Summary:
The eigensolver does not fulfil the tests for sparse and dense RDL decomposition (stationary distribution from eigenvector as well). The abbreviation is very large for the sparse code (10 orders of magnitude).
This is actually a minor issue, since NumPy 1.9 is fairly outdated and not supported by either Omnia nor Conda-forge. However people compiling the software them self could potentially run into this. So this issue is for the record only, as we can not fix it our selfs.
https://travis-ci.org/markovmodel/msmtools/jobs/254465976
Summary: The eigensolver does not fulfil the tests for sparse and dense RDL decomposition (stationary distribution from eigenvector as well). The abbreviation is very large for the sparse code (10 orders of magnitude).
This is actually a minor issue, since NumPy 1.9 is fairly outdated and not supported by either Omnia nor Conda-forge. However people compiling the software them self could potentially run into this. So this issue is for the record only, as we can not fix it our selfs.