so we've gone through a fuse simple models that kind of get you a flavour offor some of the things that agent based modeling has been useing to explore in the past in many of those contexts agent based modeling was used almost as a thought experiment to explore potential theories about the way the world works or to understand how different parameters relate through things along those lines but agent based modeling could also be used to create very complex models umm, so one natural question might be how big in advance can it get well in netlogo alone which we'll talk about here you could have tens of thousands of agents of patches in fact I've seen now models with hundreds of thousands and potentially even millions of agents and patches you can have very complex decision makers decision makers within the agent who are optimizing some sort of economic market or who are trying to make a decision based upon all the past behavior they've had you could see many different agent types so you could have five or six or dozens of agent types I puff in the model you can have models of whole cities and how they're integrated and one of the nice things about not only netlogo but many of the agent based modeling platforms out there is that they allow you to bring in additional tools that are available outside of those software packages so you can integrate for instance that statistical analysis package in order to use that to its fullest extent within the system now what constrains how big a model can get well that's an interesting question a lot of times has to do with the computational power you have the amount of time you have the complexity of the agents right the more complex the agents are the more computational time they're going to take right and you know also the environment and so for in this particular context we're going to now talk about two particular types of environments that are often used within Agent based modeling one is more of a spatial type model and the other one is more of a network type model so one of the models that I really like there one of the groups that I like that does a lot of age' based modeling is a group called the redfish group that builds a lot of models and net logo actually uses them quite successfully to understand complex problems there around them and you know into this particular model that I'm showing you is a visualization of a very complex traffic intersection so it kind of feeds naturally from the simple traffic model that we were looking at before right and now if you're still you know you've got kind of these representation of agents and things like that but allows them to kind of look at and try and understand would a traffic network such like this actually work so this is another one of my favorite models this one was also built by the redfish group under the principles of Stephen Goren no in Densmore and this particular case what they were trying to do was to help the Santa Fe safety departments with understanding a evacuation or a mass egress actually from an annual event that happened to Santa Fe every year called the Zozobra which is from what I understand a little like burning it I've never been myself but apparently at the end of it they build this big thing and I said I'm fire and you know it's something where you obviously want to control a little bit the environment around you and what I find interesting about this particular model is that you know it led to a lot of interesting conversations about how to set up exits correctly from the event but if you notice there's one particular little agent that runs off towards the north end of the screen and this kind of helps to or the top end of the screen this kind of helps illustrate one of the points about agent-based modeling is that when you construct an Agent based model and you give the agents goals if you leave any kind of gaps any kind of errors and the goals that you haven't thought through they will find it right and in this particular case this agent seems to have found some sort of break in the fence or something along those lines that allows it to exit a completely different manner from the rest, anyway it's just another great example of using agent-based modeling in a spatial setting to explore some more complex phenomenon a different project that I was involved in actually was that the tail end of the project was that it was a procedural modeling of city project right in this particular context net logo was being used to try and see if it was possible to paint cities in in other words to create cities that could have the flavor of something like Rome because you had painted a set of seeds down that would create that but not actually be Rome I just feel like Rome right or cities that might have the flavor of something like Paris but not actually be Paris and things along those lines right and in fact one of the partners on this was Maxis who actually builds the SimCity toolkit right and so we were able to take the models that were rendered in that level which is what this picture is and then put them into a GIS piece of software and then also build them up into the SimCity rendering environment and whenever possible I always try and include a reference to a paper that kind of talks about that process so there's the paper I also happen to be on the committee of a student who is doing some very interesting policy analysis of agent based modeling and again this is in the spatial context so she was actually bringing in data from a variety of different geographical sources including the real some rail networks and some description of demographics and things like that for the greater Chicagoland area and with the geographic data she built also a model of transportation decision-making for each individual agent within the model which was at the household level right and so in that model they when the agents would actually make a decision based upon various questions like how expensive it was to buy gas or whether or not how expensive was to buy a car whether they would use a private car or public transportation to get to their work destination right and what was interesting about this is that you know a lot of the data was actually held outside the ABM in the GIS landscape and then there was also the integration of actually a traffic demand model that was integrated as well outside the ABM and that the ABM only controlled kind of the integration of all these things together the policy analysis as well as the decision-making of the individual households and what she was trying to explore was to explore when you implemented different policies was it possible to revert to a more public transit world from a world which was descending into a where everyone has their own private transit system