Buongiorno:
I present you what appears to be
the first detection of the magnetic cosmic web
via the stacking of ~390,000 large pairs of galaxies, imaged in radio (MWA & LWA) and X-ray (ROSAT).
https://arxiv.org/pdf/2101.09331  by @tvern23 et al. (incl. me)
This is huge!
#magcow

So far,no detection of radio emission from filaments detached from clusters have been reported.
Which is understood because of the low particle density and magnetic fields. But that was annoying nevertheless because the magnetic web should be there, as shown by many simulations.

In this work (**submitted**), @tvern23 has used an impressive number of surveys at different wavelenghts (radio:MWA, LWA X-ray:ROSAT, microwaves: PLANCK; IR:Wise,Iras,Akari) to attempt a statistical detection of the signal from the gas between the nodes of the cosmic web.

Starting from the SDSS survey of galaxies, about ~390,000 large red galaxies were selected, with a linear separation in the range of 1-15Mpc, and on average located at z~0.14. This was largely inspired by X-ray stacking works by Tanimura+ @ByoPic_ERC ( https://arxiv.org/pdf/2011.05343.pdf)

In the stacking technique, you basically sum together N images of (what you think is) a realisation of the same class of objects, counting on the signal to noise ratio to scale as ~sqrt(N). Some astro-amateurs also do this! https://astrobackyard.com/tutorials/stack-exposures/

To know if you detected something, you also need to produce a "null" stacking sequence, in which you average N random maps, and at the end you hope to see a difference between the null and the true stacking.

The extra difficulty here is that you want also to remove the true signal of the(likely) diffuse emission from each galaxy halo-not the filamentary one we want to highlight here. So the "null" sequence contains random rotations of each galaxy in the pairs to produce a halo model.

Then you hope to see a significant signal in the region between aligned pairs, or and the control sample including only the noise and the overlap of spherical profiles. And there's indeed a significant (~5σ) excess signal there, in all MWA-GLEAM and LWA radio images!

(there are N technical and important issues, discussed in the paper much better than I can do, about removal of pointlike sources via wavelets, background etc)

This is the distribution of the signal in the true stacking and in the control sets, for various resampling, to get a statistics. The distance between control and physical pairs is always very large.

Also, very importantly: the same experiment repeats and confirm the stacking by Tanimura+ with ROSAT, indirectly confirming that the overall (complex) procedure is sound.

Ok, so... What is this significantly excessive signal?
How can we know it's the radio cosmic web?

First: the radio spectrum is very flat, α~-1.0 \\pm 0.1, which what you would expect from the synchrotron emission by electrons accelerated by strong shock waves around filaments. https://arxiv.org/pdf/1503.08983  (look at the yellowish trends in filaments between halos)

This spectrum seems too flat for unresolved AGNs.
Also: a unresolved population of faint radiogalaxies, which may in principle produce this signal,would be dominated by star forming galaxies.But the corresponding stacking of IR images showed no signal, hence we can discard this.

Given the large pair separation and the observed spectrum, this seems also incompatible with "bridge" emission (e.g. https://arxiv.org/pdf/1906.07584  or https://arxiv.org/pdf/2008.09613  by @and_botteon ), which would have instead α<-1.3

Finally: we have simulations to the rescue! @tvern23 applied the same stacking technique to random and true pairs extracted by our #magcow simulations, finding a stacked signal only a factor ~4 lower than the observation (which, believe me, is unexpectedly close!)

The same excellent comparison is found with the stacking of the X-ray emission of simulated filaments vs the true ROSAT stacking- less surprising in this case as the thermal WHIM properties are somewhat consolidated.