comprising long spans of quiescent “dead” sections of fiber for which the PMD
vectors are relatively stable, connec ted by livelier hinges which essentially rotate
those large chains as the environment changes. This model does not preclude the
“dead” sections from evolving over a long time scale as well, but on short time
scales, the hinges will dominate PMD dynamics.
The hinge model predicts different dynamics than the “fast mixing” assump-
tion, as can be appreciated by considering a long link with one active hinge. In the
fast mixing model, all the sections are capable of reorienting, so the fiber samples
the entire Maxwellian distribution. In this view, the hinge has no important bearing
on the statistics, because it is merely another retarder ...