To view this email in your browser click here.

Using SIMUL8 with Agent Based Modeling

In this issue of our Newsletter we look at a hot-topic in the world of simulation - agent based modeling. We explain how you can use agent based simulation, its potential benefits, and how SIMUL8 can be used to emulate many of its processes. Read the full article below.


SIMUL8 Support Update

Whether you’re new to SIMUL8 or you’ve been using our software for years, we know you’ll have some questions. That’s why we have a range of support services and a team of experts to help you. Click on one of the links below or contact us to discuss your options.

ONLINE HELPDESK

For free feature support or questions about our products or services.

SIMUL8 TRAINING

SIMUL8 public training courses and on-site coaching sessions.

SIMUL8 CAFÉ

Hints, tips and advice from our SIMUL8 user community.

 

SIMUL8 NEWS

 

WinterSim 2009

From resource planning to reducing waiting times, SIMUL8 has become more widely used in the healthcare industry.
 
Click here to see how John Hopkins used SIMUL8 to plan resource for their Comprehensive Transplant Center.  

Tell us how you've used SIMUL8 and you could win an iPod touch.

Following on from last years success, SIMUL8 are excited to announce our attendance at WinterSim 2009.

WinterSim will take place in Austin, Texas from December 13th to 16th and we'd love to see you there.

If you're heading to WinterSim and would like to arrange a meeting or want anymore information about the conference, get in touch using the  form here.

 

SIMUL8 RECRUITMENT

Graduate Business Analyst (Healthcare)
Supporting the Lead for Healthcare you will provide training, support and consulting to a range of healthcare modeling and simulation projects. Click here for more information or apply now.

 
September Feature Article
Using SIMUL8 with Agent Based Modeling

Recently we’ve been looking at agent based modeling - a simulation technique that’s been used since the 1990s by academics. It’s becoming more popular in industries such as travel, safety and marketing because it’s more oriented to the way those industries think. We’ve been assessing how SIMUL8 can be used for agent based modeling and whether there is enough demand and benefit in the technique for us to add specific functionality to SIMUL8, and we'd like to hear your views.

There are three types of simulation relevant to business: Discrete Event Simulation (DES), System Dynamics (SD) and Agent Based Modeling (ABM). SIMUL8 started as pure DES but in version 3 we added some SD functionality because some users need to combine the two.

ABM takes yet another perspective on simulation. Rather than modeling from an overall picture of the flow in a system, ABM is centered on interacting individuals with a view to assessing the system-wide effects of their individual behavior and interactions. Those individuals are known as the agents in a model, typically replacing work items (entities) in a DES model.

Instead of a centralized or global simulation control, ABM attaches the rules of the system to the individual agents - unlike DES where work-items are passive and the actions are defined by the activities that process them.

Thinking in terms of a DES model of an airport, passengers are pushed or pulled between check-in and security processes and it gets tough to model all aspects of the airport (some passengers stop off for a caffeine shot then browse the book store). With an agent based mindset the passengers are in control and making their own decisions on where to go and when - just like in real life. (Although we are not sure the TSA sees it that way!)

 Benefits of ABM include:

Consideration of the variations in individual behaviour.

Models situations with interdependencies between work entities.

Spatially explicit - allows agents to have a physical size and for relationships to depend on the distances between them.

Free movement - agents can move around their environment with no need to pre-define all the possible routes that can be taken.

This makes ABM particularly suitable for modeling situations where large numbers of humans are present and each make their own choice between many alternatives. It makes it easy to include individuality and see the impact on the overall system of the variations in different people's behavior.

 
Where ABM is Used

Most often ABM is used to model systems where outcomes have a high degree of dependency on the actions of humans. Common applications include the spread of diseases or information between populations; people or traffic movements, the impact of marketing campaigns etc.

 

For example in Healthcare diseases like flu and MRSA can be modeled using agents whose infection status changes over time. It starts as “uninfected” may become “weak, could be infected”, then “infected, likely to transmit” etc. Infections can be transmitted based upon contacts between people, while taking proximity and time of contact into account. Behavioral aspects can consider that people showing symptoms are likely to stay at home to limit their contacts and further spread of the disease. For hospitals and health planners these models can be used to test out the most effective strategies for stopping the spread of epidemics.

 
Agent Based Modeling using SIMUL8

Agent Based Modeling has yet to become really popular as the practical commercial applications are limited. It’s also often very difficult to obtain the data you need to make an accurate simulation. For example: What is the likelihood that I will buy a package of TV channels because I have seen one of the channels in a friend’s house?

However, like with the combination of parts of SD with DES, we see real potential in the combination of DES, with ABM. This gives the process view required to make process decisions but also gives the additional influence on the system of the mass combination of individual behavior.

With SIMUL8’s current functionality ABM simulations can be built. It is just a different way of thinking. Just like agents, work-items in SIMUL8 can be as unique as necessary. Labels can be used to represent individual attributes and preferences, and then used in incorporating behavioral differences. Of course, values for each of these can be sampled randomly from a variety of distributions or set precisely if desired.

Actions can also be performed on individual work items at particular times. Using the “Schedule Event On Object” Visual Logic command, in combination with unique work item ID labels, events can be scheduled that act on specific work items – perhaps to change a label value representing their current status of a disease.

Currently all software that offers explicit ABM functionality is difficult to use requiring programming in languages such as Java that make the learning curve really steep. But we’re experts in making simulation easy to use so we know we can overcome this. We’re interested to know what your views are on ABM and whether it’s functionality you’d like us to introduce or whether you’d prefer us to invest the time in developing you more fantastic DES features, so please take a minute to fill out our ABM questionnaire here.

 

 

You are receiving this e-mail because you subscribed to or are a registered SIMUL8 user. To unsubscribe click here.