Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories

arXiv:2411.05909 · hep-ph, astro-ph.HE · Submitted 2026-08-11 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: Next we'll be talking about the paper "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories".

Jocelyn: The paper was written by Lucca Radicce Justino, Clarissa Siqueira and Aion Viana from University of São Paulo and Chung-Ang University and Observatório Nacional.

Vera: Stay tuned as we take you through the paper and discuss its implications.

Paper discussion segment 1: Vera: So, we are looking at "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories." This isn't just a general look at dark matter; it’s specifically testing a theory called the Inert Doublet Model. Essentially, this model suggests that dark matter is made of new scalar particles—extra Higgs-like particles—that don't interact with normal matter in the usual ways, which is why they are so hard to find.

Jocelyn: Right, and what makes this specific model so interesting for the researchers is where it lives on the mass scale. While many searches focus on lighter particles, this paper focuses on a "high-mass regime," specifically looking at particles between five hundred GeV and twenty-five TeV.

Vera: Exactly, Jocelyn. And for those of us listening who aren't particle physicists, that mass range is huge! It’s much heavier than anything we can easily produce in a lab like the LHC.

Subrahmanyan: That's why the indirect detection method mentioned in the paper is so vital here. Because these particles are so heavy, if they do exist and they collide with each other in dense regions like the center of our galaxy, they should produce high-energy gamma rays that we can actually see with telescopes.

Jocelyn: And that’s exactly what this team did—they looked at the data from the H.E.S.S. telescope, which has been observing the Inner Galaxy for a long time. They weren't just looking for a simple signal; they were doing a very complex 2D likelihood analysis that looks at both where the gamma rays are coming from and their energy levels.

Vera: It’s a very thorough approach. They even compared their findings against the LUX-ZEPLIN experiment, which is one of our best tools for direct detection, to see where these different methods overlap or tell us something different.

Subrahmanyan: It's fascinating because they found that H.E.S.S. is already doing some heavy lifting here. For a specific way of modeling the dark matter distribution—what they call a "cuspy Einasto profile"—the data from H.E.S.S. actually excludes dark matter masses between one and eight TeV within this model.

Jocelyn: That's a massive chunk of the parameter space just gone! It means if the Inert Doublet Model is correct, and if it follows that specific density profile, the particle has to be either lighter or much heavier than that one to eight TeV window.

Vera: So, "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories" is essentially narrowing the search area by telling us where the particle probably isn't. But they aren't done yet, because they also looked forward to what the next generation of telescopes will do.

Subrahmanyan: Right, they projected what the Cherenkov Telescope Array, or CTAO, will be able to see. Their calculations suggest that CTAO is going to be a game-changer for this model. It’s expected to "comprehensively probe" the remaining viable space for the IDM.

Jocelyn: Basically, we are approaching a "now or never" moment for this specific theory. If CTAO doesn't see anything, then the high-mass version of the Inert Doublet Model might be in serious trouble.

Vera: It really sets the stage for why we need these next-generation observatories so urgently. We are moving from "maybe it's here" to "it must be in this tiny sliver of possibility, or it isn't here at all."

Subrahmanyan: And they even mention something called "Sommerfeld Enhancement," which could boost the signal significantly for particles between five and twenty TeV. If that effect is real, the H.E.S.S. data might already be pushing even more of the model into the excluded zone than they initially calculated.

Jocelyn: It's a race between our theoretical models and our ability to catch these high-energy photons as they fly across the cosmos. We're getting very close to some real answers here.l

Paper discussion segment 2: Vera: So, looking closer at "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories," what I find fascinating is how they use these massive telescopes to hunt for something that might be hiding in plain sight via its cousins. They aren't just looking for a single particle; they are looking at how this new Higgs doublet interacts with the particles we already know.

Jocelyn: Right, and because this model involves "co-annihilation," the signal could actually be stronger than what we usually expect from standard WIMP models. The authors found that in certain mass ranges, the annihilation cross section can reach up to one point zero times ten−25 cm3 s−1.

Subrahmanyan: That enhancement is huge for observers! If the signal is boosted like that, it means our current telescopes should be seeing something if this model is correct.

