Monday, March 9, 2015

Exploratory Modeling and GoldSim

Posted by Nick Martin

There is an interesting recent journal article, “Exploratory Modeling: Extracting Causality From Complexity” (Larsen et al. 2014), which describes the usefulness and success of exploratory modeling in the earth sciences. Although this article does not mention GoldSim, it provides a nice explanation of modeling philosophies which work well in GoldSim. These philosophies emphasize a “top-down” approach to modeling complex systems and focus on the representation of uncertainty in addressing “big-picture” issues.

In the Larsen et al. (2014) article, exploratory modeling is defined as a philosophical approach to identifying the underlying “simple processes” whose physical interactions, within certain parameter ranges, can give rise to complex phenomena. The focus on physical processes provides an earth science spin on the classic operations research exploratory modeling definition; “Exploratory modeling is using computational experiments to assist in reasoning about systems where there is significant uncertainty (Bankes, 1993, p. 435).” 

“Exploratory Modeling: Extracting Causality From Complexity” goes on to explain how the exploratory modeling approach is best implemented and has been most successfully used in the past in specific applications in the earth sciences. “Exploratory modeling works by intentionally leaving out or simplifying physical details like turbulence … , not necessarily because those details are poorly understood or computationally demanding but because doing so helps elucidate the essential causative processes …. Key to developing a successful exploratory model is selection of the appropriate level of detail, which should be just sufficient to reproduce the emergent phenomena of interest and any secondary details useful for selection between multiple models. Exploratory models generally work at the coarsest scale of process description commensurate with the problem being investigated (Larsen et al. 2014, p. 285).”


Figure 1: Graphical depiction of the interaction of complex processes and uncertainty in the "Work under Pressure" GoldSim model
As a final note related to exploratory modeling and GoldSim, the 2007 GoldSim User Conference presentation "Modeling Complex Systems – How Much Detail is Appropriate?" provides a good description, specifically related to GoldSim, of a process that can be used to determine the necessary level of detail for an exploratory model in general and a Goldsim model in particular. This presentation also provides some beneficial tips on exploratory modeling using GoldSim.

References:

Banks, S. (1993), Exploratory Modeling for Policy Analysis, Operations Research, v. 41, n.3, p. 435-449.

Esch, D. (2007), Modeling Complex Systems – How Much Detail is Appropriate?, Presented at the 2007 GoldSim User Conference, San Francisco, CA.

Larsen, L., Thomas, C., Eppinga, M., and T. Coulthard (2014), Exploratory Modeling: Extracting Causality From Complexity, EOS, v. 95, p. 282-292, doi: 10.1002/2014EO320001.


No comments:

Post a Comment