Beecroft Institute for Particle Astrophysics and Cosmology

Abstract:

We present the first detection of mass dependent galactic spin alignments with local cosmic filaments with >2σ confidence using IFS kinematics. The 3D network of cosmic filaments is reconstructed on Mpc scales across GAMA fields using the cosmic web extractor DisPerSe. We assign field galaxies from the SAMI survey to their nearest filament segment in 3D and estimate the degree of alignment between SAMI galaxies’ kinematic spin axis and their nearest filament in projection. Low-mass galaxies align their spin with their nearest filament while higher mass counterparts are more likely to display an orthogonal orientation. The stellar transition mass from the first trend to the second is bracketed between 1010.4 M⊙ and 1010.9 M⊙, with hints of an increase with filament scale. Consistent signals are found in the Horizon-AGN cosmological hydrodynamic simulation. This supports a scenario of early angular momentum build-up in vorticity rich quadrants around filaments at low stellar mass followed by progressive flip of spins orthogonal to the cosmic filaments through mergers at high stellar mass. Conversely, we show that dark-matter only simulations post-processed with a semi-analytic model treatment of galaxy formation struggles to reproduce this alignment signal. This suggests that gas physics is key in enhancing the galaxy-filament alignment.

Abstract:

We use Bayesian convolutional neural networks and a novel generative model of Galaxy Zoo volunteer responses to infer posteriors for the visual morphology of galaxies. Bayesian CNN can learn from galaxy images with uncertain labels and then, for previously unlabelled galaxies, predict the probability of each possible label. Our posteriors are well-calibrated (e.g. for predicting bars, we achieve coverage errors of 11.8 per cent within a vote fraction deviation of 0.2) and hence are reliable for practical use. Further, using our posteriors, we apply the active learning strategy BALD to request volunteer responses for the subset of galaxies which, if labelled, would be most informative for training our network. We show that training our Bayesian CNNs using active learning requires up to 35–60 per cent fewer labelled galaxies, depending on the morphological feature being classified. By combining human and machine intelligence, Galaxy zoo will be able to classify surveys of any conceivable scale on a time-scale of weeks, providing massive and detailed morphology catalogues to support research into galaxy evolution.