The models used in this program have been taken from: "Opinion Dynamics and Bounded Confidence Models, Analysis, and Simulation" by Rainer Hegselmann and Ulrich Krause (2002).
In this program we aim to simulate different models of opinion formation within an interacting population. There is an extensive collection of research exploring such models so we have chosen a few to show how polarization, the formation of stable subgroups, can occur in a population.
We consider each individual, or agent, in the population to have some opinion between 0 and 1 which they update based on the opinions of all other agents in the population. To normalize this updating each agent has a weight associated with all other agents such that their sum is 1. With this model as a baseline, we add complexity to create three more models:
- Social susceptibility model
- Bounded confidence model
- Bounded confidence and proximity model
Model (1) includes an additional weight or susceptibility between 0 and 1 which represents how influenced they are by all other agents' opinions. Model (2) includes a bounded confidence such that an agent will ignore another's opinion if the difference of their opinons is greater than the bounded confidence. Model (3) includes bounded confidence with the additional proximity limit. This aims to simulate a senario where agents only update their opinion based off of agents who neighbor them (are within the proximity limit).
Running this program will allow the user to choose the model used, initial opinion distribution, and the parameters associated with each model. It will return a plot showing the evolution of all agents opinions.
The cli program starts off with a tutorial, which will prompt the user to choose the amount of agents and the number of time stamps (they are the same number.) Next, the user will be prompt to choose the susceptibility and the value of proximity. Each of these values have an effect on whether or not there will be any converges or polarization in the opinion as shown in the examples below.
This is an example of a normally distributed opinion where the number of population is 200.
The same as above with a size of 100.
