[Forum SIS] Seminar

Simone Padoan simone.padoan a unibocconi.it
Sab 4 Apr 2015 09:34:28 CEST


Dear All,
We are glad to announce the following talk

*************************************************

DEC - Statistics Seminar - Universitą Bocconi,

Thursday, April 9th, 2015,

Room 3-E4-SR03, Via Roentgen 1 - 3rd floor

Time: 12:30pm

(The complete DEC seminars schedule is available at            
                 http://www.unibocconi.eu/statseminar)
*************************************************


Speaker

Mary C. Hill
(University of Kansas)


Title

"Understanding Uncertainty by Exploring Models"


Abstract


Numerical models are used to simulate complex systems in many fields, including environmental systems such as climate and hydrology. Models allow data, conceptualizations, and conservation principles to be brought together in ways that, it is hoped, will provide great insight for system performance and management. Yet often modeling efforts have produced results that are confusing to resource managers and policy makers, and many modeling efforts have been more expensive and less useful than was hoped. As a result, the role played by numerical models in decision making is being reevaluated. This is an important time for modelers to take a step back and consider the current status of numerical models, how we use them to quantify things like prediction, uncertainty, and risk, and how we might proceed in ways that produce more relevant results. In this talk we proceed from the perspective that while much has been accomplished, we are still at the beginning of modeling for resource management. This is because numerical modeling is a very complicated tool and computers have been fast enough to allow expansive exploration of complex models for only a short time. When compared to the evolution of other technological advances such as cars and medicine, it is clear that numerical modeling is early in its development.
Our evaluation of the present status of the field suggests that three challenges compromise the utility of mathematical models of environmental systems: (1) a dizzying array of model analysis methods and metrics make it difficult to choose how to proceed with evaluation of model adequacy, sensitivity, and uncertainty; (2) the high computational demands of many popular model analysis methods (requiring 1,000’s, 10,000s, or more model runs) make them difficult to apply to complex models; and (3) many models are plagued by unrealistic nonlinearities arising from the numerical model formulation and implementation. A strategy is proposed to address these challenges through a careful combination of model analysis and implementation methods. In this strategy, computationally frugal model analysis methods (often requiring a few dozen parallelizable model runs) play a major role, and computationally demanding methods are used for problems where (relatively) inexpensive diagnostics suggest the frugal methods are unreliable. We also argue in favor of detecting and, where possible, eliminating unrealistic model nonlinearities - this increases the realism of the model itself and generally reduces model execution time. Removal of unrealistic model nonlinearities also improves the performance of reduced and surrogate models. Reducing model execution times and using model analysis methods that require fewer model runs enables greater exploration of individual models and the more general question of what methods tend to produce more useful models. Results from recent field and theoretical hydrologic studies will be discussed. We suggest that the strategy proposed would make exploration of models easier and enable a more productive, exciting, and societally consequential future for simulations of complex systems.




-- 
-----------------------------------------------

Dr. Simone Padoan, PhD.
Assistant Professor

Department of Decision Sciences
Bocconi University of Milan
via Roentgen, 1 20136 Milano - Italy
tel. +39 02 5836 5368
Email: simone.padoan a unibocconi.it
Webpage: http://faculty.unibocconi.it/simonepadoan
-----------------------------------------------




Maggiori informazioni sulla lista Sis