Questions related to Agent Based Modeling
The Covid shock
Covid19 caught us by surprise. The previous comparable event had happened 100 years earlier and it had been named the Spanish flu, and it probably killed more people than World War I. The 1919 lessons learnt by health policy makers such as the US cities and European governments had been long forgotten.
Now we have learnt something
This time, it's 2023, and the Covid years, which are not over, have left a clear memory. We have no excuse, not to go prepared.
What about behavioural economists, looking at the exchange of value, time, tasks, anything, in human groups?
What can they tell us, of practical and explanatory interest for the next wave of Covid, or ahead of a similar event?
Behavioural economics scenarios and the pandemic
Can I suggest to look back at the micro-problem replicated a large number of times worldwide, where each household had to self-manage meals, schooling the children, etc. Trade-offs happened at high frequency between the members of the household, seen as agents in a behavioral nano-economy of the house...
The behavioual economist and a vademecum for the next wave?
What do you consider worthwhile for planning the next wave each household likely to have to isolate for while, at least now and then?
Let me share the assessment and model developed for the case of Covid19 "household lockdown":
 Agent Based Model for Covid 19 Transmission: -field approach based on context of interaction, July 2020,R. Di Francesco, DOI: 10.13140/RG.2.2.24583.83364
 "Nanoeconomics of Households in Lockdown Using Agent Models during COVID-19," Sustainability, by Javier Cifuentes-Faura & Renaud Di Francesco, 2022, vol. 14(4), pages 1, February.
 Microeconomics of intertemporal choice in zero-space during Covid-19: a behavioral economics perspective. by Cifuentes-Faura, J., Di Francesco, R., J Health Econ 23, 559–563 (2022). https://doi.org/10.1007/s10198-021-01403-z
Challenge faced worldwide: a new spread of Covid
We all have lived through Covid19, across the world. Prior to the availability of vaccines, Non-Pharmaceutic Interventions by health-aware governments were implemented, with significant success, well into a stage of lockdown, where residents of a country were asked and then required to stay at home, with stringent conditions to get out of their homes. The logistics of food supply was usually well managed, even if there were cases of people remaining isolated from food supply at times.
Anticipating the risk through propagation models
The key to not letting Covid-19 take its toll, and it actually did take its toll, especially among elderly residents of care homes, in Italy, France, the UK, etc, was to model in an anticipatory manner the spread of the disease and assess its risk realistically.
Macro-models available (statistics), but what about micro-level (few humans)?
Modelling was mostly at macro-level: cities, regions, countries. However the different context of human interaction in daily life received much less attention, although large data sets and use cases build on a number of elementary interactions, and smaller numbers of humans involved in each.
Elementary interactions of few humans
Our endeavour, which could not afford the ambition of health statisticians in larger teams, to model the spread of the disease at country level, focused instead then, in the years 2019-2020, onto elementary use cases of interaction, with few humans involved (few starting from 2). Such use cases covered elderly patients of care homes, and their interactions during joint meals in the care home meal area, with tables shared, it also covered households in close (and closed) interaction during lockdown. It also tried to make sense of large events, where many humans interact during a limited time (football game, women's day celebration, etc).
The typology of likely propagation in such use cases was modeled, and parameters of a simple but robust model were tuned to known data, and in turn simulations could be run, and such simulation could be assessed on other known outcomes (such as the observation of virus propagation among the citizen team running a polling center during elections in France).
Next steps: anticipating the wave coming, with micro-models?
Can we ask the researchgate community if anyone is interested to undertake similar micro-level models of elementary human interaction leading to a likely spread of the virus?
Could we consider building a federated collaborative project, with data fed by anyone having access to these (literature, publications, etc)?
What approach do you recommend? Have you published on the topic?
Here is a reference to the model mentioned above, with associated training/verification data:
 Agent Based Model for Covid 19 Transmission: -field approach based on context of interaction, July 2020, R. Di Francesco, DOI: 10.13140/RG.2.2.24583.83364
Topic plan: Reallocation water resources from agriculture to industry
Effective agents: a group of farmers, industry, regional water organization, ...
Question: Is it possible to use the agent-based model to simulate interactions between agents?
The fact that there are individual government agencies and a representative must be chosen for each agency, while farmers are a group and several thousands of farmers can be defined.
Does the difference in the number of agents cause problems for modeling?
Is it better to use Multi Agent system or Agent Based Model? What is the difference between Multi Agent Systems and Agent Based Model?
