Python package that implements adaptive sampling strategies to improve the quality of Markov State Models.
This collection of scripts is useful for implementing an adaptive sampling protocol to build optimized Markov models, using intelligently selected data parts, with different criteria. Among the criteria analyzed are:
-
Start from the state with the lowest number of counts
-
Start from the state that causes the best update of the transition matrix, based on a metric defined upstream.
The expected result is accelerated convergence of the model created through adaptive sampling, compared to that created with random data selection.
These scripts are created specifically for the format used to save our discrete trajectories. They were not designed with a generalizing approach.