Vera: Exactly, Subrahmanyan, and that's why the H.E.S.S. data they analyzed is so critical here. By looking at the Inner Galaxy, they’ve already managed to exclude dark matter masses between one and eight TeV if you assume a "cuspy" Einasto profile for how dark matter is distributed in our galaxy.

Jocelyn: But it's not a total shutdown of the model yet. They found that if you look at much higher masses, specifically around the ten to twenty TeV range, there are still these little pockets of viable scenarios that haven't been ruled out by H.E.S.S.

Subrahmanyan: So we have these remaining "red points" on their plots?

Vera: Yes, those red points represent the models that survive everything: the relic abundance requirements, the direct detection limits from LUX-ZEPLIN, and the current gamma-ray observations. But even those are on borrowed time.

Jocelyn: That's where the Cherenkov Telescope Array Observatory comes in. The paper projects that CTAO will have enough sensitivity to essentially "sweep" through these remaining gaps, potentially probing almost all the remaining parameter space for this model.

Subrahmanyan: It sounds like we are on the verge of either discovering this specific scalar dark matter or proving it doesn't exist in this mass range at all.

Vera: That is precisely the tension they are highlighting. Whether it's through refined observations of the Galactic Center or better modeling of how gamma rays travel through our galaxy, "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories" shows we are narrowing the search window significantly.

Jocelyn: It really puts the pressure on next-generation instruments to finally give us an answer. If CTAO sees nothing where this model predicts a signal, the Inert Doublet Model is in serious trouble.

Subrahmanyan: It’s a classic "now or never" moment for this specific theory.

Vera: We'll see if the universe decides to reveal its secrets to the next generation of telescopes. Stay with us; we'll be back after the break.fofofofofofofofofofoefoefoefoefoeufoefoeufoeufoeufoeufoeufoeufuefuoefuoeffuefuoeffuefuoeffuefufoeuffufoeufeufufeufufeufufeufufeuuufeffufeffufeffufeffefeuffeuffeufefffefeuffeufeufeeuffeeufeeuuuufefuuuuuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeffffeuffeuefuffeuefufeeuuufufeeuuufefeuuuuffuuuuuuuufuuuuuuuuufuuuuuuuuuuuufefeeeeeeeeeeeeeefeeeeeeeeeeeeeefefeffffffeffffffffeffffffffeffffffffeffffffffefffffffffefeuefufufefffeffffeeeuefufufeff

Paper discussion segment 3: Vera: So, we are deep into our discussion of "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories," and what really strikes me is how much more sophisticated this analysis is compared to previous attempts. Instead of just assuming dark matter annihilates into one specific particle, like a simple pair of taus, these authors actually simulated the full continuum spectra for every scenario. They looked at how the dark matter decays into W bosons, Z bosons, and even Higgs bosons to get a much more realistic picture of what our telescopes should see.

Jocelyn: Right, because if you assume it only produces one type of particle, you might completely miss the signal or miscalculate the energy levels. By using these detailed simulations from tools like micrOMEGAs and CalcHEP, they’ve created a much more robust map of where this model lives.

Vera: Exactly. And when we look at Figure three in the paper, which compares different observatories, it’s pretty clear that the current H.E.S.S. data is already doing some heavy lifting by ruling out a huge chunk of this model's territory. Specifically, for a standard Einasto profile, they found that dark matter masses between roughly one and eight TeV are essentially excluded by the Inner Galaxy Survey data we already have from H.E.S.S.

Subrahmanyan: It is quite remarkable to see that level of exclusion, but we must remember the caveat mentioned in the text regarding astrophysical uncertainties. If the dark matter distribution in our galaxy follows a "cored" profile—meaning it's less dense at the very center—rather than this "cuspy" Einasto profile, those H.E.S.S. limits could drop by as much as two orders of magnitude!

Jocelyn: That is a massive swing in the data, isn't it? It shows how much our understanding of the dark matter "shape" in our own galaxy dictates whether we can claim a model is ruled out or not.