My Awesomest Network , Could I ask You politely a bout recommending the best the most cross - sectional the most extended books about artificial intelligence and about agent based modelling agent modelling too respectively, please?
Thank You very much in advance
Animations are known to be a fast and very efficient way of dissemination of knowledge, insights, and understanding of complex systems. Through the animations, quite complicated research can be easily shared among all scientific disciplines.
While starting with complex systems descriptions of Dynamic Recrystallization in metals about almost 30 years ago, it had become very obvious almost instantly that animations carry with themselves a huge expressive power.
This recently led to development of the GoL-N24 open-source Python software that enables to create animations effortlessly. The user just defines the input parameters and the rest is done automatically. Share your software too.
This question is dedicated to all such animations and open-source-source software, which are producing them, in the area of complex systems.
Everyone is welcomed to share their own research in the form of animations with the relevant description.
I have a set of positive COVID-19 cases by district in a state. I want to use agent based modelling to make the infected cases move around the state and count the number of infected, number of healthy and number of immune. I have a set of code which the number of population is random and not based on the location. Tried to watch youtube for the tutorial but could not find one which really helps. the code is as below. I want to insert the map of the state, number of population for each district and number of infected in Netlogo.
create-turtles 200 [
setxy random-xcor random-ycor
set shape "person"
set color pink
set size 1.0
set illness 0
ask n-of starting-number-infected turtles [
set color yellow
set shape "X"
set size 1.0
ask turtles [
rt random 360
ask turtles [
if color = yellow [
ask turtles in-radius infect-distance [
if random-float 200 < infect-chance [
if color = pink [
set color yellow
set shape "X"
ask turtles [
if color = yellow [
set illness illness + 1
if illness > infectious-period [
set color white
set shape "circle"
set size 1.0
set illness 0
ask turtles [
if color = white and waning-immunity < random-float 100 [
set color pink
set shape "person"
I'm starting to model an urban simulation and I'm having a bit of a dilemma regarding what language to use. Have some experience in Netlogo and I'm starting to make a shift towards Repast, GAMA or MESA (geo-mesa), because it is recommended for large scale simulations.
Have been reading papers about which tool to use, but I need someone working on simulations to help me out.
Still, I have questions because:
1- The user base of MESA is scarce and i feel that dealing with issues will be dificult
2- So far i have only seen and read about limited research done in MESA. Specially, dealing with road network integrations. (move an agent along a network)
3- It seems that Netlogo is good for prototype, will not handle big data projects
Thanks in advance, and any pointers to courses or moocs would be great.
I'm doing a research about simulation modeling and I've been using anylogic for a couple of weeks now to collect information for my project.
I know anylogic is a very flexible software with a java base but I want to know if there are any issues you ran into while using anylogic or if there are any fields/industries that doesn't work with anylogic very well.
Please feel free to write down any issues you noticed while you were using anylogic and any opinions you have about the software.
Thank you for your time.
Hello everyone, I hope you all are doing well.
I am preparing for a master of science nowadays,
and I am searching about Agent Based Modeling in Construction Management for my thesis topic, I do not have any background about this topic so from where can I start to search or read?!
Complex systems are becoming one of very useful tools in the description of observed natural phenomena across all scientific disciplines. You are welcomed to share with us hot topics from your own area of research.
Nowadays, no one can encompass all scientific disciplines. Hence, it would be useful to all of us to know hot topics from various scientific fields.
Discussion about various methods and approaches applied to describe emergent behavior, self-organization, self-repair, multiscale phenomena, and other phenomena observed in complex systems are highly encouraged.
Do we know theoretical models how to share benefits between a focal company and for instance 3 tier 1 supplier's using a supply chain finance programme?
so what is ‘fair’
for instance there is a net saving of 1.5 euro using a SCF reverse factoring programme. Is the following split up fair or not, and why from a theoretical point of view:
€ 0.3 to the focal company and 3 times € 0.4 to the (3) tier 1 suppliers
I am studying walking behavior by agent based modeling. Do you see proper Anylogic software on it? Or another software?
Also, do you know any sample in which analyzed walking behavior by Anylogic?
For example, I am particularly interested in the consequences of the UK's new global policy for India, China, and Australia. Could a well-designed agent-based computer model, using appropriate data including about individual world leaders, help predict any dangerous consequences of this new policy?
