A frame dragging, in local coordinates in the vacuum regions of space time
ds^2 = ( a(y) dt + b(y) dx)^2 - (c(y) dt + f(y) dx)^2 -dy^2...
can of course be transformed away locally by a linear transformation
into Minkowski form
ds^2 = dt^2 - dx^2 - dy^2 ...
exept if on closed y-submanifolds
((a,b),(c,d))(y)
is a non trivial SO(1,3) bundle.
This is certainly the case if
1) the y=const manifolds are spacelike rotational surfaces x infinite
time or deformations
and
2) a complete turn for y=0->2pi along a these directions does not result
in an identity transformation for the local frames.
In this case one needs an atlas of more than one overlapping
differentiable coordinate patches to cover the manifold.
The linear case seems to be similar to linear rotational flow in
Newtonian fluid dynamics.
Probably it plays no role because in vacuum it will reduce to a trivial
bundle as solution to Einsteins vacuum equations.
The alternative of a infinite space, uniformly filled with matter,
rotating locally in one t-x-plane, would probably violate all
cosmological symmetry principles.
If such a solution is possible mathematically with a reasonable Ricci
tensor as momentum tensor of matter is decidable with twentysomething
lines of curvature calculation.
--
Roland Franzius