1 / 29

Monte Carlo-Markov Chain

Monte Carlo-Markov Chain. Markov Chain: A series of states of being that have probabilities associated with transitions (i.e. the state is not deterministic but has some stochastic component). Caribou – Wolf Interactions. http://www.youtube.com/watch?v=nK1JOmMQ5Fc. Issues with Simulations.

medge-mckee
Download Presentation

Monte Carlo-Markov Chain

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Monte Carlo-Markov Chain • Markov Chain: • A series of states of being that have probabilities associated with transitions (i.e. the state is not deterministic but has some stochastic component).

  2. Caribou – Wolf Interactions • http://www.youtube.com/watch?v=nK1JOmMQ5Fc

  3. Issues with Simulations • Realistic Movement • Moving toward desired areas • Keep animals from getting “stuck” • Validation • Boundary conditions • What happens at the edges? • Disappear/die (need immigration as well) • Reverse direction • Wrap to the other side (not realistic) • Model stability • Model complexity/Performance

  4. Moving Toward Desired Areas • Use Distance Raster • Desired area: • Spawning ground, feeding grounds • Destination is desired area • Pixels with lower cost are closer to desired area • Animals move to adjacent pixels with lower cost

  5. Moving Through Networks • Use polylines with network nodes • Move in direction (or against) of polylines • Need to make decisions at nodes

  6. Getting “Stuck” • Certain conditions arise and animals can become stuck and just move back and forth • Add additional randomness • Examine environmental layers for unrealistic values • Streams not flowing perfectly down hill • Bays not connected with the ocean

  7. Validation • Run models over and over again • Record locations, births, deaths, feeding • Create probability surfaces • Validate against existing datasets • Are observations/measurements within the predicted areas? • Could use likelihood/AIC…

  8. Model Stability • Density dependence • Most species struggle more when crowded • Reduces food availability, increases disease • Places a control the number of individuals • Balancing birth rates and death rates • Realistic lifecycles, predation rates

  9. Model Complexity • Typically there are lots of agents • Need to keep the behaviors and attributes simple • The group behaviors are typically more complex than expected from the individuals • Model a population • Convert “groups” of individuals to populations when they cluster

  10. Performance • Select the right resolution for rasters • High enough to be realistic • Low enough for speed • May have to just simulate a smaller area than desired.

  11. 3D and Temporal • Simulations are almost always temporally based • 3D is common but requires more hardware/time • Also requires special software and 3D data

  12. Ecological Modeling • Water, carbon, nutrient cycling • Trophic models • Population models • Predator/Prey • Disease wsu.edu

  13. Sustainable Fisheries www.niwa.co.nz

  14. Can operating rooms in Second Life teach real doctors? - Discover Sims World of Warcraft

  15. Tools • NetLogo • HexSim • MASON Multi-Agent Simulation Toolkit • Repast • Programming! • Python • Java • Books: “Agent-Based Models of Geographical Systems”

  16. Possible Simple Models • Fire w/fire fighters • Zombies and humans • Reef fish • Wolves and elk • Invasive species (w/managers)

  17. Envision – Policy Simulation

  18. Simulations • Parameterization • Based on mechanisms/theories • Typically include random effects • “Tweeked” to fit reality • Validation • Run over observed space and time • Do simulated measures match observed? • Run it over and over • Observed fit into “Confidence interval”?

  19. Simulations • Startup: • Either simulate a known situation • Or, run until reaches an expected state • Stability • Populations tend to breed out of control or die out over time • Build in realistic limitations

  20. Python SEIBM • Very simple model • Includes 3 classes: • Animal, Predator, Prey

  21. SEIBM – Main Script • Imports: Tkinter, time, Predator, Prey, Animal • Setup the GUI • Initialize animal objects in an array • Loop forever: • Update each object • Redraw the window • Let Python process events (mouse clicks) • Sleep for a bit

  22. Animals • Initializatoin: • Create our oval on the canvas • Update: • Find a random direction to move • If moving off the canvas, reverse direction • Do the move

  23. Prey • Initialization: • Call the Animal class to create a green oval • Set the time until birth • Update: • Call the Animal to update movement • Decrement the time until birth • If it’s time for birth: • Create a new object at the current coordinate • Add the object to the array

  24. Predator • Initialization • Call the Animal class to create a red oval • Set the time to birth • Update: • All the Animal to do movement • Decrement time to birth • If it’s time to birth: • Create a new Predator • Add it to the array and reset the time to birth • If any prey are close by • Delete them

  25. HexSim • Visit Hexsim.net to download • Decompress the folder • Can run “Hexsim.exe” from within the folder without other installation • See the “Examples” on the HexSim website to get started • The User’s Guide has good information on how HexSim works

  26. HexSim Basics • Set the workspace: • HexSim -> Set Workspace • Select a “*.grid” file • Double click on a scenario to “open” • Adds a tab for that scenario • “Scenario” menu items work on the currently selected tab • View spatial data: • In the “Spatial Data” panel, open items • Double-click on “-> 1” to show the data

  27. HexSim Basics • Select “Scenario -> Run Simulation” • Save an XML file for the run • Click “Start” in the window that is displayed • These examples are not yet parameterized so you may need to modify them to do something interesting

  28. Comparison of Software • http://en.wikipedia.org/wiki/Comparison_of_agent-based_modeling_software

  29. Simulation Modeling • Climate: • http://climate.nasa.gov/meteorologists • Tsunamis: • http://www.youtube.com/watch?v=_bCTa5su8II • http://www.youtube.com/watch?v=WgpXzwLuGDo • Plate tectonics: • http://www.youtube.com/watch?v=ryrXAGY1dmE

More Related