Thread by @RKLeane

New paper out today!

Dark Matter Strikes Back at the Galactic Center
arxiv.org/abs/1904.08430

We show that dark matter annihilation might explain the excess of gamma rays detected at the center of our galaxy, after all.
Mega-thread explaining our results and backstory below!

1/ Back in 2009, Dan Hooper (@DanHooperAstro) and Lisa Goodenough used publicly available data from the Fermi Gamma-Ray Space Telescope (@NASAFermi) to examine gamma rays (very high-energy light) from the center of our Milky Way galaxy.

2/ They found an excess of gamma rays with energies a billion times that of visible light. Since then, there has been extensive debate over the origin of these mysterious excess high-energy gamma rays.

3/ An exciting possibility is that we are seeing the first signal of annihilating dark matter (DM) - DM interacting with visible matter. This is an appealing explanation because:

4/ 1. The center of the galaxy has a lot of gravitational pull, and DM likely accumulates there, where it could annihilate to produce visible particles we can detect with telescopes,

5/ 2. The excess is approximately spherically symmetric around the galactic center, unlike most signals from visible matter, which look more like the disk of our galaxy. This is close to what's expected from DM annihilation if DM density follows one common estimate (called NFW)

6/ 3. The DM annihilation rate required to produce the excess matches the annihilation rate DM would need to have in the early universe to produce the amount of DM we observe in the universe today (assuming it is a type of DM called a thermal WIMP).

7/ The leading alternative hypothesis is that the excess is produced by millisecond pulsars (old stars spinning with millisecond periods). This is a popular alternative as pulsars can produce a comparable energy spectrum of gamma-rays to what we see in the excess.

8/ This is a good possible explanation, but there are still questions as to how you get the right number and distribution of pulsars to explain the excess. The point is, we really don't know the answer.

9/ So, how do we tell these hypotheses apart? If it were pulsars, can't we just measure their gamma-rays and and compare directly? Well, none of the pulsars we have detected so far explain this excess; it can't be coming from bright pulsars.

10/ If there are many faint pulsars that are too dim to be detected individually, there could be enough of them to collectively contribute enough gamma-rays to explain the excess. But they are too faint to see, so we need a technique to characterize them.

11/ An excellent method was introduced in 2011 by Malyshev and Hogg, and was first applied to this excess in 2015 by Lee, Lisanti, Safdi, Slatyer & Xue, which is called "Non-Poissonian template fitting" (NPTF).

12/ Traditional template fitting builds up a picture of the sky as a combination of "templates", each corresponding to a particular spatial distribution of gamma rays. We can calculate which combination gives the best description of the data.

13/ NPTF is an extension of this which exploits the fact that the statistics of gamma-rays produced by DM vs pulsars can be vastly different.

14/ That is, in a given pixel of the sky, the variance in the number of photons can be much higher from pulsars, as you may have several pulsars in a pixel _all_ giving you gamma-rays - or zero pulsars, giving you none. In contrast, DM can be everywhere.

15/ This increased variance doesn't apply just to pulsars, but to any "point sources" (PSs) of gamma-rays - individual gamma-ray-emitting sources.

16/ Ok, so, you can add a "NFW DM template", can see how much of the gamma rays are attributed to DM, compared to PSs distributed the same way (an "NFW PS" template), based on the fact that they can make different predictions for the number of gamma-rays observed.

17/ (Recall NFW is just the profile shape which matches the emission of excess gamma rays, so we are comparing really if DM or PS is the better fit to the excess.)

18/ In 2015 when NPTF was first applied, the result was as follows: evidence for a PS population was found, with high statistical significance, and the DM contribution was consistent with zero.

19/ We asked, what is driving this preference? We guessed that if there are some PSs really present in the data, but not described by the templates we are using, it is possible that this could bias the result away from the DM explanation.

20/ We set out to test this in two ways: (i) using simulated data, and (ii) using the real Fermi data.

21/ We first simulated data, putting PSs in a large structure in our galaxy called the Fermi Bubbles. It wasn't known before if PSs are in the Bubbles, so they are the new ingredient for testing a bias. We then simulated a DM signal, as well as all the backgrounds.

22/We then analyzed the simulated data with the standard templates (the ones to make the 2015 figure above,where there is no DM found).This does not include a Bubbles PS template.When we do this, even though we simulated DM, it's not found, and is instead misattributed to NFW PS!

23/ We added even more DM flux in our simulation, and even more flux went into the PSs!!

24/ Amazing. So, next we looked for PSs in the Fermi Bubbles for real in the Fermi data, but they aren't there. So, this exact example isn't happening in the data, but serves as a proof-of-principle example of how a DM signal can hide behind other unmodeled PS distributions.

25/We then wanted to see if something like this is happening in the real data. Of course in the real data you don't know the true distributions of all the faint PSs, which is a complication. But, in the simulated data, we found that adding more DM kept pushing up the flux in PSs.

26/ So, if this sort of biasing effect is present in the real data, it is likely not saturated in its ability to absorb DM flux. So we injected an artifical DM signal to the real data, to see if it was recovered, or if it was _also_ misattributed to PSs.

27/ In the real Fermi data... it was also misattributed to PS!!

28/ We add more and more, still totally misattributed until at least a factor 5 larger than the excess size itself is injected!

29/ What do we learn from this? We could have DM explaining this excess, and previous studies using the NPTF to disentangle what was producing the excess might have missed it. DM might fully explain the excess, after all.

30/ To be clear, this is not positive evidence for dark matter, but evidence of a biasing effect which can drive down the DM flux, and push up PSs. This bias can hide a real DM signal! DM is back in the game, though maybe it is still PS. Lots of interesting follow-ups to be done.

I also want to say a massive thanks to my awesome collaborator Tracy Slatyer!!