Acknowledgments

We thank Marcus Weber and the Computational Molecular Design (CMD) group at the Zuse Institute Berlin (ZIB) for the longstanding and productive collaboration in the field of Markov modeling of non-reversible molecular dynamics. M. Weber, together with Susanna Röblitz and K. Fackeldey, had the original idea to employ Schur vectors instead of eigenvectors in the coarse-graining of non-reversible transition matrices. Further, we would like to thank Fabian Paul for valuable discussions regarding the sorting of Schur vectors and his effort to translate the original sorting routine for real Schur forms, SRSchur published by Jan Brandts, from MATLAB into Python code, M. Weber and Alexander Sikorski for pointing us to SLEPc for sorted partial Schur decompositions, and A. Sikorski for supplying us with an code example and guidance how to interface SLEPc in Python. The development of pyGPCCA started - based on the original GPCCA program written in MATLAB - at the beginning of 2020 in a fork of MSMTools, since it was planned to integrate GPCCA into MSMTools at this time. Due to this, some similarities in structure and code (indicated were evident) can be found. Further, utility functions found in pygpcca/utils/_utils.py originate from MSMTools.