10:42 AM
Apparently that thing I noticed where the synchronization structure isn't actually a proper connection has been already expanded upon in some paper
They called it a non-principal Ehresmann connection
"Although a general Ehresmann connection on a fiber bundle on which no Lie group structure is defined or considered has already been formulated (see e.g., Refs. 1 and 2). Without a group structure, the connection form is described in terms of an endomorphism field, and the curvature tensor can be reinterpreted in terms of Frölicher-Nijenhuis bracket. "'
"Note that, by the Frobenius theorem, the space distribution $S$ is integrable if $\tau \wedge d\tau = 0$ (which in coordinates becomes a rather index-trashed equation, $g_{ab} T^a \partial_i (g_{jk} T^j) + \mathrm{cycl}(b, i, k) = 0$)."
"The fiber bundle (2.7) actually is a principal fiber bundle: the group ${\exp tT }$ defines action via transport along the integral curves of T . However, this group action is not—in general—congruent with the horizontal distribution; that is, $\kappa$ is not a principal connection. The magnitude of the deviation from being principal can be measured by the Lie derivative of $\kappa$ along $T$ ; call it the torque of the observer"