A demonstration of the expanding universe, where every object becomes further away from every other unless they are gravitationally bound.
Weak lensing in action, where foreground structures (blue) distort the background shapes, shearing them tangentially, magnifying them, and displacing them. Galaxy shapes that are distorted in this way can trace this foreground structure, even if that structure is entirely invisible to our instruments.
When observing a passive galaxy spectrum, we see a sharp drop off in flux around 400 nm. As the galaxy becomes further away, we can see this break shifting to higher wavlengths. When we only have the images collected by filters, rather than the entire spectrum, we see this feature as the galaxy vanishes from our image. The correlation of when the galaxy disappears is closely linked to its distance!
A star forming galaxy with the O-II doublet (in blue) at increasing distance and redshift away from us. When the blue line passes inot the gray region, it is undetectable by most silicon-based detectors and measuring the redshift from a given spectrum becomes difficult.
A depiction of gravitational lensing in an idealized background shows how light gets warped around an (invisible) foreground dark matter halo. As the lensing mass moves towards and away from the observer we can see the shapes of galaxies change.
A depiction of strong gravitational lensing in front of the Hubble Ultra-Deep field, producing so-called Einstein rings when a galaxy is directly behind the lensing halo.
A depiction of a self-organizing map (SOM) learning the shape of a 3D data manifold, and reducing the dimensionality to 2D map efficiently.
A depiction of a self-organizing map (SOM) at work, discretizing a 3D space into the 2D map seen on the right. This unsupervised machine learning technique preserves the geometry and distribution of the data.