Friday Forum Presentation

<slide header="none" center="both">

Open World Project

Friday Forum Oct. 19, 2012 </slide>

<slide title="Talk Plan">

• Other Projects of Interest
• Motivation and Vision
• Core Elements
• Case Study Projects
• Unified Model of Everything
• Next Steps

</slide>

<slide title="Other Projects of Interest">

• Carbon Transition Working Group
• Himalayan Melt and Flooding
• Peruvian Spatial Fisheries
• Web-Weaver
• Slider Extension

</slide>

<slide title="Introduction"> Many of the human behaviors that drive climate change and environmental degradation are deeply embedded in our society, economy, and government, and are mutually reinforcing. Better modeling of human-natural systems can help in many ways:

• Analyzing feedback loops can help identify leverage points, where small policy changes can have pervasive

impacts.

• Allowing models at diverse scales and contexts to

interact can help scientists integrate knowledge.

• Interactive models can facilitate communication

with policymakers and make complex problems intelligible.

The Open Model is a modeling framework aimed at these issues. </slide>

<slide title="Applicability">

• Systemically intractable due over-determined, reinforcing drives, and spatially heterogeneous.
• Environmental and public health issues
• environmental degradation, agricultural practices in poor countries, obesity, substance abuse, groundwater use, and fishery management
• rebound effects and cross-border shifts (e.g., carbon leakage)

</slide>

<slide title="Big Proposals">

• Fisheries Project
• Climate Behaviors

</slide>

<slide title="Amalgamated Modeling"> </slide>

<slide title="Amalgamated Modeling"> Coupling causes feedback, and models are defined at different scales.

Amalgamated modeling allows models to interact, specialize, and "overlap". </slide>

<slide title="Bayesian Coupling"> </slide>

<slide title="Bayesian Coupling">

       For a variable <latex>\theta</latex> described by multiple models, each
       model provides both a PDF across values at a given time <latex>t</latex>
       when run in isolation, <latex>p(\theta, \bar{S}^i)</latex>, and a
       distribution that includes feedback effects, <latex>p(\theta,
       \tilde{S}^i)</latex>.  The final distribution is
       <latex>\[
       p(\theta | \cdot) \propto p(\theta) \prod_i p(\theta |
         \bar{S}^i)^\lambda p(\theta, \tilde{S}^i)^{1-\lambda}
       \]</latex>

</slide>

<slide title="Multiple Networks"> Models use multiple networks simultaneously

• Different paths on which stocks flow
• Structured disaggregations into classes
• Capturing network properties

</slide>

<slide title="Multiple Networks in Ohio"> </slide>

<slide title="Networked System Dynamics"> </slide>

<slide title="Networked System Dynamics"> </slide>