ABM get-together summary November 2014

Here’s a quick reminder of the talks from the ABM get-together on the 14th November 2014. We were ~30 people this time – just about maximum for the room in the FHI, thanks for squeezing in!

Ridhi Kashyap described a model she is creating to help understand why in some parts of the world there is a 130:100 males to females ratio. Ridhi is drawing on the Schelling’s segregation model to show how relatively small preferences for males before birth (i.e. abortion) can lead to such gender imbalances. Andreas mentioned the similaries with his population and cemetery simulator for constructing living populations from graveyard records.

Matteo Richiardi described the problems modellers have in calibrating agents in ABMs and evaluating the usefulness of a model, especially when it generates data that doesn’t correlate with empirical oberservations. Matteo’s work is focused on labour markets and he is working on JAS, a new Java based ABM simulation environment for discrete event modelling to help with this research. Felix suggested the Scaling and criticality in a stochastic multi-agent model of a financial market paper. Slide: Presentation – Matteo Richiardi (slide contains transitions i.e. press forwards!) Matteo also has some course slides and literature on ABM in economics

With Uri Wilensky we discussed how ABM is part of most courses at Northwestern University although there are similar controversies about this system of representation in economics. Uri brought up the topic of restructuration (Wilensky, U., & Papert, S. (2010). Restructurations: Reformulations of Knowledge Disciplines through new representational forms. In J. Clayson & I. Kalas (Eds.), Proceedings of the Constructionism 2010 Conference. Paris, France, Aug 10-14. p. 97.) e.g. it being easy to add and subtract but not divide and multiply with Roman numerals. Uri also asked Wybo to send him a link to AgentBase because there are similarities with the forthcoming javascript implementation of NetLogo. Uri will publish a new book in March 2015: Wilensky; William Rand (in Press). An introduction to agent-based modeling: Modeling natural, social and engineered complex systems with NetLogo. Cambridge: MIT Press.

Wybo Wiersema showed us his new tool called AgentBase that can be used to create ABM entirely within the browser. Wybo intends to use Agent Base to model recent uprisings or political rebellions. Slides: AgentBase: Agent Based Modeling in the Browser

Gemma Poulter told us about a model she is building with colleagues at the Rutherford labs and Sheffield University that aims to support patient pathway planning in hospital A&E departments. The model is currently running on real live data using the FLAME software. Due to Eduroam issues Gemma wasn’t able to show us the system but she has kindly included some screen shows in the slides: Modelling patients’ pathways through A&E with FLAME

Posted in Meetings | Comments Off

Next Oxford agent-based modelling get together

Interest in Oxford in agent-based modelling continues to grow, so we thought it is time to have another informal gathering to:
·  see who’s working on what sorts of problems
·  find out what software and packages are being used
·  see who has what expertise
·  create some new collaborations
·  assist those who are new to the technique
·  reinvigorate the network of existing informal contacts
This time we are very lucky that Uri Wilensky, the ‘father’ of NetLogo, visiting from Northwestern University will join us.
Everyone is welcome whether you’re an undergraduate, postgraduate, research fellow, or academic staff. It doesn’t matter which department you are in, what topics you’re interested in, or what level of knowledge or experience you have.
We are offering 5-minute presentation slots to anyone who wants to discuss thoughts on ABM – particularly problems that might be well-suited for ABM or work-in-progress. This time we will be strict about the 5 minute limit but are encouraging people who may just want to ask questions to sign up for a slot. Please send email to kenneth.kahn@it.ox.ac.uk
Date:            Friday 14 November
Time:            2:00pm – 4:00pm (though in previous meetings some stayed past 5pm)
Venue:         Future of Humanity Institute, Suite 1, Littlegate House (1st floor, on the left)
                      16/17 St Ebbe’s Street, Oxford, OX1 1PT
                      http://www.fhi.ox.ac.uk/about/contact/
Cake and refreshments will be provided.
Let your colleagues know about this.
If you’re unable to come along but would like to keep in touch, please send an email to kenneth.kahn@it.ox.ac.uk
We hope to see you on the fourteenth of November.
Ken Kahn and Howard Noble (Research Support, IT Services)
Andrew Snyder-Beattie and Anders Sandberg (The Future of Humanity Institute)
Posted in Meetings | Comments Off

New release of the Behaviour Composer

This release fixes bugs and updates the home page and related pages. The Behaviour Composer to NetLogo tool has also been updated to fix a bug that was triggered by plot, monitor, or other widgets with an ampersand in the label. No need to download the new version unless ampersands in titles of NetLogo widgets are needed. Full details of source code changes are here.

