Short presentation

You are currently viewing a placeholder content from Vimeo. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.

More Information
See a short presentation of the module by the lecturer (webinar recording).

Module
Information

Assoc. Prof. Dr. Gudrun Wallentin

UNIGIS Program Director

Gudrun Wallentin has her main research focus on spatial simulation modelling, including but not limited to the ecological domain. Gudrun graduated from the University of Innsbruck with a diploma in Ecology. After working several years as the executive director of the Nature Park ‘Zillertaler Alpen’ in Austria, she continued her academic studies at the University of Edinburgh, from where she holds a MSc in GIS. In her PhD at the GIScience Institute of the Austrian Academy of Sciences in Salzburg, she applied methods of GIScience and spatial simulation modelling to the domain of ecology. Since 2011 she works at the Department of Geoinformatics – Z_GIS at the University of Salzburg. Gudrun Wallentin is program director of UNIGIS Salzburg.

Upon successful completion of this module, you will be able to design your own simulation models and implement them with AI assistance while staying in full control of the code. You will also be able to analyse and validate them to solve applied problems and conduct research with simulation models.problems and conduct research with simulation models.

Basic competences in Geoinformatics (UNIGIS modules 1 – 3) are required. Additionally, you should have successfully completed the UNIGS module “Basics of Application Development” or have equivalent competences. See also the information under “software”. If you are not sure whether you qualify, please contact gudrun.wallentin@plus.ac.at.

The module provides a broad overview of existing theories and approaches for spatially explicit simulation modelling. The focus is on hands-on experience with designing and coding of geosimulation models, which is a spatially explicit modelling approach that integrates agent-based modelling and cellular automata.

The software that will be used for model development is GAMA, which stands for GIS and Agent-based Modelling Architecture. It is an open-source programming framework that was specifically developed for the design and the implementation of spatially explicit simulation models. GAMA models are coded with GAML, an object-oriented programming language that was specifically developed for the purpose of agent-based modelling.
You will work with AI-assisted coding. A subscription to a pro-version of an AI platform that supports agentic AI therefore is highly recommended. The tutorials are based on Claude Pro from Anthropic.

Lecturer assessment reflects student’s achievements in this module and is conducted through assessing module assignments. Assignments (English language!) must be submitted in written format (.PDF) within the required time period. Exercises are designed to enforce students’ knowledge and skills. These should be completed to allow students to assess their own progress and are not included in the module assessment.

Duration

3 months

Language

English

Value

6 ECTS

Costs

The module is free of charge for UNIGIS students working to meet their elective subject requirements. Included are:

  • all related study materials
  • supervision and assessment
  • module accreditation according to the curriculum

ClubUNIGIS members can register at a price of € 420,-.

Included are:

  • all related study materials
  • supervision and assessment
  • course certificate upon completion of the module

Lessons
of the Module

1

Lesson 1 – Why simulate?

What is a model, and which problems call for simulation and which do not? An introduction to simulation modelling, including your first “Hello World” model in GAMA.

2

Lesson 2 – Systems thinking and complexity

From System Dynamics with its stocks, flows and feedbacks to complexity theory: why a system is more than the sum of its parts, and how emergence and self-organisation arise.

3

Lesson 3 – Choosing a modelling approach

Spatial System Dynamics, partial differential equations, probability-based models or geosimulation? Five guiding questions help you match the right approach to your spatio-temporal problem.

4

Lesson 4 – Geosimulation

In this lesson you build your first geosimulation model with geographic data, combine agents with a cellular automaton, and learn when entities should update synchronously or asynchronously.

5

Lesson 5 – AI-assisted coding

Generative AI as a coding partner, with the human in the lead. You set up a coding agent with git, learn to prompt effectively, and keep the model design in your own hands.

6

Lesson 6 – Code verification

Did I build the thing correctly? You learn systematic ways to find syntax, runtime and semantic errors, and build observation dashboards that expose the subtle errors typical of AI-generated code.

7

Lesson 7 – Cellular Automata

Cells, neighbourhoods and transition rules, from Conway’s Game of Life to deterministic, stochastic, constrained and probabilistic CAs of real geographic space.

8

Lesson 8 – Dynamic environments

Urban sprawl, wildfires, invasive species, erosion: you work with building-block models for the local processes behind landscape change, including growth, spread, diffusion, segregation and movement.

9

Lesson 9 – Model design

You design your model before coding it: problem definition, scales, boundaries and process schedules, formalised in a UML activity diagram and documented with the ODD protocol.

10

Lesson 10 – Agent-based models

Agents that sense their environment, query their inner state, decide and act. You learn how bounded rationality and the way agents perceive space shape the patterns that emerge.

11

Lesson 11 – Movement

From random walks and Lévy flights to purposeful movement through dynamic landscapes and social movement such as flocking. You also see how agents interact with their environment and with each other.

12

Lesson 12 – Parameterisation

You learn to assign, calibrate and test input parameters, and to quantify model uncertainty with sensitivity analysis and Monte Carlo batch experiments, while avoiding the trap of overfitting.

13

Lesson 13 – Validation

Back from the model world to the real world: you meet levels and strategies of validation, up to pattern-oriented modelling, to judge whether a model is fit for its purpose.

14

Lesson 14 – Scenarios: what if…?

Putting the model to use: How many cows can an alpine pasture sustain? How does street infrastructure impact bicycle safety? A validated model becomes your virtual laboratory for comparing alternative options, both to support decisions and to answer research questions that no real-world experiment could.