Cybernetic Simulation Applet - Instructions
Abstract: This page includes user's instructions for an applet program made in HUT Control Laboratory. The aim of the program is to simulate and illustrate a cybernetic ecosystem that is based on local adaptation. The program implements the "go for the resources"-behavior. The emergent behavior arising from this strategy will coincide with sparse- / principal component analysis (with suitable configuration). More information of the theory is found from the neocybernetics page.
The program can be easily used without advanced knowledge in neocybernetics. Below you will find basic instructions to use the program. For more ambiguous usage, advanced settings can be configured; these are explained at the end of this page.
Contents:
The ecosystem considered consists of plants and herbivores which live in a closed environment. The environment is considered as a rectangular area (cf. a half hectare forest) which has a variety of basic resources (or just resources) which are spatially distributed along the area.
Environment, "The half hectare forest".
The environment has been spatially discretisized so that the resources are constant in one grid rectangle. The amount of different types of resources is predetermined (default 8). These could be eg. amount of sunlight, humidity, fertilizers and so on.
As the environment is spatially discretisized, the whole environment can now be defined as a 3 dimensional matrix in which the resources are one dimension. In the program this is visualized with a rectangular map in which the weighting of different colors illustrates the resources (as shown below).
Environment in the actual program
The resources are varying in time so that the variation of different resources is independent, where as there can exist connections between them (cf. high sunlight results in low humidity). The amount of one resource is varying equally along the map. Compare to the amount of sunlight; it varies from place to place, and from year to year but the temporal variation affects equally along the whole forest.
The resources have nominal values. That is, their average values in time. These nominal values vary spatially. Compare e.g. that the humidity close to a river is in average higher. In the program these nominal values are randomly generated so that they form clusters. In one cluster the relative distribution of nominal resources are almost equal.
The plants (and herbivores) are characterized by their profile. The profile is a vector of n elements (n resources) that tell how much the plant benefits of each resource. Compare to a specific plant that requires little sunlight and much humidity, the profile would have a small weight on sunlight and a large on humidity (e.g. for [sunlight humidity] profile [0.1 2.0]).
The ecosystem simulation consists of two phases; the local adaptation and the profile adaptation or the evolution.
The local adaptation timescale is considered relatively fast. In this adaptation the population sizes and leftover resources are determined independently for each point. Notice that the profile is considered constant. Compare to a scenario where e.g. two known plants are competing for the same resources, the plants will reproduce for as long as the resources last, and the plant with characteristics better fit for the resources will reproduce more. Iteration is continued until equilibrium is found. The plant characteristics will not change in local adaptation.
1.5 Profile adaptation (evolution)
The profile adaptation, that is, the evolution is the adaptation of the plant core characteristics. The local adaptation is considered to have reached eqilibrium before the profiles are adapted. Practically, most plants and herbivores reach local adaptation equilibrium in 1 year, thus the profile adaptation would happen once a year. The time for the profile to reach equilibrium is hundreds, thousands of years.
The visual outlook of the plants is a function of the profile, even though the physical characteristics do not reflect the resource usage.
The program runs on java runtime environment 1.4 or newer, with Internet Explorer or Mozilla browsers. If you encounter problems with the program please install the newest java runtime environment from http://java.sun.com/javase/downloads/index.jsp (Java Runtime Environment (JRE) 6 ).
To start the program, click on the link start simulation. The program will start on a new window. A button "Start Simulation" is displayed in the middle of the window, press this button to start the simulation.
First, try Quick start button. It will provide a new resource profile with all settings chosen by computer. If you want to change these setting, try Build New World button.
The environment now created consists of n (default 8) different resources that are spatially distributed in an area (20 x 20).
There are two different views available for viewing the resources. The default view is the profile view (below on right). In this view the resources are shown as collected average over the overall map (20 x 20). The map view (below on left) shows different resources with different colors spatially placed on the map. The view can be changed by pressing Toggle view button while the plants are evolving.
The resources are varying around their nominal values with a chosen variance (default = 1). The different types of resources are varying independently, but there is some correlation between them.
The basic resources could be seen to represent sunlight, humidity, fertilizers and so on.
When the resources have been initialized a button START will appear green on the plants section. Press this START button to start plant evolution.
