Dear Samuel, I think that the NetLogo is a very good modelling tool for the biology. Maybe, it looks like a toy :-), but it is possible to create useful MAS models. Try to look on the wepage http://ccl.northwestern.edu/netlogo/models/ containing models created in the NetLogo. The next alternative can be Repast. It is free as a NetLogo. It offers different programming languages: Groovy, Java, state-charts and Relogo (similar syntax as in NetLogo). The GAMA generic multi-agent environment can be also useful (http://code.google.com/p/gama-platform/). There are good tutorials and examples. It uses quite untypical language GAML, but I think that it is not hard for learn as in case of NetLogo. The epidemiological model SIR was created with the aid of GAMA (for example). I hope that these advices will help to you :-). Best regards, Martina Husáková

Have you heard of AnyLogic? What I like about it, is that if you need something very visual (not just mathematical), you have it. You can stay on a basic level by playing with the things that you can simply click on, drag on. It can be as simple as dragging your components to the model in the screen, connecting them with arrows, and letting AnyLogic calculate the differential equations and all the rest of the math for you. When you are ready for some serious tools, you can learn a bit about how to customize things more with Java Script. Nathaniel Osgood has great tutorials and examples on his webpage, or atleast used to last time I checked. And I think he offers a "modeling camp" every summer. You don't need to program in Java, just have enough examples to understand which commands you need. The whole thing is very visual. I myself always did my modeling with Matlab or R, but I was exposed to AnyLogic during my PhD, and I have seen what some other students did on it, and it seems very good for people who don't come from a strictly mathematical background and are trying to get into modeling.

@Fernanda: Thanks, Looks interesting! In the end I hope to find something more open source:ish, and I wonder how well it supports biological modeling(?), but for playing around, this looks really nice!

NetLogo has a large community behind it and lots of models (a lot related to biology) in all levels of complexity and it is mostly open. http://ccl.northwestern.edu/netlogo/models/

Samuel: AnyLogic is not only not free, but also very pricey. I just realized that we are in the same city. We meet once a week at SVA to discuss R, and so far we don't have a modeling group, but there are a couple of us interested. Please do write if you think it would be a good idea to start a studying group or something like that: [email protected].

Rafael: Thanks, yes, saw about NetLogo. I first thought it looked a bit "too simple", but maybe it is because most screenshots are from 2D simluations. I saw there are 3d ditto as well. But then of course, pre-made models are valuable for studying how it works.

Fernanda: Thanks! I'm not really working "hands on" in this area yet, but I'm highly interested to go in this direction, so I'll definitely let you know if so, to see if there could be some group formed. meetup.com could be a very nice platform for that (There's an R useR group in Stockholm there already, etc).

Dear Samuel, I think that the NetLogo is a very good modelling tool for the biology. Maybe, it looks like a toy :-), but it is possible to create useful MAS models. Try to look on the wepage http://ccl.northwestern.edu/netlogo/models/ containing models created in the NetLogo. The next alternative can be Repast. It is free as a NetLogo. It offers different programming languages: Groovy, Java, state-charts and Relogo (similar syntax as in NetLogo). The GAMA generic multi-agent environment can be also useful (http://code.google.com/p/gama-platform/). There are good tutorials and examples. It uses quite untypical language GAML, but I think that it is not hard for learn as in case of NetLogo. The epidemiological model SIR was created with the aid of GAMA (for example). I hope that these advices will help to you :-). Best regards, Martina Husáková

I'd like to throw support behind NETLOGO as a (free) environment for learning simple frameworks. There's even a nifty text that gets you up to speed on ABM and illustrates how to use NETLOGO: S. F. Railsback & V. Grimm (2012), Agent-based and Individual-based Modeling: A practical introduction, New Jersey: Princeton University Press. It's good because it is very visual (so makes for good figures), and easily accomodates the number of agents increasing/decreasing. But it can difficult to model complicated decisions or to pull large amounts of data from your simulations (these are possible, but take quite a bit of effort). The NETLOGO language is amazingly easy to pick up (so you won't spend much time learning) and actually kind of fun when you show people the code:

ask turtles [

set color white

forward 2

]

%agents turn white and move forward 2 spaces.

I also use MATLAB when I need to do something complicated (it's easier to manage the output, but it doesn't allow the number of agents to change very easily). I already knew how to program in MATLAB before I started ABM research (in Economics), however, so I found it easy to transfer my skills over.

Thanks Martina and Daniel for the additional suggestions to look at NetLogo.

