Large-scale patterns of galaxy spin rotation show cosmological-scale parity violation and multipoles

TL;DR

This study analyzes the spin directions of over 64,000 galaxies, revealing significant cosmological-scale asymmetries and potential violations of isotropy, with findings supported by a fully automatic, bias-free data analysis process.

Contribution

It provides the first large-scale evidence of parity violation in galaxy spins and introduces a fully automatic, bias-resistant method for analyzing galaxy orientation data.

Findings

Significant asymmetry in galaxy spin directions at cosmological scales.

Redshift-dependent homogenization of galaxy spin distribution.

Photometric differences between galaxies with opposite spins, normalized by redshift.

Abstract

The distribution of spin direction of ~6.4*10^4 spiral galaxies with spectra was examined. The analysis shows a statistically significant cosmological-scale asymmetry between galaxies with opposite spin direction. The data also reveals that the asymmetry changes with the direction of observation, and with the redshift. The redshift dependence shows that the distribution of the spin direction of galaxies becomes more homogeneous as the redshift gets higher. The data also show photometric differences between galaxies with opposite spin patterns. When normalizing the data by the redshift, the photometric asymmetry is eliminated. However, when normalizing the data by the magnitude, statistically significant differences in the redshift remain. These evidence suggest a violation of the cosmological isotropy and homogeneity assumptions. Fitting the distribution of the galaxy spin directions to a quadrupole alignment provides fitness with statistical significance >5 sigma, which grows to >8 sigma when just galaxies with $z>0.15$ are used. The data analysis process is fully automatic, and it is based on deterministic and symmetric algorithms with defined rules. It does not involve neither manual analysis of the data that can lead to human perceptual bias, nor machine learning that can capture human biases or other subtle differences that are difficult to identify due to the complex and non-intuitive nature of machine learning processes.