The new Modelling4All home page

The new Modelling4All home page

 

Posted in Software development | Comments Off

Agent-based modelling at the University of Oxford

IT Services offers an agent-based modelling (ABM) service – sound esoteric? Not so!

ABM is a simulation technique that emphasizes modelling the interaction between autonomous agents or individuals. Think about the way ants forage – each ant might have a behaviour to move randomly until it finds some food, at which point it heads back to the nest and secretes a chemical signal (pheromone). The pheromone attracts other ants towards the food source, and guides them towards the nest too. An ant by itself would be extremely inefficient at gathering food, but many ants with this collective chemical memory are very effective. But it is more complicated than this – the ants have to secrete a pheromone that evaporates and diffuses at the right rates!

You can run an ant foraging model here.

Ant foraging is just one example from zoology, but researchers are using ABM in nearly all disciplines, but especially ecology, epidemiology, economics, political sciences, anthropology and public policy.

A video of a model of the Spanish Flu Pandemic that our team built illustrates the use of ABM in both epidemiology and history.

So why are researchers creating models? It must be to predict the future right? No so, and one of the fathers of ABM went out of his way to point out that there are at least 16 good reasons to create ABMs other than prediction.

Students learning to model need guidance and an example is this tutorial for modelling an epidemic that spreads over a social network (does not work with Internet Explorer 9 or below).

How do we support researchers at Oxford? The research support team provides specialist advice in a number of key areas aligned to the IT Services strategic plan for research. In each domain we actively engage the research community by offering taught courses through the ITLP, presentations for research groups, and web-based resources. In all cases we aim our services at the long-tail i.e. the population of researchers that need support getting started, or research groups that want to experiment with this technique and need hands-on practical support (someone to write the code).

In the case of the ABM specialism we maintain and support a web application called the Behaviour Composer, a pioneering tool for creating and sharing NetLogo simulations online. The Behaviour Composer was created in the modelling4all.org project with funding from the Jisc and Eduserv between 2007-2010. We won a University of Oxford teaching award in 2009. We also meet researchers on request, and charge for bespoke teaching for research groups, and for supporting work packages on research projects.

Here’s a few examples to illustrate the number of different ways ABM is being used by researchers at Oxford:

  1. Created an ABM to explore the Modes of Religiosity theory, the model later became part of a doctoral thesis (Institute of social & cultural anthropology, European Commission project – Explaining Religion project).

  2. Co-designed and teach 3 hour practical on using ABM to think about the spread of a virus through different social network topologies (Zoology)

  3. Help teach a 2-hour practical on using ABM to think about wealth distribution (Said Business School)

  4. Designed and built a game-like ABM to discuss cropping decisions with farmers in Cameroon (Institute of social & cultural anthropology)

  5. Support researcher in creating a model of living populations that are based on data from medieval graveyards(Archeology department)

  6. Created the Epidemic Game Maker a tool for creating models of the spread of an infection for the Royal Society Summer Science Exhibition (Zoology department)

  7. Supported researcher in getting started with a model of the Arab Spring and commented on his transfer document (Oxford Internet Institute)

  8. Bespoke ABM course for agricultural genetics researchers from Edinburgh and Alberta Universities (Joint Canadian and ESRC project)

  9. Taught a course on how to teach ABM in UK universities (Life Sciences HE Academy)

  10. Supported postdoc in creating a model of cancer cells and genes (Applied computational genomics group: http://www.medsci.ox.ac.uk/newsletters/april-2014/it-news#section-2 )

  11. Created a model to constuct counter-factual histories of the Spanish Flu Pandemic: http://ww1centenary.oucs.ox.ac.uk/body-and-mind/agent-based-model-the-spanish-flu (Jisc WW1 projects)

  12. Supported Prof. Cindy Skatch in designing a model to think about the consequences of the French government’s Burka Ban (Politics and International Relations)

  13. Supported researcher in building an network formation model (Economics)

  14. Supported researcher interested in modelling the spread of Lyme disease and ticks in UK national parks (Zoology)

  15. Support researcher interested in modelling migration between India and the UK ( International Migration Institute)

Besides the projects we have supported there are many other ABM projects within Oxford including

  1. Institute for New Economic Thinking CRISIS project

  2. The Insect Vector Project in the Zoology Department

  3. FHI-Amlin Research Collaboration on Systemic Risk of Modelling by the Future of Humanity Institute

  4. Several projects led by Felix Reed-Tsochas in the Said Business School

We also co-host a university-wide agent-based modelling get-together each term.

