How does the model get from one time step to another. What are the rules to follow in each time step?
In CA’s and L-systems, the dynamic is completely described in the rules. In a systems model, it is described by equations. In agent-based models, it is represented by interactions between agents.
Ted’s Project
Phenomenon: Endosymbiosis. Mitochondria generate energy for our cells. How did they arise? We now belive that they used to be separate organisms that were eaten by other cells and not digested.
The question: Why did the endosymbiant survive? Why didn’t it reproduce until the host burst, like viruses, and why didn’t it die out. So, the phenomenon is the synchronization of reproduction rates. Co-evoution of population growth rates. (Not sure that this *actually* happened, but that doesn’t matter at this stage).
Possibilities: STELLA: populations as actors. Chose an agent-based model. Will track the numbers of agents in each populations.
Interactions: What is the mechanism of growth and synchronization? (This is the science part).
1. Interacting pairs: host-host endosymbiant-endosymbiant host-endosymbiant self-self (dying, etc.). 2. Kinds of interactions: eating, reproduction, effect on population growth rate.
Think back to the logistic growth models we discussed. The actors were population size, growth rate, and carrying capacity, and they were related by an equation.
What would an agent-based model of logisic growth look like? Each individual has a probability of reproducing. There could be a resource (--> floating, or fixed (spatial heterogeneity)). There are mating interactions and eating interactions. The resource might also have a replenishment interaction. The probability of mating could be dependent on the amount of resource available. This can’t really be described by equations, you have to write time step interactions possibilities.
**Irina & Ananya’s project: Stock Market Crashes, a systems model. Two variables are stock price fluctuation (amplitude), and optimism. Higher rates of fluctuation decrease optimism. Do the model by industry stocks. The fluctuation of each of these will reflect a typical stock in a given industry. Note: there is no formula for optimism. The equation will be determined by the opinions of the group doing the modeling.
Ted - Use of a scale is a way to quantify things without getting precise data. If the interactions between two actors are very complex, maybe you’re choosing the wrong ineractions. Maybe you need to explain this complicated behavior in terms of simpler behaviors. Maybe you just need to make a model in you head to explain the shape of the curve. You need to be able to justify it on your poster.
Modeling is about taking what we do know and seeing what comes out of it. You could say “Prof. X says this about optimism, and Prof. Y says this. It turns out consistent with what Prof. Y says, so our model supports his hypothesis about optimism and the stock market.”
Maybe you’re trying to see what structure of the optimism vs. fluctuation graphs makes the stock market behave the way it does.
Rob - Instead, you could look at probabilities of selling, buying an keeping stocks, depending on these variables.
**Christina and Hannah’s project: Agent-based model of colony founding by multiple queens. Question: are we trying to have too much mechanistic realism?
Ted -Imulse is to take all the information and put it into the model. Instead, it should be to oversimplify and then add.
We know ants can communicate with other ants in a particular way. Does this alone explain ... divison of labor according to caste in an anto colony.
**Arun & Carolyn’s project: Difference in AIDS transmission rates in San Fransisco and New York. Different groups of people, such as homosexual men, intravenous drug users. Using a systems model in STELLA. Focusing on male populations. Ex. drug users won’t get infected if they’re rich. Economics and education are major factors. An example of a stock is “poor drug users who don’t have HIV”.
