Code & Data
Our software packages and datasets are archived with a citable DOI, so results can be reproduced against the exact version they were produced with.
andreanovoa
echostatenetwork
Echo state networks in pure numpy: ridge-regression training with chaotic recycle validation, parametric inputs, Bayesian hyperparameter search, closed-loop prediction, and Jacobians for data assimilation, with a parallel layout of one reservoir per patch for spatially extended systems. The reservoir core behind qlroms and romda. Includes a live demo that builds a network from your choices.
ntsa
Nonlinear time-series analysis of dynamical systems from a single long trajectory: Lyapunov exponents, delay-embedding analysis, and classification of the dynamical regime (fixed point, limit cycle, quasiperiodicity, chaos), with diagnostic figures for each case. Works with any model following the dynamodels protocol.
romda
Real-time reduced-order modelling and bias-aware data assimilation: ensemble Kalman filters including the regularized bias-aware EnKF, physical and data-driven forecast models (ESN, POD-ESN), bias estimators, and POD/SPOD decompositions, with tutorials building up to full real-data digital twins.
qlroms
Quantized-local reduced-order models: the state space is partitioned into charts by k-means, each chart gets its own local POD basis, and an exact shared atlas stitches them into one global coordinate system, with the local dynamics advanced by an intrusive Galerkin projection. Includes a live KS-1D demo.
dynamodels
Dynamical-system models with history tracking and pluggable integrators. Currently includes Van der Pol, Rijke, annular thermoacoustics, Kuramoto–Sivashinsky, Lorenz 63 and Lorenz 96.
Data
A real-time digital twin of azimuthal thermoacoustic instabilities
Acoustic pressure data from an annular combustor, recorded by four microphones at 0°, 60°, 120° and 240°, for four equivalence ratios between 0.4875 and 0.575 at a constant power of 6 kW. Originally created by Indlekofer et al. (2022); also used by the tutorials in real-time-bias-aware-DA.
