>An agent-based model (ABM) is one of a class of computational models for simulating the actions and interactions of autonomous agents (both individual or collective entities such as organizations or groups) with a view to assessing their effects on the system as a whole. It combines elements of game theory, complex systems, emergence, computational sociology, multi-agent systems, and evolutionary programming. Monte Carlo methods are used to introduce randomness. Particularly within ecology, ABMs are also called individual-based models (IBMs),[1] and individuals within IBMs may be simpler than fully autonomous agents within ABMs.
Right now I'm making a model that uses the lanchester equations to predict -at least I'm theory- the casualties sustained in a two way infantry engagement
Julian Morales
This just looks like liquid phase separation
Ryan Moore
It's the schelling model - it demonstrates how widespread patterns of group segregation can emerge despite a largely tolerant population
Jose Ward
Bump for life
Christopher Collins
How do i into?
Aaron Johnson
Any book, please I beg you?
Adam Gray
Growing Artificial Societies is one of the OG books but it's hella expensive cause I think it's out of print. The textbook for my class is called Introduction to Agent Based and Individual Based Modeling by Railsback and Grimm. Note that Railsback and Grimm is geared towards NetLogo specifically, but if you're looking for a way to introduce yourself to ABM, NetLogo is what you should be using anyway
Ian Myers
What I think is neat about ABMs is that some companies will pay top dollar for really detailed supply chain simulations
I feel like if people like Pareto and Schelling had access to modern supercomputers we would understand so much more about the world
Oliver Stewart
Bonus link for thread enrichment:
George Mason university has a whole bunch of smart people working on applying computation to social science and promoting the viewing of social systems as complex adaptive systems.
A buddy of mine told me that the rumor is that there's a professor there who recently ran an ABM with over 3 million agents in it at once.
what the hell is a complex adaptive system. I hate the term complex systems because it's overused and doesn't mean much.
are you bad mouthing netlogo? Because netlogo is fucking awesome. Although being able to define your own functions is something most languages have.
Leo Barnes
Used netlogo for a semester under a PI. First half was nice. Second half was implementing increasingly obscure work-arrounds. Next semester it was abandoned and a real language was used.
Is nice for getting started quickly I suppose.
Jonathan Davis
Oh, i meant CAN'T. You can't make your own function. What is up with that trash.
Or something else really fundamental. Haven't touched netlogo in two years.
Easton Robinson
>what the hell is a complex adaptive system
The term is overused because it is still in the early stages of being defined. Roughly speaking, it is a system that exhibits emergent nonlinear properties, is composed of many individual agents operating at various strata, and involves many simultaneous markov chains
James Ortiz
What was the second language you used?
Dylan Harris
Agent based modelling is a huge red pill, as the agents that only tolerate their own kind grow to dominate the whole space. Mean while the altruistic agents get annihilated early on
Benjamin Hughes
what other kind of models are there. Why Should I choose ABM over others?
Brandon Jackson
ABM and maths/stats in general reveal some pretty grim patterns in society. Like Pareto's discovery of wealth distribution and that one equation that led the guy who discovered it to kill himself (i forget his name)
Jacob Ward
ABM is usually contrasted with equation based modeling and system dynamics. Where EQB and system dynamics usually deal with the relationships between the stocks and flows of aggregate variables like GDP, GINI coefficient, or HDI, ABMs are designed at the micro-level.
When someone designs an agent based model, they program only the rules of behavior for however many individuals they choose to put into the model. Once the rules for the agents are laid out, each agent follows its programmed behavior on a spacial plane. Usually, the result is something that one would not expect given the prescribed rules at the individual level, a property called emergence.
What is interesting about ABMs is that they need not be exclusive to social systems. Anything you can think of could be counted as an individual in an ABM so long as you can describe its behavior. These individuals could be cells in a culture, individual muscle fibers, or molecules in a solution.
In short, it allows a researcher to create a space where components of a system act according to safe assumptions or known rules about their behavior, and see how it affects large-scale measurements
Jacob Fisher
Growing Artificial Societies is on libgen for free
Nathan Hernandez
Sounds a lot like cellular automata or kinetic monte carlo.
One advantage of continuum models in the physical sciences is that they aren't typically dependent on grid geometry.
Cameron Perry
>Sounds a lot like cellular automata
exactly. CA was a huge influence in the development of ABMs as we know them today
Well if you shape your rewards properly, you can get multiple agents to work together to accomplish something really cool. youtube.com/watch?v=cq8lFO5Mzi0
>>it is still in the early stages of being defined So it's basically bullshit because we can define it to whatever we want it to mean >>emergent oh boy another one of those overused terms that means practically nothing en.wikipedia.org/wiki/Emergence
>>is composed of many individual agents operating at various strata, and involves many simultaneous markov chains Well I've never heard of the word agent being used in this definition before, nor the markov chains. So I take it that your research involves those two things.
Nolan Rogers
bump
Liam Nguyen
do you think this is important in finance? does it have a major role in creating mathematical models, forecasting? This seems like a really interesting subject to learn and I'd like it even more if it had potential for interesting projects or for increasing employability or for using it for financial investment and stuff like that.
Sebastian Baker
rump pump bump
Jeremiah King
>creating mathematical models Technically these are already mathematical models, just ones that can't be described by a small number of neat equations (equalities). >forecasting If you mean forecasting in the statistical sense, then you can do Monte Carlo (as pointed out by the user here ) but you shouldn't be expect things like the law of large numbers, or population sampling, to carry over unchanged. But then again, I don't work with ABMs for a living, so maybe the people who do will know something that I don't.
Tyler Reyes
>Agent-based modeling Sounds like an assumption laden """mathematical modelling""" people use for getting bullshit published, also something politicians might love to fund due to their ignorance.
Austin Walker
It only uses as many assumptions as you want boo boo
Luke Ramirez
can bayesian into abm? i've been doing lots of bayesian shit recently, abms look cool, can into?
Jack Williams
You can give each agent a prior and update it iteratively, so in that sense Bayesianism is quite compatible with ABMs.
Getting closed-form expressions for the individual posteriors is probably not possible, though you might just be able to calculate the posterior of a "representative agent" with a couple of ergodicity assumptions.
Landon Jenkins
Finance can go suck a dick for all I care. I just want to make communist robots to take over the world.