dipole said:
Hi, I'm taking an ecology course and I need to model some kind of population/system for a final project. The problem is, I'm a physics major and know almost nothing about real biological systems, so I'm looking for suggestions of real systems that are simple enough to model fairly realistically in a school project.
Well-studied systems with a good amount of literature behind them would be ideal, but anything I can find good data on will work. There's no real guidelines, it could be aquatic, terrestrial, whatever! It doesn't even have to be strictly biological. I'm open to suggestions, so if you have any ideas please share!
The hard part of modeling is finding an appropriate data base. The models are generally simple, but finding data appropriate to a simple model is difficult. There are tricks for finding data to analyze.
One thing that I find useful when searching for data is the “images” option in the google search. If you key in a quantitative subject, and then do an “image search”, what often comes up is a selection of graphs. Each graph has data. The data can be modeled any way you like. You don’t have to model it the way the author modeled it. In fact, it could be more fun to model it in a way that has nothing to do with the theory that the original investigator intended.
Although the predator-prey model is over-used, it is very general. I don’t know what your programming skills are like. However, maybe you would prefer to vary an overworked problem rather than to construct a model from scratch.
One thing you could try is an inverse problem. Start with a mathematical model that is fairly well known. Find a data set where the dynamics are unknown, but where the data vaguely resembles solutions to the well-known model. Then, try to find input parameters that approximate the data. Different ways of “trying out” parameters could be part of your report. The data you are modeling does not have to have a clear physical connection to the physical model.
Predator-prey model is a classic. You know that strange oscillations are typical of the solutions. Find a data set with oscillatory data for any system at all. Try to fit the data set to by varying input parameters of the predator prey-problem. The data set does not have to have a known connection to predators or prey.
This is a slightly different approach than what you are asking for. I am suggesting that you start with a predator-prey model, and search for a graph that can be “fitted” to it.
I did a search using the image option on Google of “turbidity versus time”. I got a large selection of graphs. One graph that made me think of population studies was in a study of “aqueous mixtures of pectin versus chitosan.” The abstract is in this link.
http://www.sciencedirect.com/science/article/pii/S0014305705001072
You can write a predator-prey program, but keep the source code. You can vary any line of the program that you like.
You can also generate solutions to the predator-prey problem. Make lots of graphs. Then, look through the literature for an experimental graph that happens to look like one of the graphs from your predator-prey program. Then read the article where the data came from, and then make up a physical explanation of why the two match.
This is not cooking. Try to find new insight into the data this other investigator found. His model may be correct. However, any investigator can miss something obvious.
One scientist that I know was working on a program using molecular dynamics to model chemical reactions. The reactions generally resulted in stable molecules. He made a copy of the program and erased an arbitrary line. The simulated molecules exploded. So he started a successful project simulating explosives.
Maybe you could simulate explosives with a predator-prey model.
