Experiments In Multiresolution Modeling (MRM)
This study describes the motivation for multiresolution modeling (MRM) within a single model or a family of models. After introducing a new measure of consistency for models of different resolution, the study discusses in some depth obstacles to and methods for multiresolution modeling (also called variable-resolution modeling), illustrating issues with a detailed military example involving precision fires. The study highlights the value of visual design, array formalism, formal mathematics to identify natural aggregation fragments, integrated hierarchical variable resolution (IHVR) yielding "trees" of variables, estimation theory, alternative aggregate representations called out in a user interface, "stretcher variables," and computational methods to identify natural phase transitions and facilitate calibrations. The study recommends that major Department of Defense models such as JWARS emphasize MRM and related research on family-of-models issues. It notes that MRM is not an all-or-nothing matter and that incorporating even some multiresolution features can be quite useful. Finally, the study notes that MRM is a frontier issue and suggests directions for further research.
Table of Contents
Definitions and Technical Background
Why Is MRM Important?
Obstacles to MRM
A Research Approach: Experimenting with MRM Design
An Illustrative Problem: Halting an Invading Army with Precision Fires
Conclusions and Recommendations
Model of the Halt Problem