Markov Modeling¶
Scripts to estimate, validate and analyze Markov models. Most of these scripts are based on PyEMMA.
For constructing Markov models, the intended workflow is as follows (see also the supplementary information of Prinz et al., JCP, 2011, 134, 174105):
Simulation: Simulate your system with equilibrium MD simulation.
Discretization: Discretize the MD trajectory (see Discretization for scripts that discretize MD trajectories).
Implied timescales: Calculate the implied timescales of the discretized trajectory as function of lag time with
scripts.markov_modeling.pyemma_its. Choose the lag time at which the timescales of interest (usually the few largest timescales) converge as lag time for the estimation of the Markov model. If the lag time at which these timescales converge is not significantly smaller than the timescales themselves, repeat step 2 with another discretization.Markov model: Estimate the Markov model with
scripts.markov_modeling.pyemma_mm.Chapman-Kolmogorow test: Check the Markovianity of the estimated model by conducting a Chapman-Kolmogorow test with
scripts.markov_modeling.pyemma_cktest.Analysis: Analyze the Markov model.
Todo
Change how the scripts read files containing the bin edges used for
discretization. (Background: The output of discrete_pos was
changed and all the Markov modeling scripts were initially build
upon the output of discrete_pos)