Actually, I now realized that NetLogo is used in the course "Model Thinking", that I'm taking on Coursera (Here: https://class.coursera.org/modelthinking-003/lecture/16 , just after 4:23 in the video or so). Indeed looks like a nice introductory system. Guess I'll have to take a look.

Dear Samuel, the Coursera courses are great. The NetLogo tool is used also in the course Social Network Analysis (https://www.coursera.org/course/sna). Lada Adamic leads this course very well :-).

Dear Samuel,

I will advice you to start a simple prototype with NetLogo (basic in modelasation but simple to use). Then, when you will be able to embrace most of things related to ABM, you would be able to switch to any other framework (usually hard to use but allow coplex modelisation).

You can read this in order to get an idea: https://www.researchgate.net/publication/235899116_Turtles_are_the_turtles?ev=prf_pub

After growing up for a few years, now this project is now running on a Data Processing/Computing Centre in our University... (http://dsiun.univ-reunion.fr/moyens-techniques/centre-de-calcul/projets/)

Edit: In this article, we propose an agent-based model to study some of the numerous questions regarding green sea turtles (Chelonia mydas) migrations in the South-West Indian Ocean. This paper aims at presenting the simulation model but also the approach.

Conference Paper Turtles are the turtles

Hi Samuel,

I am a proponent of NetLogo as well. It is very helpful to look at the example models in NetLogo for understanding the correct syntax to use and how to incorporate some of these variable into your own model. I started out with no knowledge of how to use NetLogo and over the past two years have been able to build a complex and realistic model of how natural selection may be operating in the species I study. I find the forum online through google groups extremely helpful. I have normally been able to find solutions to my questions by looking through the archives, but I have also posted questions and received great suggestions. My favorite part of NetLogo is the user-friendly interface. I know immediately if I mucked something up in the code by seeing the model run without have to run a full simulation and waiting for the results a couple days later. Good luck with your model - Kristine.

+1 NetLogo is good for learning (indeed, I believe it was built with teaching ABM in mind as a primary goal)

Dear Samuel,

I also would like to support the suggestions from Martina, Kristine and Daniel about NetLogo. I have worked with it to model group dynamics in ecological systems. It is definitely a good tool to start and you can build quite complex models. On top of this it is free, very user friendly (was designed to help school kids) and has an extensive example library.

Maximum success with your models,

Hagen

Nelogo can be a good choice.

But, as Martina said, you can try GAMA as well (http://code.google.com/p/gama-platform/). As Netlogo, GAMA provides an IDE with a high-level modeling language. The GAMA's modeling language is close to the Netlogo's one (as simple - but more agent/object oriented).

Moreover, in comparison to Netlogo, GAMA is more powerful in terms of spatial and visualization features and more adapted to "big" models.

Dear Samuel,

In support of and additions to the suggestions from Martina, Kristine and Daniel; An Agent-based computational model represent a collection of computational models for simulating autonomous agents. Netlogo is an essential and paramount agent-based computational model which has the capabilities for simulating biological systems. You will find these latest 2013 articles about Netlogo very useful. Check the respective links:

http://dl.acm.org/citation.cfm?id=2445518

http://link.springer.com/chapter/10.1007/978-3-642-34300-1_13

http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6455790

http://www.mathnet.ru/eng/mbb114

NetArg is another agent-based social simulator: http://lia.deis.unibo.it/~pt/Publications/aamas13demo.pdf

I agree with Martina. NetLogo seems to probably be the easiest. Repast is also popular and much used, but is probably a bit more overhead and also a bit more powerful.

I'm sure everyone has seen this, but since only three ABM platforms have been mentioned in the discussion I wanted to make sure the full range of options was provided.

http://en.wikipedia.org/wiki/Comparison_of_agent-based_modeling_software

Agreed. Netlogo is a great starting point. However, I find success dependent on individuals ability to relate to the terminology used by the netLogo primitives. I typically have individuals start with Repast. Obviously lots of tradeoffs in these types of decisions. I like establishing new coders/modellers with a development environment and a language that they can use for other purposes in addition to ABM (e.g., eclipse and repast).

Best of luck (good discussion).

Derek.

I agree with Derek Robinson that it is better to chose the agent-based tool supporting environment or language which can be used also for other purposes. Decision is not easy. It depends on your programming knowledge, too. In my point of view, python is a very easy programming language and can be used not only with agents. It has support in Breve environment, but the documentation is not satisfactory for me. The Repast also supports Python, but I think that there are some restrictions. SimPy package is not directly for agent-based modelling, but I read somewhere that the agent-based models can be created in SimPy, too. My experience shows that the Java language is used in most of agent-based environments (As Mr. Robinson said - in Repast), Mason, Jade, Anylogic (in this case you do not create whole model in Java, but you create only pieces of codes in Java), Ascape etc. I think that Java is quite chatty :-). If you want to create agent-based model quickly, mmm, I would not chose Java language. Maybe others readers can have different opinion ...

