Bridging Equations and Agents
Bridging Equations and Agents
Embedded Files
An introduction to the world of modelling and simulation with a live example.
What is Simulation?
What is Simulation?
A simulation is a representation that imitates real-world systems in order to study or predict the phenomena and interactions that happens in these systems. Researchers use these simulations in a wide range of fields from urban traffic, weather systems, pedestrian walking routes, and in the case of our blog, the spread of the COVID-19 virus.
In our blog, we will be presenting an agent-based model representing the SEIR model from the referenced research paper. The model used in the paper is actually an equation-based model which use mathematical equations to simulate the spread of the COVID-19 virus based on the real-world data gathered during the pandemic on the NCR, Davao City, and Baguio City. However, we decided to convert the model into an agent-based model to better visualize how simulations work.
How It Works
How It Works
Firstly, let's talk about equation-based models, since that is what the paper originally uses.
A lot of simulations in science—especially in physics—are built using equations. That’s mainly because these fields already have solid theories that can be turned into math, often using differential equations to describe how things change over time.
There are usually two ways this is done. One is particle-based, where you’re tracking individual objects and how they interact with each other. The other is field-based, which looks at how something continuous—like a fluid or some kind of field—evolves over time instead of focusing on separate pieces.
You’ll find these kinds of simulations everywhere, from trying to understand how galaxies form to modeling things like severe storms.
In the case of COVID-19 spreading or any other virus, an equation-based model that will be used to simulate them will be particle-based. Essentially treating each simulated human as a particle that will interact with each other. For a deep dive to the actual model used by the paper, you may go to this link.
Next, let's talk about agent-based modelling.
Agent-based simulations tend to show up more in the social and behavioral sciences, but they’re also used in areas like artificial life, epidemiology, and ecology—basically anywhere you’re trying to study how lots of individuals interact within a system.
They’re kind of similar to particle-based simulations since you’re still dealing with many separate “entities” or individuals. The big difference is that there isn’t one set of overarching equations controlling everything. Instead, each individual follows its own set of rules, and the overall behavior of the system just emerges from all those interactions.
Agent-Based Modeling (ABM) is basically a bottom-up way of building simulations. Instead of starting with big, overall equations, you start small—by modeling individual “agents” that act on their own inside a simulated environment, each following a set of simple rules.
What makes ABM stand out is how detailed it can get. Traditional models that rely on equations or probabilities often smooth things out and miss the finer details of complex systems. With ABM, you can actually represent individual behaviors and how agents continuously interact with each other—whether they’re people, animals, or even bacteria.
If you want to explore of the agent-based model that we have created based on the equations of the SEIR model, you may run the live simulation below. If you also want to know the explanations behind each test setting, you may also find them below.
Explore the Simulation
Explore the Simulation
Live Simulation: Run the NetLogo model directly in your browser
Test Settings: Understand the specific parameters, constants, and variables we used to calibrate the model for the Philippines.
Page updated
Google Sites
Report abuse