This repository contains notes and code relating to spectral- (i.e. frequency) domain models of macro-scale neural dynamics.
The primary focus of this type of neurophysiological model is to reproduce two key features of the M/EEG power spectrum:
i) Spectral peaks (alpha, theta, etc.); number, location, magnitude
ii) Power law scaling exponents
We are interested in three things:
a) understanding - derivation of, motivation for, and behaviour of various spectral-domain neural models
b) simulating - using existing, modified, and novel models
c) fitting to empirical M/EEG power spectra, and assessing alternative optimization techniques (scipy.optimize
, scikit-optimize
, tensorflow
, STAN
, etc.)
notes/
- Technical descriptions and general reflections on models, data, and science
code/
- The beating heart. Core functions (mostly python code) for simulating power spectra and data fitting
data/
- Empirical power spectrum recordings from various sources
scratch/
- Miscellaneous and work-in-progress. Unapologetically messy
Check these out:
Interactive widget and model fitting for the Abeysuriya-Robinson model (live binder notebook)