Multiobjective Groundwater Management Using Evolutionary Algorithms

Sustainable management of groundwater resources is of crucial importance for regions where freshwater supply is naturally limited. Long-term planning of groundwater usage requires computer-based decision support tools: on the one hand, they must be able to predict the complex system dynamics with sufficient accuracy, on the other, they must allow exploring management scenarios with respect to different criteria such as sustainability, cost, etc. In this paper, we present a multiobjective evolutionary algorithm for groundwater management that optimizes the placement and the operation of pumping facilities over time, while considering multiple neighboring regions which are economically independent. The algorithm helps in investigating the cost tradeoffs between the different regions by providing an approximation of the Pareto-optimal set, and its capabilities are demonstrated on a three-region problem. The application of the proposed methodology can also serve as a benchmark problem as shown in this paper. The corresponding implementation is freely available as a precompiled module at http://www.tik.ee.ethz.ch/pisa.     

A generalisation of the interval Newton single-step method for nonlinear systems of equations

In this paper interval iteration methods for solving large nonlinear systems of equations are considered. Already well-known methods are combined to a new one, whose enclosing properties are better than those of previous methods. Convergence of this new method is shown, based on a new convergence proof for the interval Newton single-step method. The central concept in this case is the fixpoint inverse of an interval matrix. Practical tests with nonlinear systems of equations arising from discretisation of certain elliptic partial differential equations show the efficiency of the new method.     

Recursive retrospective revaluation of causal judgments.

Recursive causal evaluation is an iterative process in which the evaluation of a target cause, T, is based on the outcome of the evaluation of another cause, C, the evaluation of which itself depends on the evaluation of a 3rd cause, D. Retrospective revaluation consists of backward processing of information as indicated by the fact that the evaluation of T is influenced by subsequent information that is not concerned with T directly. Two experiments demonstrate recursive retrospective revaluation with contingency information presented in list format as well as with trial-by-trial acquisition. Existing associative models are unable to predict the results. The model of recursive causal disambiguation that conceptualizes the revaluation as a recursive process of disambiguation predicts the pattern of results correctly. (PsycINFO Database Record (c) 2010 APA, all rights reserved)