithm benchmarking -- Lake problem benchmark; Environmental monitoring -- Computer programs -- Periodicals; Ecology -- Computer simulation -- Periodicals; Digital computer simulation -- Periodicals; Computer software -- Periodicals; Environmental Monitoring -- Periodicals; Computer Simulation -- Periodicals; Environnement -- Surveillance -- Logiciels -- Périodiques; Écologie -- Simulation, Méthodes de -- Périodiques; Simulation par ordinateur -- Périodiques; Logiciels -- Périodiques; Computer software; Digital computer simulation; Ecology -- Computer simulation; Environmental monitoring -- Computer programs; Periodicals; Electronic journals; 363.70015118
• DOI: 10.1016/j.envsoft.2015.07.020
• Languages: English
• ISSNs: 1364-8152

Abstract

Abstract: This study contributes a stochastic, multi-objective adaptation of the classic environmental economics Lake Problem as a computationally simple but mathematically challenging benchmarking problem. The Lake Problem considers a hypothetical town by a lake, which hopes to maximize its economic benefit without crossing a nonlinear, and potentially irreversible, pollution threshold. Optimization objectives are maximize economic benefit, minimize phosphorus in the lake, maximize the probability of avoiding the pollution threshold, and minimize the probability of drastic phosphorus loading reductions in a given year. Uncertainty is introduced through a stochastic natural phosphorus inflow. We performed comprehensive diagnostics using six algorithms: the Borg multi-objective evolutionary algorithm (MOEA), MOEA/D, epsilon-MOEA, the Non-dominated Sorting Genetic Algorithm II (NSGAII), epsilon-NSGAII, and Generalized Differential Evolution 3 (GDE3) to evaluate their controllability, reliability, efficiency, and effectiveness. Our results show only the self-adaptive search of the Borg MOEA was capable of performing well on this nontrivial benchmarking problem. Highlights: Thresholds have not been heavily addressed from a multi-objective standpoint. We pose a multi-objective environmental threshold problem for benchmarking. Even on this simple problem, many multi-objective evolutionary algorithms fail.