Introduction
People
Notes
- 2007-04-15
- Introductions, objectives of the workshop, logistics
- Mike Goodchild
- Sessions I and II are swapped.
- Session II: Modeling Complex Spatial Systems
- Mike Worboys
- Snap and Span
- Discrete Object States and Continuous Events
- States and Actions
- Sensors responding to dynamic fields
- Process Algebra
- A general form of Cellular Automata (CA)
- Richer more formalised than some CA
- Useful list of desired characteristics of models
-
- Plenary discussion
- Discussant: Nigel Gilbert, University of Surrey
- Session I: Representation in and of Complex Spatial Systems
- David Bennett, University of Iowa
- Discussion
- Discussant: Georgios Theodoropoulos, University of Birmingham
- May Yuan, University of Oklahoma
- Demonstrations and posters
- 2007-04-16
- Session III: Validation and Verification of ABMs
- Validation, Verification, Calibration and Confirmation
- David O’Sullivan, University of Auckland
- Complex systems are irreducible
- We can generally talk about prediciton but not specifically, e.g. Where is the next road accident?
- Verification is impossible is argued by Oreskes (1994) Verification, validation, and Calibration of Numerical Models in the Earth Sciences. Science 253 641-6
- There are many models that can produce the same results and these cannot easily be recognsed as being equivallent. This is known as a problem of equifinality.
- The arguement here is for multiple models :)
- Calibration and confirmation.
- I think I agree with Oreskes take on this as does David.
- Much of what we learn about a system in modelling it is in the exploration and building models.
- Recommmended read: Bevan (2002) (1993)
- Steve Bankes 2002, 1993 Exploratory models operational research
- Oh good David realises that he can have a space-time-attribute hypersphere rather than hypercubes :)
- Are you a roundy or a squarey?
- We need a way of capturing exploration:
- The key to this is provenance and archives.
- Interesting diagram of scientific reasoning and modelling.
- Discussion
- When does a simple model become unable to represent a set of observation. When does it's explanatory power need to be enhanced?
- Dawn Parker, George Mason University
- Complementary goals:
- Building theoretical frameworks and integrated models
- Building specific models for specific purposes
- (I may have not got these right)
- Sensible focus on the scientific method.
- Recommended reading Grimm et al 2005 What patterns are you interested in?
- Emergence Organisation in to recognisable macro scale patterns
- Interesting use of quilt type patterns to show emergent patterns.
- A model is not validated or not, it is validated to a degree under a specific condition or set of conditions. Verburg.
- Discussion
- Parameter free ABMs
- ABMs as an integrative tool
- Plenary Discussion on Validation and verification
- Discussant: Mike Batty
- The space of plausible models.
- In 1960 there used to be a tight link from theory through to application.
- Parsimony - as simple as possible to fit the observed data.
- In the 1970s Wilson started exploring "the space" of multiple entropy or utility maximising models.
- A model built on a system should not be validated with that system.
- Have we got validation and verfication the right way round, is one logic and the other math?
- It is confusing terminology as people are mixing up validation and verification and use the terms interchangeably.
- Maybe we need to agree on a terminology.
- What are model assumptions?
- Use Cases and purpose (Is it fit for purpose?)
- Break-out group discussion
- Mike Goodchild
- We are to split into 3 groups and report back to appreciate the similarity/difference in a discussion of the following:
- Important topics
- Spanning application domains
- Within the domain of GIScience
- Contributions to research infrastructure
- cyberinfrastructure
- e-Science infrastructure
- tools data etc
- What other mechanisms might enhance the field?
- Can we reach a consensus on the methodological questions?
- Epistemological Ontological
- Is space special?
- Validation and verification.
- The board room group
- People:
- Li An
- Richard Aspinall
- David Bennet
- Matthew Collier
- Tom Cova
- Andy Evans
- Nigel Gilbert
- Kathleen Hornsby
- Naicong Li
- George Malanson
- Richard Milton
- Donna Peuquet
- Victor Schinazi
- Andy Turner
- Mike Worboys
- Important ABM and Complex System Modelling (Research) Topics
- Richard Aspinall: Space time data representation.
- Mike: Distributed systems applied in a real world context.
- How different are agent based models and other models?
- Significance of results.
- Understanding results of an ABM.
- Developing a great application.
- Developing better analytical tools to analyse changes in the model.
- Andy Turner:
- The computational and data challenges of large (global) scale models that operate at an individual actor/agent level and with a high level of spatial and temporal resolution.
