Support for eigenvalue uncertainty based adaptive sampling

[hnadeem2]

Hello!

I was looking to implement the adaptive sampling technique from here. The idea is to start sampling from states which maximally reduce the variance in eigenvalue of the transition matrix.
The computation involves calculation of the sensitivity vectors among other quantities. I wanted to ask if we can use Bayesian MSMs in deeptime to achieve this computation, or if there are other methods that can be helpful for this.

Thanks,
Hassan

[clonker]

Hi Hassan, I am not familiar with the method but you may be able to use this:
https://deeptime-ml.github.io/latest/api/generated/deeptime.markov.tools.analysis.eigenvector_sensitivity.html#deeptime.markov.tools.analysis.eigenvector_sensitivity
Other than that you can of course use bayesian sampling to estimate the variance around quantities derived from a msm. But I feel like the analytic method above is - if applicable - probably preferable.
Cheers, Moritz.