Microstate label and connected set discussion

[franknoe]

I've just received a comment from a PyEMMA user:

In this work, we assigned an additional label to each microstate in the unused digit position of the state labels, which is cumbersome and also confused functions like active_state_fraction. We think some internal mechanism to assign additional labels for the microstates could be helpful in some use cases, though it might not have general appeals.

I think we should reconsider the use of labels. We have discussed this recently in the context of connected sets. Can we have some discussion here how to sort this out / do you already have a prototype or draft how to handle connectivity?

[marscher]

If we provide a numpy string array with the same shape as the full state space, we can perform index tricks by passing the desired label to a "select_subset" method, which then sets the state space for subsequent evaluations.

The projection to the full state space is already provided by the "Subset" mixin, but in general we can also put the labeling functionality there (when we have defined what we really want here).

[cwehmeyer]

We could provide two simple methods to

1. prune a user-provided set from the full state space (remove states not in the active set), and
2. map a user-provided set from the full state space onto their internal states.

@franknoe do you know why the cited user wants to assign labels?

[franknoe]

no

Am 14/07/16 um 12:13 schrieb Christoph Wehmeyer:

We could provide two simple methods to

1. prune a user-provided set from the full state space (remove states not in the active set), and
2. map a user-provided set from the full state space onto their internal states.

@franknoe https://github.com/franknoe do you know why the cited user wants to assign labels?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/markovmodel/PyEMMA/issues/866#issuecomment-232624272, or mute the thread https://github.com/notifications/unsubscribe/AGMeQgYEjBcfILNhaNiRzFiGx27zbB69ks5qVgvJgaJpZM4JJfvE.

---

Prof. Dr. Frank Noe Head of Computational Molecular Biology group Freie Universitaet Berlin

Phone: (+49) (0)30 838 75354 Web: research.franknoe.de

Mail: Arnimallee 6, 14195 Berlin, Germany