Writing

Notes on computational neuroscience and applied machine learning. Older posts (2016) are from my PhD years; new posts will pick up the thread.

• 2026-05-14

  Turning recorded D&D sessions into comics

  A small pipeline that records a session on a Zoom H1 Essential, transcribes the Icelandic audio with ElevenLabs, and renders each session as a graphic novel with gpt-image-2 and a set of reference images that keep the cast looking consistent across pages.

• 2016-05-15

  Leaky integrate and fire neurons (LIF)

  A simple, widely-used membrane-voltage model of biological neurons. We motivate it from Hodgkin and Huxley's experiments, write down the differential equations, and explore the dynamics with an interactive D3 simulation.

• 2016-04-29

  Percolation with inhibition - part 2

  Modelling the percolation-with-inhibition process as a Markov chain on the potential of a single vertex, with an interactive D3 visualisation of the chain, a histogram of stopping times, and a running estimate of the probability of activation.

• 2016-04-25

  Percolation with inhibition - part 1

  Adding inhibitory vertices to bootstrap percolation. An interactive simulation with green and red nodes shows how competing positive and negative messages shape activity in the graph.

• 2016-04-08

  Asynchronous bootstrap percolation

  Bootstrap percolation without synchronous rounds — when edge delays are random the activity percolates much faster, in constant expected time.

• 2016-03-27

  Bootstrap percolation

  A graph-based process that models the spread of activity in a network, with an interactive vis.js simulation on the Erdős–Rényi random graph.