We call Donald Trump and Boris Johnston and his cat and the mice that the cat hunts and the fleas on the mice and the coronavirus etc LIVING. But Boris's desk and his TV and the computer program that beats him at chess and and his electronic android servant (soon!) and windmills (cf Don Quixote) and holograms and ghosts we call NOT LIVING. But what is the difference? As a mathematical kind of person I would really like to know ... and rather precisely... it matters for agent based modelling etc
Note -- the difference is surely not about reproduction -- computer programs and machines can easily generate and distribute copies of themselves.
Basically I would like to know if the FEM solution obtained by Comsol could be used as input that could guide agents to behave in a certain way in the same common mesh. For example stem cells on a scaffold with specific material properties that undergoes a certain type of deformation. What is the most appropriate ABM Software? Any guidance or tip on the procedure that must be followed in order to interface it with Comsol Multiphysics 4 would be valuable
Offline reinforcement learning is to learn from a fixed dataset without further interactions with the environment. There are many papers focusing on offline RL for a single-agent setting. e.g. https://offline-rl-neurips.github.io/papers.html, However, no literature research on how to learn in multiagent settings. I have tried several state-of-the-art offline reinforcement learning approaches, but it doesn't work well in a multiagent environment. My code can be found here: https://github.com/SHITIANYU-hue/multiagent-offline-RL. The main challenge is how to learn multiagent distribution.
The evolutionary game model and agent model can be used to study the decision-making of micro individuals. What are the similarities and differences between them? What scenario is applicable?
Recently, I read a few papers that used agent based models. Now for me, what seems beautifull about this type of simulations is that they are less rigid than simulations of differential equations using traditional methods, but somehow the papers always begin with the set of differential equations that the author wanted to describe.
needless to say, I don't think the presentation of the equations is necessary
Why is that?
I am studing about predicting pedestrian behavior when we change density and land use paterns and ect with agent based modeling. If you know any book or articals about it, please inform me.
Many countries are in lockdown to reduce the impact of Covid19. This method has been proven to work in China (Wuhan) and previously in those US cities which triggered it early and firmly in the 1918 influenza pandemic.
Now what are the exit scenarios?
The French government has said last week (first week of April) that it was evaluating multiple scenarios to exit the lockdown, step by step, region by region.
Some articles have been published casting light on some aspects :
What are the possible exit steps?
Is testing a must prior to this as shown in an article by Economic Nobel Prize Winner Paul Romer?
Please share your views and input, with needed references
Nowadays, R play an outstanding role in Ecological research. It provides a free to use platform for innumerous statistical analysis and a huge collaborative community of developers and users that share codes and help each other with programming and statistical issues.
For many technical reasons, R is not a computational efficient language and, because R is not much popular among computer scientists, progress in computational efficiency is slow. Differently, Python is very popular among computer scientists, it experiences constant computational efficiency improvements and it is much more efficient than R in many aspects. Also, Python has gained some attention in Ecological research similarly to R.
So here is my question, with the increasing computational requirements of current analyses (e.g., Heavy Monte Carlos resampling procedures), will Python become as popular as R in Ecology?
The reward is necessary to tell the machine ( agent ) which state-action pairs are good, and which are bad.
Please help me to understand the behavior of the discount factor or reward in terms of reinforcement learning.
What I don't understand is why the discounted reward is necessary? Why should it matter whether a good state is reached soon rather than later?
I am interested in creating a multi-layer mechanical network. Therefore I would like to find a software where you can visualise nodes and links moving around in 2D and 3D space.
In many publications on the topic of modelling residential mobility I read, agents are regarded as households and can only make decisions (here relocating) as one single entity. The question whether an agent with given characteristics has a propensity for changing housing or not is often answered for an entire household. Restricting households to only be able to stay or move as a whole seems not appropriate. An alternative would be to compute/model on the individual level and take the household type into consideration. However, individuals as part of a household should still be able to move together which would be rather unlikely in this case. How is this taking care of in research?
Agent Based Model has been used for different simulations. but i do not know that can we use this method to policy testing at household level or not?
I am modeling the collective behavior of random walkers (using CTRW) on a 2d lattice and I am having trouble finding a "correct" rule that won't eventually violate the uniformly random motion of my agents. Any ideas?
I am developing an agent based model about citizen's decision making of daily mode choices which I have already 2 variables that affect the choice of people and I am looking for a new factor to evaluate its impact on mode choice.
I'd appreciate your answers
I'm trying to use Agent Base Modelling to show how the effective the different behavioural change mechanisms in the China Social Credit System are at influencing society. These mechanisms are surveillance, rewards and punishment and gamification (scores are made public).
Could anyone point me in the direction of which models would be good to look at in order to build something like this?
