I fear you will require that particle expansion accelerates with the inverse of particle separation, which will result in path-dependent anisotropic particle size variations.  I can measure the size of atoms after path variations, and compare them, to invalidate such a theory, on uniform curvature assumptions.  Something akin to general relativity might rescue it, however.  This is the path that leads to epicycles.

IMO this approach is a more interesting innovation: http://arxiv.org/pdf/1001.0785v1.pdf