When the START button has been pressed, the evolution will go on until STOP is pressed. The evolution time (iterations) is shown on the left side of the window. Each iteration counts for a "year" and only the final balance is shown.
At the beginning of every iteration new current resources are generated on basis of initialized nominal resources and predetermined variance. The current resources are shown on the resource section along with the free resources. The free resources represent the resources that remain after feedback.
While the evolution is going on you can at any time reset the plant profiles, initialize new resources or toggle view. Resetting of profiles will also reset the evolution time.
The bar right to the plant profile (or map) represents the current size of the population. The average size is also shown.
To stop the evolution press STOP. The evolution will stop and you can either continue the plant evolution, or start the herbivore evolution.
To start the herbivore evolution, press the green START button in the herbivore section.
As with the plants, the herbivore evolution will go on until the red STOP button is pressed. While the evolution is going on you can reset profiles, toggle view, toggle visualization or initialize new resources. Resetting the profiles will also reset the evolution timer.
The bar next to the profile represents the size of the herbivore population.
Notice that the herbivores are now using the plants as their resources. The plants are though not directly varied with time, but through the basic resources.
While evolution is going on, you can also click on any of the herbivore profiles (or maps) and a new window with additional information will appear. In this window, you can choose to make the herbivore profile constant by pressing Set constant profile button. You can adjust the profile by moving the yellow markers. The profile will be automatically scaled so that its norm is equal to 1. When you are ready, click ready.
When you have set a profile for a herbivore it will appear among the other herbivores. The profile of the constant herbivore will be scaled as the evolution goes on. The scaling will be positive if there are free resources in the direction of the profile, and negative if there is lack of resources in the direction of the profile. The appearance of the herbivore does also change as the profile is scaled.
When you click on the constant herbivore again, a window will appear that allows you to change the profile. You can also make the herbivore cybernetic by pressing make cybernetic. If the evolution is going on as you press make cybernetic it will stop when you press it. When you start the herbivore evolution again the changed herbivore will act as cybernetic, starting to evolve from the profile that it had before it became cybernetic.
To stop the herbivore evolution, press the red STOP button.
When the evolution is stopped, you can continue the evolution by pressing the START button again, continue the plant evolution or add a new herbivore system by pressing Add Herbivore System. This will add a new, parallel herbivore system that will evolve simultaneously, independently of the previous herbivore system. By clicking on the different systems you can get the system herbivores visible.
The program allows you to adjust many simulating parameters. Notice that changing these parameters might change the emerging behavior of the populations.
To change advanced parameters click "Advanced settings" and "Set parameters" on the top menu bar. Any on-going simulation will be terminated.
The following parameters affect the simulation:
Cut-off negative resources (Ubar) |
By selecting this Ubar (current free resources) negative values are set to zero. |
Cut-off negative populations(Xbar) | By selecting this Xbar (population sizes) negative values are set to zero. |
Set constant resource variance | By selecting this the variance added to different resources is of equal magnitude, otherwise it is decreasing in power-scale from first to last resource. |
Set plant-resources connection factor q1 |
Local adaptation connection factor for plants, see eq.(1) below |
Set plant-resources connection factor q2 | Local adaptation connection factor for plants, see eq.(1) below |
Set herbivore-plants connection factor q1 | Local adaptation connection factor for herbivores, see eq.(1) below |
Set herbivore-plants connection factor q2 | Local adaptation connection factor for herbivores, see eq.(1) below |
Set plant evolution factor lambda | Plant evolution converging factor lambda, see eq.(2) below |
Set herbivore evolution factor lambda | Herbivore evolution converging factor lambda, see eq.(2) below |
Set variation in plant initial profile | Set the initial variation in plant profiles, also applied when reset is pressed |
Set single-point adaptation factor | Local adaptation converging factor lambda (1&2), see eq. (1) below |
Set single point accuracy | Threshold value for ending local adaptation algorithm, |
(1) | ||
population size , free (visible) resources, original (hidden) resources | ||
(2) |
Notice that the above equations include the populations collected in vectors. The algorithm can also be written in scalar form and the calculations can be performed independently for each population (in one iterating loop).
2.6 Contact
If you encounter any bugs or have questions about the program implementation, please contact jan.tiri@tkk.fi.
Ideas on how to develop the program are also welcomed.
Last updated in September 5, 2008 -Jan Tiri