Research
Most of this is about cellular automata and their cousins on networks: fully discrete systems where a handful of local rules produce behaviour that ranges from dead to chaotic, and where the interesting question is usually how to tell those regimes apart before you run the thing. Earlier detours went through epidemic modelling during the pandemic and gravitational-wave data analysis before that.
Published
- A comprehensive taxonomy of cellular automata
Five families of cellular automata defined in one place, with a map of which exotic-sounding variants are the same thing wearing a different hat.
- Convolutional neural networks for automated cellular automaton classification
Earlier networks that sorted cellular automata by behaviour were quietly cheating by learning the local rule. Ours looks at the actual pattern and still gets it right nearly always.
- Essential metrics for Life on graphs
The Game of Life generalised to arbitrary networks. A few simple metrics predict how it behaves, well enough to synchronise a firing squad nine times out of ten.
- Efficient simulation of non-uniform cellular automata with a convolutional neural network
Non-uniform cellular automata are secretly convolutional neural networks, so we ran them in TensorFlow and simulated many large automata at once, fast.
- Mobility and the spatial spread of SARS-CoV-2 in Belgium
Early in the pandemic Belgians moved less as cases rose, and how connected your district was to the first outbreaks predicted its excess deaths six weeks later.
- A stochastic mobility-driven spatially explicit SEIQRD COVID-19 model with VOCs, seasonality, and vaccines
A province-level COVID model with variants, vaccines and phone-tracked mobility. Travel between provinces barely mattered; cutting social contact beat restricting movement.
- Assessing the effects of non-pharmaceutical interventions on SARS-CoV-2 transmission in Belgium by means of an extended SEIQRD model and public mobility data
Mobility data plus hospital records showed that under lockdown, homes and schools drove transmission, and reopening schools could push R back above one.
- A morphology-independent data analysis method for detecting and characterizing gravitational wave echoes
Without assuming what an echo looks like, a wavelet-based search can still pick gravitational-wave echoes out of detector noise and measure their spacing and decay.
Under review
- Cellular Automata: 17th International Conference on Cellular Automata for Research and Industry, ACRI 2026, Ghent, Proceedings
- Exact Lyapunov spectra of affine cellular automata and the parity rule on networks
For a linear rule you can write the whole chaos spectrum down by hand. Code to reproduce every figure is public.
- The MACCA-GW groundwater flow model: application to De Wilde Landen, the Netherlands
- Probing dynamical importance of network nodes under maximal agnosticism
Which node matters most when you know nothing about the dynamics except that they are local and deterministic?
- Shape irregularity of Life-Like Network Automaton rules as an indicator of classification performance
In progress
- Machine learning loses to simulation: rule reconstruction as the benchmark for reading cellular-automaton damage response
A negative result: the learned model never beat just running the thing.
- Exhaustive enumeration of symmetric still-life tiles in Life-like cellular automata via orbit-reduced SAT solving
The still lifes behind the mosaics on this site, counted properly.
Talks
| Year | Where | Title |
|---|---|---|
| 2026 | ACRI & AUTOMATA, Ghent (co-organiser, proceedings co-editor) | The MACCA-GW groundwater flow model |
| 2025 | Conference on Complex Systems, Siena | Impact maximisation: a network automaton perspective |
| 2025 | AUTOMATA, Lille | Small changes with maximal impact |
| 2025 | CompleNet, Fortaleza | Small changes with maximal impact |
| 2024 | ACRI, Florence | Bridging the gap between cellular automata and deep learning |
| 2024 | Conference on Complex Systems, Exeter | Classifying cellular automata: a deep learning approach |
| 2023 | Summer Solstice, Florence | Pattern recognition for noisy 1D cellular automata with CNNs |