Abstract
Abstract
Circles of low-variance and Hawking points in the Cosmic Microwave Background (CMB), resulting from black hole mergers and black hole evaporation, respectively, in a previous cycle of the universe, have been predicted as possible evidence for the Conformal Cyclic Cosmology model (CCC) introduced by R. Penrose. We present a high-resolution search for such low-variance circles in the Planck and WMAP CMB data, and introduce HawkingNet, our machine learning open-source software based on a ResNet18 algorithm, to search for Hawking points in the CMB.
We find that spots consisting of a few unusually bright (high-temperature) or dark (low-temperature) pixels, erroneously lead to regions with many low-variance circles, and consequently sets of near-concentric low-variance circles, when applying the search criteria used in previous work [1]. After removing those spots from the data, no statistically significant low-variance circles can be found.
Concerning Hawking points, also no statistically significant evidence is found when using a Gaussian temperature amplitude model over ∼ 1° opening angle and after accounting for spots of unusual brightness.
That the unusual spots in the data are themselves remnants of Hawking points is not supported by low-variance and/or low-temperature circles around them.
The absence of such statistically-significant distinct features in the currently available CMB data does not disprove the CCC model, but implies that higher resolution CMB data and/or refined CCC based predictions are needed to pursue the search for CCC signatures further.