Project overview

Msm class code

The main class for the Markov state model (MSM) is located in src/msm_playground/msm.py. This class manages creation and operation of the MSM. An MSM can be created based either on a trajectory with clustering algorithm or a custom_labels integer-valued trajectory describing in which state a system is at each time step. There are also alternative ways to create an empty MSM and set correlation (transition) matrix manually, please refer to the tests/test_msm.py for examples. A created MSM can be used to compute various properties, such as:

• transition probabilities
• committor functions and corresponding linear equations for diagnostic purposes
• mean first passage times (MFPT) and corresponding linear equations for diagnostic purposes
• stationary distribution

Diagnostic scripts

double well system

• diagnostics/double_well/

triple well system

• diagnostics/triple_well/

Müller-Brown system

• diagnostics/mueller_brown/mb_diagn_pipeline_committor.py
• diagnostics/mueller_brown/mb_diagn_pipeline_mfpt.py

Plot registry

Plot Description	Script
Comparing naive vs stopped process committors on triple well system	diagnostics/triple_well/1D_triple_well_diagnostics.py
Committor approximation error on Müller-Brown system	diagnostics/committor_error_new_states.py
MFPT approximation error on Müller-Brown system	diagnostics/mfpt_error_new_states.py