General prescription for selecting observables for the observable drift method

Develop a general prescription for selecting observables used to compute the observable drift in order to ensure that the drift and its variance provide meaningful diagnostics of regularity versus chaos across different systems.

Background

The observable drift framework requires choosing a measurable observable whose time-integrated deviations from its mean serve as a random walk; inappropriate choices (e.g., constants of motion or overly localized observables) can yield uninformative results.

The authors explicitly state that they cannot currently provide a general prescription for choosing such observables, highlighting a methodological gap that limits routine application and comparability across systems.