Mathematical Sciences: Reduced-Form Modelling for Integrated Assessment

综合评估越来越多地被用作在环境问题上弥合科学与政策之间差距的一种方式。一个复杂问题的每个方面的模型往往过于繁琐，无法直接集成。相反，通常更有吸引力的是为每个组件模型开发简化形式的模型，这些模型足够小，可以理解、灵活，并且容易相互连接。简化形式的模型以更紧凑和可理解的形式提供与从中提取模型大致相同的行为。这个项目的目标是开发和评估技术，以帮助建模者分析他们的模型并构建简化形式的模型。具体目标是：1。设计建模人员的工作平台并制作原型，以方便对不同类型和大小的现有模型的输入、输出和假设进行规范和文档化；2. 提供一系列现有的和新的抽样技术，以产生模拟这些模型的情景；3. 提供一系列现有的和新的模型分析技术，以识别关键输入和参数，识别非单调和定性不同行为的区域，从而帮助开发简化形式的模型；和4。通过将这些技术应用于简化形式模型和综合评估模型的开发，对这些技术进行实验评估，从而指导哪些技术最有价值，适用于哪种模型。本项目在综合评估方法和模型资助机会下得到支持。

Integrated assessment is increasingly used as a way to bridge the gap between science and policy in environmental problems. Models of each aspect of a complex problem are often too cumbersome to integrate directly. Instead, it is often more appealing to develop reduced-form models for each component model that are small enough to be comprehensible, flexible, and easily linked to one another. A reduced-form model provides approximately the same behavior as the model from which it is extracted in a much more compact and comprehensible form. The goal of this project is to develop and evaluate techniques to help modellers analyze their models and construct reduce-form models. The specific objectives are: 1. To design and prototype a modellers' workbench to facilitate specification and documentation of the inputs, outputs, and assumptions of existing models of various types and sizes; 2. To provide a range of existing and new sampling techniques to generate scenarios to exercise these models; 3. To provide a range of existing and new model analysis techniques to identify key inputs and parameters, identify nonmonotonic and regions of qualitatively different behavior, and so help develop reduced-form models; and 4. To evaluate these techniques experimentally by application to the development of reduced-form models and integrated assessment models, and so to provide guidance on which techniques are most valuable and for what kinds of models. This project is supported under the Methods and Models for Integrated Assessment funding opportunity.