Any recommendations or suggestions would be appreciated as I am now currently learning how I do this.
I am doing agent based modeling and was wondering, if I can have a model that does not specify a resource the agents use, but apply a density dependent function to decide how likely each individual is going to reproduce? So the probability of reproduction by each individual is determined by agent density around it. Or should I add in another parameter, which represents resource in the grid, which can be consumed and regenerate? But that will make my model so complicated and very slow to run.
I am looking for an accessible, but non-trivial introduction to agent based models in transportation, preferably with exercises and assigmnets. Of course, you can find a lot of information on Google, but it is rather difficult to identify a source that is up-to-date and that meets all requirements.
What are some topics related to sustainable manufacturing that are still in need of development and future research?
My masters thesis must incorporate some type of system modeling geared toward reducing energy Costs, however I am having diFficulty finding a topic that is unique and under developed. my Literature review consists of topics such as: production planning, predictive maintenance, buffer optimization, reduction of peak demand, systems which utilize renewable energy and or chp systems, agent based modeling, machine learning, etc.
I started a new project on ABM for criminology and bumped on GIS and crowds movements. What I need is to simulate the movement of people commuting to work every day in a map extracted from OpenStreetMaps (OMS) platform. I program in Python, so a solution in Python would be ideal (and very convenient) for me. What library/toolbox/guide would you suggest for a newbie in GIS and simulation of crowds?
On the recent 73rd anniversary of the Hiroshima nuclear attack I took part in a informal discussion targeted at exploring possible ways of preventing such terrible catastrophes in the future.
I give below my sketchy notes for my contribution to the discussion. I feel that they may form a useful start point for a discussion in the RG forum.
CONTRAST this pleasant environment with cooperative people enjoying afternoon tea -- some staff working, true, but they are moderately fairly compensated -- with the horrific destruction and killing of 70,000 in Hiroshima in 1945.
How do such things come about (eg Battle of Kursk, Tai Ping Rebellion, Spartacist Revolt)?
I look to human basics: greed, pride, ambition, family loyalty, aggression, and other fundamentals.
Adopting an evolutionary perspective, I ask: is it just human beings – and what is the evolutionary benefit? Consider the behavior of other great apes and chimpanzee “war parties”.
The late Steven Hawking suggested, as I recall: that initially in human evolution aggression was useful, but now more empathy is urgently needed.
A KEY FACTOR – massive development of technology! Consider the origins of nuclear weapons – Einstein, WW2 and Manhattan Project
How possibly control to avoid human self-destruction in foreseeable future.
Well --- computer modelling of the human global social system, what it is, its feasibility (or otherwise) in this context and the problem of how to control the seemingly massively complex global human system – which, alas, we are part of, including our decision making!
Best answer yet - experiment assuming LIMITED intervention possibilities in the complex system, and model to explore most effective (or least ineffective) intervention strategies.
Time is not on our side.
I came across papers by Sigglekow, Knudsen, Gavetti, etc and they use agent based simulation to generate NK based complex models. From the few paper that I read, none of them mention the software that they use. So, I am looking for recommendations - as a beginner, what would be the best software to learn to be able to simulate such complex models in strategy?
We are working on a project on developing methods for functional agent-based modelling. We have implemented a SIR model using Haskell and Yampa, a functional reactive programming library.
Is it possible to do some correctness proving / formal verification in agent based modelling? Can you guide us to the relevant literature that explains the process?
I am working on a project "Agent Based Modelling for Flood Risk Management under Different Climate Change Scenarios" and need some research papers or book chapters for literature review in similar fields.
It'd be greatly appreciated if you could provide any paper/chapter/link?
I am looking for examples of the combination of ABM, MO optimization, and game theory, preferably the ones that have been used for practical purposes.
In the context of electricity market modelling, what are the main differences between ABM and SD?
I know there are a number of general differences between these two approaches. For instance, regarding the systems containing active objects (such as people or organizations), ABM can go beyond the limitations of SD. However, I am looking for a more specific answer with regard to the characteristics of electricity market.
I would like to use differential equations for ABM based simulation in organization studies. Can anybody provide me with suitable examples of such research?
In the context of urban (spatial explicit) modelling, what is the difference between agent based simulation (ABS) and micro simulation (MS)?
Both are individual based modelling approaches. Are ABS models typically theory driven while MS models are typically data driven? Do ABS models contain intelligent and non intelligent objects (have a memory; capable of making decisions based on their own judgement; capable of learning) while MS models only contain non intelligent objects (making decisions based on distributions derive from historic data)? What is the simulation execution algorithm they follow - do they differ?
