Home

The Most Probable Path Algorithm Package

The most probable path (MPP) algorithm helps to reduce the number of a state trajectory. Based on the dynamic relationship between the states, the least stable state is merged with its most similar state (e.g. to which the transition probability is highest) until all states have been combined. This so-called lumping tree is parsed in a second step and divided into macrostates satisfying a user-defined minimum metastability and population. In contrast to other clustering algorithms like generalised Perron cluster-cluster analysis, the number of macrostates is not provided as a parameter but arises from the boundary conditions (requested metastability and population) and the lumping tree. This often allows for a more detailed state partitioning while the states are still distinct.

Installation

python -m pip install mpp-lumping