Paper day: Did you see the @eBOSSurvey results and think “Nice, but I wish they had even more data”? Then this is for you.

We show how to effectively quadruple the data volume of the LRG survey for free – no new observations! – using info from voids.

https://arxiv.org/abs/2008.06060 

One way you can get more cosmological information from the large-scale structure of the Universe is to measure more of it – build a better telescope, survey a larger area, observe more galaxies.

This is what @desisurvey will do over the next 5 years.

A second way is to improve your theory so you can use more small-scale data: this is what the EFT approach to large-scale structure achieves.

A third way is to measure *differently* – use a new observable in addition to galaxy 2-pt statistics. That's what we do.

Voids are regions of space with low matter density, where galaxy motions can be modelled by linear perturbation theory alone.

These motions cause distortions to the apparent shapes of voids so they no longer appear symmetrical.

Because we can model this distortion very well, we can use voids as "standard spheres": the observed asymmetry in their shapes tells us about cosmological distances, and how fast structure is growing.

This is information we can't get as accurately from galaxy clustering alone!

As a result, if you take the current state-of-the-art, top-drawer results from the combination of several different BAO and full-shape galaxy clustering techniques ...

... adding voids on top *doubles* the precision we can achieve!

Technically, this is a 55% reduction in the allowed parameter space. As the uncertainty goes roughly as the square root of the survey volume, that's equivalent to analysing a survey 4x as large ...

Since I forgot to add it earlier in the thread, here's a simplified picture of the model we use and the data it is fit to. Have a look at the paper for more details! https://arxiv.org/abs/2008.06060