How can galaxies that are barely stretched map dark matter?
Sandbox Physics ·
Mass bends light, so every distant galaxy seen through a cluster or a filament of dark matter arrives slightly stretched, by a few percent at most, and always along circles around the mass. One galaxy cannot show this, because its own shape is about ten times more elongated than the stretch. Hundreds of neighbouring galaxies can. Their own shapes point in random directions and average away; the shared stretch does not. Turning that leftover stretch, patch by patch, into a picture gives a map of all the mass in front of the galaxies, whether it shines or not.
Mass acts like uneven glass
Light from a galaxy far behind a cluster does not travel in a straight line. Gravity tilts its path a little toward the cluster, and the tilt is larger for light passing closer to the mass. A small galaxy image is therefore not just shifted. Its near side is pulled more than its far side, so the image is magnified slightly and stretched along the circle around the cluster centre. The effect is the same as looking through a sheet of glass of uneven thickness.
Close to the centre of a massive cluster the stretch can be extreme, producing long arcs. That is strong lensing. Almost everywhere else the stretch is a few percent or less. That is weak lensing, and it is the kind that covers the whole sky. Weak Gravitational Lensing, Sec. 3
One galaxy tells you almost nothing
Galaxies are not circles. Seen on the sky, a typical galaxy is an ellipse whose elongation, measured along one direction, scatters by about 0.26 from galaxy to galaxy. The Dark Energy Survey quotes exactly this number for its 100 million galaxies. Cosmological Constraints from Galaxy Clustering and Weak Lensing, Sec. II
A large cluster stretches galaxies a few arcminutes from its centre by roughly 0.03. A single galaxy that looks elongated could simply be an elongated galaxy. There is no way to tell from one image how much of its shape came from lensing.
Averaging hundreds of galaxies
The way out is that galaxies do not know about each other's orientation. Average many neighbouring galaxies and their own shapes, pointing every which way, cancel. The lensing stretch, which all of them share, survives.
The leftover noise falls as one over the square root of the number of galaxies. With a scatter of 0.26, measuring a stretch of 0.05 with an error one third of that takes about 240 galaxies. A deep telescope image holds about 20 measurable galaxies per square arcminute, so 240 galaxies come from a patch roughly 3.5 arcminutes across, about a ninth of the Moon's width. This is why weak lensing maps are blurry: each point of the map is an average over a patch of sky.
The pattern points to the mass
Lensing leaves a specific pattern. Around a lump of mass, images line up along circles, like the bark rings of a tree seen end on. Around an empty region they line up along spokes. A pattern of swirls, where images line up along spirals that curl one way, cannot be made by lensing at all.
That rule is useful twice. In 1993 Nick Kaiser and Gordon Squires showed how to turn the measured stretch pattern back into a map of the projected mass. The same calculation produces a second map, built from the swirl part of the pattern, which should hold only noise. If it holds structure, something other than lensing is shaping the galaxies, such as an imperfect correction for the telescope's own blurring. Discriminating weak lensing from intrinsic spin correlations using the curl-gradient decomposition
What the map cannot tell you
A lensing map shows differences, not totals. Add a thin, perfectly uniform sheet of mass across the whole field, weaken every lump by a matching factor, and no galaxy changes its shape, only its size and brightness. Shapes alone therefore leave the overall level of the map open. Calibrating it needs the sizes or brightnesses of galaxies, or a field so large that its average density is known.
The map also depends on where the galaxies are. Mass behind them bends nothing they send to us. Mass about halfway between us and them counts most. A map made from nearby galaxies and a map made from distant ones show the same matter with different weights.
Finally, galaxies can line up on their own. Neighbouring galaxies form in the same tidal field and may point the same way before any lensing happens. Christopher Hirata and Uroš Seljak showed in 2004 that this alignment can even partly cancel the lensing signal, because the tidal field that lines up nearby galaxies also lenses the ones behind them. Intrinsic alignment-lensing interference as a contaminant of cosmic shear
From one cluster to the whole sky
The same averaging works on the faint stretch that ordinary large-scale structure gives to every galaxy, about one percent. That is too weak to map lump by lump, but it can be measured statistically: how similar are the stretches of two galaxies a given angle apart? The Dark Energy Survey measured this for over 100 million galaxies across 4143 square degrees in its first three years. From the shapes alone it pinned down how clumpy matter is today to about 3 percent. Cosmology from Cosmic Shear and Robustness to Data Calibration
The weak-lensing mass map Lab puts two hidden halos in front of a synthetic field of galaxies. Turn off the shape scatter and the circles appear; turn it back on and only the averaged map finds the halos.