Read more about the ABM Service and ongoing work in the Modelling4All blog. If you are interested in learning more about ABM send an email to researchsupport@it.ox.ac.uk

 

Posted in Uncategorized | Comments Off

ESSA 2014

Modelling4All workshop at ESSA 2014

Modelling4All workshop at ESSA 2014

I attended ESSA 2014 in Barcelona Sep 1-5th. Unlike previous ESSA conferences this brought together social scientists, historians and economists each having their own strand in the 30 minute paper presentations. Whilst I stuck mostly to the social sciences talks, it was interesting to see how other discplines are using and discuss agent-based modelling. It is also heartening to see ABM being used in the Humanities (an area under-represented in our engagements at Oxford).

I earned my bacon by giving a workshop about using the Behaviour Composer as a meta-modelling tool (modelling4all project). I think I got the idea across, and there were a lot of in-depth questions during the practical part of the session. We built the Sugarscape model in homage to the keynote the next day by Joshua Epstein but later found out he’d cancelled due to illness, which was a real disappointment because I really wanted to meet him.

The conference website brings together all the details about the talks and we were given the papers and posters on our memory sticks. Here are the three memories that stand out most from the conference:

  1. First and foremost, this year was the first to feature a serious games strand. The first presentation by Alexander Verbraeck (you’ll have to use google translate) was about their use of various kinds of simulation in business games. Alexander gave some important words of wisdom with regards to how great graphics shouldn’t trump game design, and the simulations that really work are those that are uses creatively within a workshop. What I mostly took from this session was that there are lots of other scientists doing work with ABM.
  2. Hidden amongst the posters (I somehow managed to walk past it several times) was a description of work by Tomas Trescak et al at the University of West Sydney who have recreated a 3D virtual model of the Uruk civilisation. Not only that, they’d given people in the model behaviours that meant they trade, get hungry and move about the city in a realistic manner. But where this most matches work here is that they’ve done this by writing open source plugins to Unity. Actually I found out about this work by bumping into Mike Bithell at Cambridge, who used to be in Geography here and who is also interested in participatory games.
  3. Stefano Picascia has created a GIS based simulation of manchester where the map is populated with historical house price data by postcodes. The aim of the model is to try to predict which areas will regenerate through private capital, and which will need public sector intervention. I liked this paper because it reassures me NetLogo can do GIS modelling, and the model is based on real world data but being used to try to extrapolate forwards. If successful there’s an obvious role for such a model in policy-making.
Posted in Meetings | Comments Off

Report from Constructionism 2014 – Days 3-5

Thursday’s plenary keynotes began with Gary Stager who talked about constructionism and the maker movement.  Great examples at http://sylviashow.com/episodes from a young girl.

Brian Harvey talked about the history of Snap! which he wishes could be called Scratch Senior (since Snap makes Scratch a proper programming language by adding lists and first-class procedures). Note that Mitchel Resnick gave a workshop on Scratch Jr (aimed at 5-7 year olds, iPad only just now, pretty well designed simplified version of Scratch). Brian also gave a workshop on Snap! where we learned that Snap! is well-designed for introductory programming courses at the high school or university level. Brian talked about the compromises he has been making with his principles in helping to design a standardised high school advanced placement course that universities will accept for credit.

Wolfgang Slany presented Pocket Code, a Scratch-like programming system that runs on smart phones and tablets so one can program smart phone apps on smart phones (unlike App Inventor where you author on a PC). He also ran a workshop where we learned how easy it was to make cool apps on your phone.

Jennifer Jacobs gave a nice talk about Dresscode that implements a nice ability to move back and forth between direct manipulation and code. And the code was typically used to generate designs for digital fabrication and physical making.

