PlantMoosePredator System Simulation Anastasia Mironova CS 470 Project
- Slides: 11
Plant-Moose-Predator System Simulation Anastasia Mironova CS 470
Project Overview • Goal: – Develop an agent-based simulation to model plantmoose-predator system behavior – Client is Dr. Donald Spalinger, Department of Biological Sciences, Professor and Chair • Why: – To assist scientists in codifying current hypotheses and axioms regarding the growth and defense responses of plants to herbivore foraging patterns and foraging efficiency, herbivore productivity and habitat carrying capacity, and moose-predator interactions – Limited capabilities of existing systems
Desirable vs. Possible Priority list for the desirable version: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Flexible Design 2 D Landscape implementation Agent “Plant” implementation Agent “Moose” implementation Plant-Moose interaction implementation Expansion of scope (space and number of agents) Optimization (memory, run-time) Agent “Predator” implementation Moose-Predator interaction implementation Complex Landscape Patterns Artificial Intelligence Behavior Patterns for Agents Main decision: little of everything vs. thorough specifics
Implementation Decisions Effort was geared toward more thorough implementation of specific components according to the initial priority list: – Flexible, reliable system architecture, capable of fully supporting all the functional specifications known at the time and allowing for future expansions – Emphasized Simulation Components: • Landscape model • Plants
Simulation – PMPSim. java • Timer • Landscape: Landscape. Segment[][] • Plants – Plants. Hash. Table: Plant, hashkey = plant. ID – Plants. Array: plant. ID • Moose – Moose. Hash. Table: Moose, hashkey = moose. ID – Moose. Array: moose. ID • Predators – Predators. Hash. Table: Predator, hashkey = predator. ID – Predators. Array: predator. ID • Other parameters – Growth. Season, Precipitation, etc.
Simplified System Architecture Simulation. Parameters. Main. Frame Simple. Landscape. Specifications. Panel Simulation. Main. Frame Advanced. Landscape. Specifications. Frame Simulation. Menu. Bar Simulation. Tool. Bar Landscape. Tool. Bar Individual. Segment. Display Simple. Plants. Specifications. Panel Preset. Plant. Properties. Frame Simple. Moose. Specifications. Panel Custom. Plant. Properties. Frame Simple. Predators. Specifications. Panel
GUI Representation of the System
Landscape Model • Main design decision: – Partitioning • Characteristics: – – – Dimensions Elevation Contours Bogs/Wetlands Forest Habitat Riparian Habitat Plant species-specific patches – Custom View
Plants • Preset vs. Custom • Groups vs. Individual • Characteristics: – – – Plant ID Species Name Coordinates Elevation Range Aspect Preference Special Habitat Associations – Habitat Restrictions
Potential Extensions Existing system design allows for: – Intuitive implementation of any additional types of agents, such as • Moose • Predators – Potentially, supporting very high levels of complexity, necessary for incorporating extensive amounts of research data and, hence, giving the simulation the capabilities to represent a more realistic model – Future optimization in both memory usage and run-time via • Landscape partitioning • Threading
Conclusion • Developed a prototype of fundamental components for customizing and running plant-moose-predators simulations • Lessons Learned: – More about working with a real client – More about design – More about simulations