Please provide some simple examples that clarify your explanations.
In my point of view the simulation of autonomous vehicle needs to consider the interaction between the vehicles and associated infrastructures. Hence, I wonder what types of communication(interactions) in your project is going to be study through an agent based modeling.
I am looking for tool/s or game/s that may be actual or simulation of agent-based decisions/behavior to test or model decision making patterns.
One such example is IBM's Watson which is working in various fields such as Healhcare systems, education etc.
I would like to know if there is any attempt to model cooperation inside organizations, corporate culture (a la Kreps), or in general Principal-Agent models, where agents rely on private/public information.
Both agent-based modelling and cellular automata begin with individual agents or cells and model a large number of interacting agents leading to complex behaviour. However, what are the differences and pros and cons of each approach?
I am phd student and a newbie in the domain of trafic engineering. I am working on modeling road trafic in urban areas. One of the questions of my work is how do vehicle arrive on a specific road based on lane, vehicle type and traffic load.
I am not sure if my question is well put or is it too much detailed for a research. So I am asking you.
Is it really interesting to work on a random road trafic generation on a microscopic level? is it useful in studying all kind of roads or is it specific on roads where accuracy of arrival is needed like road pricing ?
Regarding innovation diffusion, is there any model for knowledge transfer? i.e. inform technology A diffusion process with technology B process?
The data of the input and output is not present we only know the behavior and interaction details how to validate the model in this case
I want to calculate reputation and trust based on direct information and witness information in my simulation. I was wondering if anyone has an experience about these two subjects. Or, if you know any samples\ software regarding to these areas let me know.
Thanks in advance,
As much as I have read, most of the work on multi-agent-systems and thereby,on design of an agent, JADE (or other similar platforms, say JANUS,GAMA,etc) has been extensively used to model a single agent and the entire agent-based-framework.
My question is:
Is it acceptable/standard/suitable to design/model an agent as a user-defined function/class (taking-in some input arguments and yielding some outputs), whose some of the inputs may/can be outputs of other agents(also modeled as functions/classes) and its outputs may/can be inputs to other agents(also modeled as a functions/classes), without using the JADE or similar platforms?
System dynamics being a quantitative approach I'm not sure whether the social aspects can be included. But if there is an option please suggest me.
For instance, FLAME, MAMID, EMA, GRACE are computational models of emotions. What are the different parameters on basis of which, output of these models can be compared.?
What are the different ways to validate such model?
Can anyone help me to find how to measure the health insurance policy holders behaviour? What are the determinant factors for purchasing or not purchasing the health insurance? How to involve Heuristics, Endowment effect, loss aversion ,prospect theory, satisficing and strategic thinking as a construct for preparing the questionnaire? What are the other constructs or variables available for measuring the behaviour of health insurance policy holders. Whether is it possible to apply agent based modeling for knowing about behaviour and goal of agent.Please suggest some variables?
I'm trying to introduce an opinion formation model in social networks using agent based modeling. so how could i evaluate our model to verify and validate it? must i compare it with some samples of real world? is there any different evaluation method?
what i see in the literature is researchers introduce a model based on some theories and explain the behavior of social or system (by simulation or analytically). they have no comparison with real world or other models. so how can we determine a model presents more realistic results and work more appropriate than other models?
Build a generic ontology for Agent based modeling and simulation and generate any Agent based model by configuring / specifying the ontology in a particular for a particular domain. In other words how to move from a general ontology for Agent based modeling & simulation to a domain specific ontology that captures Agents and their behavior in a given domain?
When one should use Agent based models for doing "What-If analysis"? What is unique about Agent based models with respect to "What-If analysis"?
I am working on on Complex Adaptive Systems. So far that I have studied the tool suggested for studying the Complex Adaptive Systems is Agent Based Modeling. I have not seen the Agent based Modeling yet but just a query in my mind. Does Agent Based Modeling come in the qualitative domain or quantitative domain?
Also, kindly suggest me some good book or source for learning the agent based modeling.
The agents are in a simplistic trade environment were an agent can buy a resource from potential customers (having various prices). However, the aim is to train the agent on when to buy (optimizing his payoffs) given the limits of time, energy and competition among other agents.
I'm interested in designing a market model with flexibility on both demand and supply side. Agent based modelling approach is considered for the work. Could anyone suggest such existing modelling attempts? Would be a good guideline for my work.