Carina Girvan talked about how creative teachers could be in Second Life using Scratch for Second Life. Second Life (and OpenSim) seem to continue to have interesting things happening on university run islands. But much of it is private. Many universities did move to OpenSim when Second Life stopped their university discount.

There was a good panel discussion of the concept of powerful ideas.

David Weintrop gave a good demo of BEESMART - A Microworld for Swarming Behaviour and for Learning Complex Systems Concepts. Implemented in NetLogo. David presented his paper on ‘Situating programming abstractions in a constructionist video game’  but I was presenting in a parallel session so missed it. He also presented Arthur Hjorth’s ‘Redesign Your City – A Constructionist Learning Environment on Regional Development’  but I was chairing a different session at the time. David also gave workshop of NetLogo’s network analysis primitives but I had yet another conflict.

I gave a talk about ToonTalk Reborn and got good feedback. Also gave several demos during the conference.

In another session I demonstrated MoPiX - a system for being creative with algebra that I built 6 or 7 years ago.

Pavel Boytchev gave a nice talk about ‘deconstructionism’ — the role that the ability to break things into parts plays in learning and problem solving. Nice demos.

Richard Noss and Celia Hoyles talked about why teaching programming to children. ToonTalk and the Playground Project (1999-2002) that I was part of was a major example.

Videos of the plenaries are available online now.

ken-at-constructionism

Constructionism 2016 will be in Thailand (17 of the 120 conference attendees were from Thailand).

Posted in Uncategorized | Comments Off

Report from Constructionism 2014 – Day 2

Chronis Kynigos talked about constructionist e-books.  Technically the idea is to have e-books that can have embedded in this live constructionist tools. Authoring tools, student tracking and learning analytics are  part of this large EU-funded project called MCSquared.

17 participants are from Thailand and they gave a presentation of constructionism in Thailand in schools (including a special school designed around constructionism), rural communities, and learning in industry. Nice examples of each. Meditation is part of the story — seems to enhance the introspective phase of constructionist learning.

A team from the University of Cyprus presented ‘cognitive processes enacted by learners during co-construction of scientific models’ which led to a very interesting discussion about student-designed models versus the ‘correct’ model. This issue is acute in physics. Physics teachers worry that students building wrong models is counter-productive. Many pointed out how different biology or social science is in this regard. And some argued that even in physics there is a learning trajectory where students build models that are correct only in special situations and then refine and improve them iteratively.

Lots of interesting talks about programming systems for young children.

Posted in Uncategorized | Comments Off

Report from Constructionism 2014 – Day 1

Mitchel Resnick began with a talk about the four P’s of creative learning: projects, peers, passion, and play. He argued that students should be encouraged to choose their own projects that they are passionate about. He is worried that too much emphasis is placed on puzzle solving and less about being creative and expressive. His Scratch project is succeeding exceedingly well in the ‘peer’ dimension (6 million projects uploaded, 1.5 million comments per month). The web site has areas where the more advanced children help others (and are trained to support others instead of simply giving them complete answers). He argued against gamification except where it is peer-to-peer (e.g. ‘likes’ and ‘favourites’). He praised a book call Drive about the short-term benefits and long-term costs of gamification.

Edith Ackermann talked about the role of humour in creativity and argued that the playfulness that is important in learning and creativity (and doing science?)  is more than the process of tinkering — it includes fantasy, whimsy, and humour.

Karen Brennan talked about constructionism in the classroom and how to avoid technocentrism. Many teachers worry that they need to be experts on every aspect of a programming language before they can begin teaching it.

In the panel discussion there was a good deal of concern that the educational establishment is taking constructionist tools (e.g. Scratch or NetLogo) and using them in instructionalist teaching. Though several people thought this was OK so long as there was a balance between instruction and construction.

Posted in Uncategorized | Comments Off

New version of Behaviour Composer

This release updates the documentation, sample models, and fixes a few minor bugs.

Posted in Software development | Comments Off

New releases of Behaviour Composer and BC2NetLogo

To use the latest NetLogo (version 5.1.0) there is a new version of BC2NetLogo.

Minor bug fixes were made to the Behaviour Composer and it was upgraded to use the latest versions of Google App Engine and GWT.

Posted in Software development | Comments Off