"Inhomogeneous space-times in loop quantum cosmology have come under better control with recent advances in effective methods. Even highly inhomogeneous situations, for which multiverse scenarios provide extreme examples, can now be considered at least qualitatively."

In the conclusion:

"In view of these new results, we must revise our multiverse scenario sketched in the beginning of this article, based on cosmological bounces. (See also [57].) We observed that inhomogeneous collapse combined with a transition to expansion (a “bounce”) may lead to causally disconnected regions, expanding within a larger multiverse. Inhomogeneity of this type is extremely hard to control with present-day non-perturbative quantum gravity, but good effective methods are now available to help us understand the relevant space-time structure. Loop quantum gravity, it turns out, implies radical modifications at Planckian densities, with a quantum version of 4-dimensional Euclidean space instead of space-time.

In Euclidean space, initial-value problems are illposed and there is no propagation of structure from collapse to expansion, as assumed in bounce models. Expanding patches that may result are causally disconnected not just from their surrounding space-time, but also from their predecessor which was collapsing. Instead of a bounce, loop quantum cosmology, once inhomogeneity is taken into account consistently, gives rise to a non-singular beginning of the expanding Lorentzian phase we can observe. The transition from Euclidean to Lorentzian signature, when [itex]\beta = 0[/itex], is a natural place to pose initial conditions, for instance for an inflaton state. These initial values are unaware of what happened in the collapse phase, so that the picture of dense collapsing patches bouncing first is not realized.

The new signature-change of loop quantum cosmology shares with bounces the combination of collapse with expansion, but the collapse phase does not deterministically affect the expansion phase. As a consequence, there is no entropy problem because no complete information is transmitted through high densities. And yet, the model is non-singular [58].

One may still view the possible collection of expanding universes within one space(-time), combining Euclidean and Lorentzian pieces, as a multiverse. However, any causal contact realized is even weaker than what is usually possible in multiverses, and it may seem more appropriate to talk of separate universes instead of one however connected larger structure. Each of these expanding patches has its own beginning when space-time emerges by signature change from 4-dimensional space, giving it a clear status as a universe of its own."