Vera: And that brings us to the big prediction for the future with CTAO, which stands for the Cherenkov Telescope Array Observatory. The paper projects that CTAO is going to be a game-changer because its sensitivity is nearly ten times better than what we have now. They suggest that CTAO will be able to probe almost every remaining viable scenario of this Inert Doublet Model, especially those with small mass splittings.

Subrahmanyan: Indeed, the paper highlights that CTAO’s ability to map larger regions of the sky makes it much less vulnerable to those "cored" vs "cuspy" profile uncertainties than H.E.S.S. was. While H.E.S.S. might lose a lot of steam if the center isn't dense, CTAO's wider reach means it can still find that signal even in more spread-out distributions of dark matter.

Jocelyn: It really feels like we are standing on the edge of a definitive test for this specific model. If CTAO doesn't see anything, the Inert Doublet Model is going to be in serious trouble across almost its entire high-mass parameter space.

Vera: We are essentially moving from an era of "maybe it's there" to an era of "if it exists in this form, we are absolutely going to see it." It’s a high-stakes moment for particle physics and cosmology alike.

Jocelyn: Absolutely. And before we move on, let's keep in mind that this paper, "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories," really underscores how much we rely on the synergy between different detection methods—combining direct searches like LUX-ZEPLIN with these massive gamma-ray arrays.

Vera: Coming up next, we'll look at how these results might shift if we account for even more complex effects like Sommerfeld enhancement. Don't go anywhere. manufacturing a signal where there is none is the nightmare of every astrophysicist!

Jocelyn: Haha, exactly! We'll be right back.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift through! Stay with us.mountains of data to sift

Conclusion: Vera: Well, it’s a sobering look at the landscape for the Inert Doublet Model. This paper really highlights how we are narrowing the window of where dark matter might be hiding by using these high-energy gamma-ray observatories to squeeze the parameter space from multiple sides.

Jocelyn: It really does. It's fascinating that while direct detection experiments like LUX-ZEPLIN are doing incredible work at lower masses, it’s these massive, ground-based telescopes looking at the Galactic Center that are starting to push back against the multi-TeV scenarios.

Vera: Exactly. The authors have essentially shown that if this model is correct and follows a standard Einasto profile, H.E.S.S. has already ruled out a huge chunk of it—specifically those masses between one and eight TeV.

Jocelyn: And the real kicker is the projection for CTAO. The paper suggests that once that next generation of telescopes is fully operational, it’s going to be a definitive test; it will likely probe or even exclude almost all the remaining viable scenarios for this model.

Vera: It really sets the stage for what we can expect from future observations. We've spent a lot of time today on "Constraints on the Inert Doublet Model of dark matter with very high-energy gamma-ray observatories," and it’s clear that the hunt is only getting more precise.

Jocelyn: Absolutely. It’s a great reminder that even when we don't find a signal, those "null results" are actually providing massive amounts of information about what the universe can and cannot be.

Vera: Well, on that note, we're going to take a quick break. When we come back, we have another fascinating paper to dive into. Stay with us!

Jocelyn: We'll be right back after this. Don't go anywhere!le

Lucca Radicce Justino, Clarissa Siqueira, Aion Viana

University of São Paulo · Chung-Ang University · Observatório Nacional

hep-ph, astro-ph.HE

Submitted: 2026-08-11

Comments: 13 pages, 6 figures. Matches the version published in PLB

DOI: 10.1016/j.physletb.2026.140809

Code: https://github.com/RadicceJustino/IDM-indirect-detection

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 65/100

The gist: This paper investigates constraints on the Inert Doublet Model (IDM) of dark matter using data from recent and future very-high-energy gamma-ray observatories, specifically the H.E.S.S.

Key concepts

Inert Doublet Model (IDM)
This is a theory suggesting dark matter is made of new scalar particles, which are extra Higgs-like particles. These particles do not interact with normal matter in the usual way, making them difficult to detect.
High-mass regime
The paper focuses on a mass range for these dark matter particles between five hundred GeV and twenty-five TeV. This is a very heavy range, much larger than what current particle accelerators like the LHC can easily produce.
Indirect detection method
This method involves looking for high-energy gamma rays produced when these heavy dark matter particles collide in dense regions, such as the center of our galaxy. These gamma rays can be observed by telescopes.
Sommerfeld Enhancement
This is a phenomenon that could significantly boost the signal from dark matter particles between five and twenty TeV. If this effect is real, it might push the H.E.S.S. data to exclude more of the model's parameter space.

