/generators

Early work on the AJC and coherence (2019/2020) also with weather data

Primary LanguageJupyter Notebook

Generators

Experiments around the EAMC and space-time committors / coherence. The state of this repository is pretty messy and is left here for archival use. The Augmented jump chain code has been rewritten for python and julia in cmdtools and birthdeath

Notable features: Coherence of meteorology data, PCCA+, Bickley Jet

Contents

Source

  • Galerkin discretization of the Augmented Jump Chain (called Embedded augmented markov chain (EAMC) during that time): src/galerkin.jl
  • EAMC reconstruction of the Koopman operator by space-time committor: src/committor.jl
  • EAMC reconstruction of the Perron-Frobenius operator by jump activity: src/perronfrobenius.jl
  • Gillespie like simulation of trajectories of the EAMC: src/gillespie.jl
  • Opimization of space-time committors: src/main.jl
  • PCCA+ implementation: src/pccap.jl
  • Some interactive (?) Makie visualization: src/plot.jl

(Nonautonomous) Models

  • 4 Cell barrier on-off switching: barrierswitch.jl
  • The bickley jet: bickleyjet.jl
  • Fluid flow governed by vectorfield: continuity.jl
  • Overdamped Langevin via SQRA, also shifting and rising doublewells: langevin.jl
  • 1d sqra: mysqra.jl

Notebooks

  • example*: EAMC plots of different systems
  • bickley*: bickley coherence, plots not saved
  • weather-inspect: look at the weather data
  • weather-kmeans/pcca: clustering of the eamc for coherent sets
  • weather-optimize: clustering via optimization criterion