- Measuring uncertainty. Not just sensitivities.
- Can we measure the complexity or expressive power of a model? Can we show that two different models are as powerful?
- Provenance and archiving: ensuring results replicability/reproduceability.
- I articulated this one to the group as it seemed we could have one each and I chose this one to make Edoardo happy.
- A registry of exogenous factors.
- What is my simple model not including that might help better explain the behaviour observed?
- Can I pick one from a list?
- Using the Semantic Web as a modelling tool...
- How do we collaborate?
- Linking information to form a community of those interested in Agent Based Modelling of Complex Spatial Systems (ABMofCSS).
- Should we drop "Agent Based" and consider ourselves a community intersted in "Complex Spatial Systems Modelling"?
- Nigel: Methods and tools for exploring model "spaces".
- Kathleen: Boundaries on the limits of predictability of complex systems - computability and unsolvability.
- Andy: Community negotiation of truth, belief.
- Donna: Note the differences or conflicts in meaning and belief.
- What contributions/developments to research infrastructure might help?
- Discussion and thoughts:
- Discussion and thought of important factors is relevant.
- Development of a community for research can be supported by various means
- Can we model our community?
- Andy Turner:
- What standards are important?
- High Level Architecture (HLA)
- Repository for model frameworks and models and specific simulation.
- Here is a model, based on these software, here are the results if you do this.
- How can we register our models and simulation runs in a way that makes them easily accessible.
- Corpus of work.
- What other mechanisms might enhance the field?
- Conference sessions and workshops on:
- Agent Based Modelling of COmplex Spatial Systems
- Sessions on complex social spatial systems?
-
- Session on land use change and agent simulation.
- Structure a set of workshops to address specific research topics as outlined above.
- Can we reach a consensus on the methodological questions?
- Epistemological Ontological
- Is space special?
- Validation and verification.
- Andy Evans mentioned some work presented at GeoComputation a couple of years back by a computer games programmer come phd student who was modelling taxi's.
- Let us map ourselves individually onto the list of things we generate to show:
- Which ones are we interested in...
- What are we working on...
- What ideas we have...
- Feedback session
- Richard Aspinall presented from Group 1 :)
- Steve Manson presented from Group 2 :)
- David Maguire presented from Group 3 :)
- Plenary Discussion
- Nigel is part of a consortium that has been funded by the European Commission to build a virtual world
- Catherine Dibble
- Mike Goodchild
- Null Hypotheses
- Wants a library of buttons which give a possibility to run simulations of the world under different conditions.
- Generated randomness replaced by models.
- Someone to rewrite Harvey 1969 to show how science is done today.
- Close
- Great workshop, thanks everyone!
Blog
- 2006-04-16
- Some attempts at definition
- What are Agents?
- Objects with state and ability to change state based on the state of other agents and its own state.
- Typology/taxonomy of agents
- In an Agent Based Model (ABM), the environment is the collection of all agents and a space time frame in which agents interact.
- On terminology:
- Complex and complicated
- Process and service
- What was being described as a process yesterday, others might call a service.
- 2006-04-15
- Added and linked in a NCeSS Wiki : ABMofComplexSpatialSystemsPeople page
- Talked with Mike Worboys to say that I thought the abstract model he presented was a general model and thus potentially very powerful. He seemed pleased with that feedback.
- Talked with Nigel Gilbert about using models to undertake social shaping by government and policy.
- Chatted with various folk about a perceived housing crisis in affordable housing in the UK and what might be done about it.
- 2006-04-13
- Read through the available Bio and Position Papers of participants
- Began searching for and compiling a list of participant Web Home Pages and summarising their interests.
- I want to generate some tag clouds...
References
Agenda
People
Alan Glennon's Blog Workshop Notes
Conferences to target:
- COSIT
- World Congress on Social Simulation at George Mason organised by Dawn Parker et al.
- GIScience, 2008
- GeoComputation, 2007
- 3rd International Conference on e-Social Science, 2007
![[Validate RDF]](http://www.geog.leeds.ac.uk/people/a.turner/images/rdf_w3c_button.gif "W3C RDF Validation")
![[Validate CSS]](http://www.geog.leeds.ac.uk/people/a.turner/images/vcss.gif "W3C CSS Validation")
![[An image of Andy Turner]](http://www.geog.leeds.ac.uk/people/a.turner/a.turner.png)