/cmdtools

Python implementation of severals tools (PCCA, AJC, SQRA, P/Q estimation) for the analysis of dynamical systems from the transfer operator perspective.

Primary LanguagePython

Binder Documentation DOI

cmdtools

This Python library implements a suite of tools used and/or developed in the Computational Molecular Design group of the Zuse Institute Berlin.

Installation

Install with pip install cmdtools (If you want to use the SLEPc library for sparse Schur decompositions install cmdtools with the extra slepc, i.e. pip install "cmdtools[slepc]")

Note: It seems that pip has problems installing SLEPc. However, using conda you can install it with conda install -c conda-forge petsc slepc petsc4py slepc4py.

Contents

  • pcca: An implementation of (generalized) PCCA⁺ using the Schur decomposition
  • ajc: A sparse implementation of the augmented jump chain
  • diffusionmaps: Diffusionmaps with sparse support and out of sample extensions
  • galerkin: Trajectory based estimation of the transfer operator using a Galerkin projection onto Gaussian RBFs
  • gillespie: Trajectory simulation from a generator
  • newton_generator: Multi-step estimation of the generator via the Newton polynomial
  • picking_algorithm: Given a set of datapoints, pick n points such that they are distributed as evenly / equidistant as possible
  • sqra: The Square Root approximation, estimating the generator for the diffusion in a given potential
  • voronoi: Voronoi clustering of trajectories and estimation of the transfer operator with different metrics and center strategies.
  • diffusion: A collection of dynamical systems (So far the double- and triple-well)

If you use cmdtools in your research, please cite our work.