Dear Peer-Olaf,

I think we might have met in EASSS 2009 in Turin... but maybe I'm wrong.

Anyway, from my perspective, it all comes down to your first question... and the answer is probably "because most people actively working in this area do not really come from Computer Science"...

From my understanding, moreover, it is relatively simple to use a class diagram to capture static aspects of a domain, which in this context would mean to represent agents, relevant aggregated abstractions, relevant 'objects'... it would also help giving an explicit account of additional "things" (in a better English abstractions) you put in the model to manage some mechanisms... but you'd need additional diagrams to capture dynamic aspects of the model, and this would make things more complex (you probably need to be quite good at UML modeling to really capture all the necessary aspects)... and frankly I think it would be difficult to achieve the level of formalism you can get with equations.

I guess you might be interested, on one hand, in all the systematic efforts carried out in the AOSE community about meta-models and modeling methodologies. On the other hand, I think that the idea you propose of defining a systematic way of translating equations into agents is original, and you might actually exploit work carried out in the AOSE community.

Sincerely

Giuseppe

Dear Peer-Olaf,

I think we might have met in EASSS 2009 in Turin... but maybe I'm wrong.

Anyway, from my perspective, it all comes down to your first question... and the answer is probably "because most people actively working in this area do not really come from Computer Science"...

From my understanding, moreover, it is relatively simple to use a class diagram to capture static aspects of a domain, which in this context would mean to represent agents, relevant aggregated abstractions, relevant 'objects'... it would also help giving an explicit account of additional "things" (in a better English abstractions) you put in the model to manage some mechanisms... but you'd need additional diagrams to capture dynamic aspects of the model, and this would make things more complex (you probably need to be quite good at UML modeling to really capture all the necessary aspects)... and frankly I think it would be difficult to achieve the level of formalism you can get with equations.

I guess you might be interested, on one hand, in all the systematic efforts carried out in the AOSE community about meta-models and modeling methodologies. On the other hand, I think that the idea you propose of defining a systematic way of translating equations into agents is original, and you might actually exploit work carried out in the AOSE community.

Sincerely

Giuseppe

Hi Giuseppe

Your name rings a bell :).

I usually get away with using UML class and state machine diagrams to create my conceptual models and I use the same diagrams for the implementation (in AnyLogic, a multi-paradigm simulation package). These diagram types seem to be relatively easy to learn (at least this is what my students say).

I think the reason that people do not use UML is that they have traditional ways of setting up models which they don't like to change (perhaps it is difficult to get published if you use an odd way of describing your models) - or they don't see a need or benefit in using a different method. But I might be wrong ...

I have some experience with searching for systematic ways of translating equation based models into object oriented agent-based models (UML state machine diagrams). I worked with a colleague from Systems Biology on this topic, trying to translate mathematical immunology models (ODEs) into state machine diagrams. We concluded that rather than translating the models it is often best to build the agent-based model from scratch (and use the ODE models for calibrating / validating the agent models) - as we encountered a lot of problems, in particular when models became more complex. The underlying assumptions that are valid for the mathematical models are not always valid for the agent-based models (e.g. in an ODE model a cell can be fractional dead while in an agent-based model (as well as in reality) the model is either in an alive or dead state (it is a binary decision).

I will definitely have a look at the AOSE website to see if I can find some ideas there. I also have a PhD student who has started investigating the topic, so I hope he will come up with something that helps to promote the application of UML in Social Simulation and Economics.

Thanks again for sharing your thoughts - much appreciated :).

Some related papers:

Siebers PO and Aickelin U (2011) 'A First Approach on Modelling Staff Proactiveness in Retail Simulation Models'. Journal of Artificial Societies and Social Simulation, 14(2) 2.

Figueredo GP, Siebers PO and Aickelin U (2013) 'Investigating Mathematical Models of Immuno-Interactions with Early-Stage Cancer under an Agent-Based Modelling Perspective'.

I have recently given a presentation at the GTMooS-SIG (http://www.cs.nott.ac.uk/~pos/gtmssig/) to some Behavioural Scientists from NIBS (http://nottingham.ac.uk/esrc-nibs/). It provides some more explanations and perhaps clarifies my questions and lists some resources related to the topics. You find it here: http://www.cs.nott.ac.uk/~pos/gtmssig/slides/gtmssigSiebers2013-09-13v3.pdf

Are you aware of this paper:

Hugues Bersini (2012): UML for ABM. Journal of Artificial Societies and Social Simulation 15 (1) 9

?

Some very subjective answers from me:

(1) Why is UML not used more frequently in those fields?

--> Because most people working on agent-based economic models are not computer scientists

(2) Can we use the UML instead of (or besides) equations to describe (better conceptualise) agent based models in those fields?

--> UML-diagrams will probably help, as they might complement and enhance the flowcharts often used to illustrate the model.

(3) Are there any (obvious) rules to translate equations into agents?

--> I wouldn't know of any that I followed in my work so far.

Well, a straight and a plain 'Yes' is the probable solution to your answer.

Translation from UML to any mathematical model is possible and there are many such models available. And then further these can be tuned in as per your requirements. The last question about mapping of equations to agents can be done using rules while software development.

I have used such mixed approach in my research; came up with a mathematical notation and used translation rules for UML to CPN and then used it in implementation through agents. I hope this helps. Best Luck