This release adds support for NetLogo extensions (e.g. the Geographic Information System extension) and the ability to add NetLogo declarations. A few bugs fixed and small improvements added as well.
Today we gave a lunchtime seminar as part of the IT Learning Programme’s ‘make’ series. These lunchtime talks are informal and stress creativity and exploration. We presented agent-based modelling and Modelling4All as a way to ‘play god’ in that one creates a universe, populates it, and then the computer brings it to ‘life’. We showed and discussed lots of sample models. I demonstrated how one can build Schelling’s Segregation model in the Behaviour Composer (using only generic micro-behaviours!). A nice discussion followed. Course description is here: http://courses.it.ox.ac.uk/detail/TM13J
Most of the changes make it easier to install and use the new Behaviour Composer to NetLogo tool. The release includes significant improvements to the ability to make some edits (including moving, resizing, and customising NetLogo interface elements and the main view) and have them automatically reflected back to the Behaviour Composer. This release supports a way to switch to another model via URLs. Also experiments created using the BehaviorSpace tool are now stored in the Google cloud so that they are available if you resume a session on another computer. Improved the layout of many sample models using these new functionalities.
Howard and I were invited to give a seminar about Modelling4All and to meet academics in York. The visit was organised by John Forrester
The morning consisted of interesting discussions about how Modelling4All might fit with a project interested in science games and how in the context of artisanal fishing in Kenya how to integrate public policy choices (e.g. what regulations to impose) and the local day-to-day fishing decisions of individual fishers. And how might one integrate a large-scale aggregate ecological model of fish stocks with an agent-based model of fishers. (Not easy)
In the afternoon we gave a talk about Modelling4All to a large audience. Slides at tinyurl.com/m4a-slides. Ended with some interesting questions about what Modelling4All says about how to do experiments with the models produced. (Our answer is to rely upon the NetLogo community for that with tools like BehaviorSpace and BehaviorSearch.org.) Another question was whether there was a danger that policy makers would misuse the Behaviour Composer by building poor quality models and justifying decisions based upon them. (Not so different from the current situation. But models should be open and transparent and then the quality can be judged by ‘all’.)
…and I think field trip is definitely the right word to use.
I think we achieved what we set out to do:
- Collaborated with a computer science group that specialise in GIS, ABM and computer vision algorithms in Yaounde 1 University
- Visited the rector at the University of Dchang to thank him for my invitation to Cameroon and explain what we’re up to.
- En-route to Somie we met the Herman the catholic priest (who has been struck by lightening three times) and took pictures of the very fertile Oku valley
- Piloted and completely re-wrote the farming model several times, amassing gigabytes of audio, video and XML log data in the process
- (On my day off I spent 10 hours trying to find a troop of chimpanzees that live in the surrounding forest).
In Somie we ran three modelling sessions, and the final one was the most formal where we had 8 farmers (all men 20-30 years). Each farmer was trained for 1 year (in the simulation) by a fellow farmer (who we paid) and then they had to run the next 9 years by themselves. We told the farmers the aim was simply to survive the ten years and make money. Downloads: (a) the versionof the Somie model we used on the last day, (b) first draft of the ODD (coming soon!)
The next set of tasks are:
- Find a tool to analyse the XML log files e.g. xml.log.toolkit, the XML package in R (- with tutorial)
- Try to use the Digital Replay System maintained by Nottingham University to piece together all the audio, video and log files into a coherant view
- Look at what is involved in programming LinCity i.e. pick one of its numerous forks such as LinCity NG
So much more to write but unfortunately other unrelated jobs are catching up with me. To be continued…
Today we had 9 participants (only 3 from Oxford) in our full-day workshop on how to teach ABM (with a focus on the life sciences). We made heavy use of the Behaviour Composer and the new tool for connecting directly with NetLogo. Here are the slides. I think it went very well.
Perhaps some of the participants will try using the Behavior Composer in their teaching.
Gave a talk to the Computer Science Department at the National University of Singapore. Title was
Agent-based modelling in education, public engagement, policy making, discussions, art, entertainment,
decision making and research
Didn’t talk much about decision making and research. Slides are here but much of the talk relied upon interactive demos of NetLogo and Modelling4All.
Got some good questions about the concurrency model of the Behaviour Composer and about validating agent-based models.
I was invited to lead a discussion at the University Scholars Program at the National University of Singapore. Here’s the poster description:
Title: I simulate, therefore I am not
Can we simulate the human brain, maybe after spending 1 billion Euro over the next 10 years, as pursued by the Human Brain Project? Will it
be able to speak, and if so, what will it tell us? Can we simulate Earth’s climate, maybe even the Earth itself? If civilizations are likely to develop realistic simulations of universes, then—so argues Nick Bostrom—we are likely not real, but living inside a simulation ourselves. Ken Kahn will take a closer look at Bostrom’s simulation argument, and at the value of simulation today and in our (simulated?) future.
We had a lively two-hour discussion. First we talked about how climate models work and what they good for (and what the controversies are). Next we talked about the new ten-year one billion Euro project to simulate the human brain - http://www.humanbrainproject.eu/. This led to discussions of substrate-independent intelligence, uploading, moral status of simulated brains and more.
Then I summarised the ‘simulation argument’ of Nick Bostrom (Oxford philosophy professor) that we are likely to be simulated: http://www.simulation-argument.com/. This was followed by a discussion of John Barrow’s ideas about how physicists might detect that we are inside a simulation: http://www.slideshare.net/UnitB166ER/living-in-a-simulated-universe-by-john-d-barrow. Finally we discussed Robin Hanson’s essay about reasons that we should behave differently if we think there is a good chance we are being simulated. http://hanson.gmu.edu/lifeinsim.html
The discussion was a good fun and a good advert for my computer science talk about agent-based modelling and Modelling4All on Thursday.
A simple set of assumptions I can make about the market and selling:
- Procedure run whenever the vendre button is pressed. The code will use the quantite NL slider to read how much of a specific crop is to be sold – selected using the a-vendre NL chooser
- Look-up market value data (or derive using an equation) for specific crop at this period during the year (this will be done simply now, but will be more relevant with the larger model where there are many different families farming. the main thing to note is that the earlier a crop is brought to market, the higher the price (within a range obviously). (It is riskier to plant early for reasons related to the weather).
- The procedure will report how much money has been earned, update the household account and reduce the amount of the crop in the reserve set of monitors. I could also warn farmers that their household account is getting low and they should consider selling something.
- Add the idea of a household-need threshold for each crop i.e. how much the family needs to keep for their own nutritional needs. this is a fairly major assumption though – it could be true that farmers can buy everything they need i.e. sell at a high price and buy at low prices as needed. this might be especially true for the richer farmers, or farmers that don’t have storage. (will update data sheet now with respect to storage).
The Behaviour Composer direct to NetLogo tool has been enhanced with new scripts to develop models in 3D NetLogo. Also the tool now stores experiments created in the BehaviorSpace tool with a model so that experiments are not lost as the model is updated. Also updated the tool to use the latest version of NetLogo (5.0.4).
Links and documentation at http://resources.modelling4all.org/Home/behaviour-composer-direct-to-netlogo-guide have been updated.