Mmm, maybe try to look on the following possibility - Agent Modelling Framework (http://eclipse.org/amp/). This is the project which uses the Eclipse environment for the agent-based modelling. In the past, I tried it and the models development does not seem to be difficult.

Martina, thanks for the amp link. I had not come across this before and I will investigate.

I'm sure there is better material elsewhere online, but I've been creating some tutorials for students for installing repast, eclipse, and various plug-ins that may be of interest to people (http://vserver1.cscs.lsa.umich.edu/~dtrobins/abm/abm.html). Cheers.

Mr. Derek Robinson, thank you very much for your useful link :-).

The problem of specific software for agents based modeling even more potent as REPAST or MASON is that impose significant restrictions on the design of the models. If you are flexible on the issue to study any of the options outlined is adequate. But if you have to model a clearly defined problem may be necessary to resort to pure programming languages ​​like Java,Python or Matlab.

@Martina, is there any Agent Modelling framework for Netbeans ?

Dear Muhammad Arif Darmawa,

thank you for your question. Well, I do not know any multi-agent environment using the Netbeans, but I think that Jade as a framework for building multi-agent-based models can be used (except for Eclipse) in the NetBeans, too.

Thank you martina

You are welcome :-).

I agree completely about Netlogo being an ideal start point. But you should be also aware of the fact that if your model's size is big (i.e. in number of agents and/or their connections) Netlogo's performance overhead can make the running time of your model runs (experiments) unacceptable. In those cases, you can use Netlogo to prototype your model and once satisfied with the result re-implement it in a general-purpose PL like Java or Phyton. Netlogo language also lacks the feature of hierarchically defining your agents in an object oriented way (i.e. deriving subclasses from super/base classes). If your model contains a variety of agents which can be modelled in this way than RePast Suite may be a better and similar alternative.

Thanks all for the additional input! I had almost forgot this thread, but is happy to find all your input here now, and will go through it shortly!

I should actually add that I also started to look a lot at the Go programming language, which should be very easy for implementing agents that can run in a concurrent way.

It does of course not provide any framework for space/time etc as of yet, but maybe there will be frameworks for that in the near future.

Anyway, I'll take a look at all your suggestions!

+1 for NetLogo, except if you need a full ecology of agents that you want to build using inheritance, as signaled by +Izzet.

Otherwise, it's language is quite complete and easy to learn (my Ph.D. students learn to master it very quickly in general).

Moreover, its comes with facilities for doing exploratory analysis, and we have supplementary platforms that allow for the execution of NetLogo models on a cluster or a grid (one of them, still in development, but already functional, is OpenMole - http://openmole.org).

I have programmed my models in the past using C++, Java, Repast and AnyLogic, but definitely prefer the convenience offered by NetLogo (open source and freeware moreover).

If you need a massive ABM with millions of agents, you will definitely need to write your model in a speedy language, C continuing to be the quickest it seems. Java seems to compare rather quite well with C++ nowadays, but not with C yet.

I have forgotten to add: There are two books by Volker Grimm and Steven Railsback on individual modeling in ecology and they give examples using NetLogo. One of them aims to provide a tutorial to NetLogo in ecology.

I am just reading the book "Buckland (2005) Programming Game AI by example". In chapter 2 it explains how to build finite state machines (agents) using C++. So if you plan to build a model from scratch perhaps you want to check out this book :).

I am adding one comment to feedback of Murat Yildizoglu. I think that one of the book written by Railsback and Grimm is: . This book is able to give you good basic insight into the usage of the NetLogo for modelling different phenomena :-).

NetLogo should be a good starting point. I wouldn't suggest diving into building your own framework for multi-agent modelling as a good starting point unless one is already well versed in programming.

If you *are* a fluent programmer, be aware that there is a lot of hype about what languages are particularly "good" or otherwise, ... most of it is driven by fashion. I've written multi-agent models in Scheme (lisp), C++, and C, and each had their strengths and weaknesses. The biggest difference between them was where you had to be cunning to get what you wanted out of the language. Unless you set out to be "compatible" with other code, pick a language you are comfortable with.

Good luck!!

I am in complete agreement with Martina on NetLogo. Also, I have used the Railsback and Grimm text she references, as the basic textbook for a graduate topics course on using ABM for society modeling, and it was very successful (easy to teach from, informative, and received well by the students).