Terminology

Summary

This paper investigates constraints on the Inert Doublet Model (IDM) of dark matter using data from recent and future very-high-energy gamma-ray observatories, specifically the H.E.S.S. telescope and the upcoming Cherenkov Telescope Array Observatory (CTAO).

The IDM is a minimal extension of the Standard Model (SM) where a new Higgs doublet is added, stabilized by a discrete Z2 symmetry, allowing the lightest neutral particle to be a dark matter candidate. The model has four physical parameters relevant to dark matter phenomenology: the heavy scalar mass m H (the dark matter candidate), the mass splittings Δ+ and Δo of the co-annihilating particles, and the physical coupling λ345 of the dark matter particle with the SM.

The paper focuses on the high mass regime (m H ≳ 500 GeV), where the correct dark matter relic abundance can be achieved for masses between approximately 500 GeV and 25 TeV. This regime is dominated by co-annihilation processes, which elevate the thermal relic velocity-weighted annihilation cross section to the range of 0.5 to 1.0 × 10−25 cm3 s−1, enhancing the potential gamma-ray signal from dark matter annihilation. The authors note that in this regime, the DM particle is expected to annihilate, producing gamma rays in the TeV energy scale and that the annihilation cross section is enhanced by co-annihilation effects in the high mass limit, leading to a higher gamma-ray signal.

The authors performed two scans across the parameter space. The first scan generated 250,000 scenarios with parameters logarithmically sampled in the ranges: 0 < λ345 < 2π, 300 GeV < m H < 30 TeV, 0.5 GeV < Δ+ < 10 GeV, and 0.5 GeV < Δo < 10 GeV. Constraints were applied sequentially: (1) unitarity and inertness constraints, (2) the measured relic abundance by the Planck satellite (Ωh2 = 0.1200 ± 0.0012), and (3) upper limits at 90% C.L. from the LUX-ZEPLIN (LZ) direct detection experiment. The micrOMEGAs 5.3.35 software was used to evaluate relic abundance.

For indirect detection limits, the authors computed the full continuum gamma-ray spectra for each model scenario based on detailed simulations of all relevant annihilation channels and their branching fractions, rather than using simplified pure-channel assumptions. These spectra were injected into a binned two-dimensional likelihood analysis (energy and spatial dimensions). A circular 1° region centered at the Galactic Center was considered, subdivided into seven concentric rings, each 0.1° wide, starting at 0.3° from the GC, with a mask within galactic latitudes b = ±0.3° applied. A cuspy Einasto profile was assumed with parameters ρ s = 0.079, r s = 20 kpc, and α = 0.17, calculated assuming a local dark matter density of 0.39 GeV/cm3.

For H.E.S.S., the authors used ON and OFF counts from real data corresponding to 254 hours of observation, extrapolated to 546 hours to match the cumulative observation time of the 2014-2020 Inner Galaxy Survey. For CTAO, the dataset was simulated using the updated Alpha Configuration IRFs (prod5 version v0.1), with 525 hours of observation time considered.

After applying all constraints, approximately 800 scenarios (red points) satisfy relic abundance and direct detection limits. The LZ exclusion provides a curve λ345 ∼ m H 3/2 which considerably restricts the available parameter space. Direct detection excludes some models with m H ≲ 10 TeV.

The authors find that dark matter particle masses within the 1-8 TeV range are excluded by current data, assuming a benchmark cuspy Einasto profile. Specifically, H.E.S.S. excludes masses between ≈ 1 TeV and ≈ 10 TeV, for all benchmarks. The paper states: the most recent indirect detection data exclude most IDM scenarios from the Inner Galaxy Survey of the H.E.S.S. telescope.

The projected CTAO limits are almost one order of magnitude stronger than those from H.E.S.S., and they would exclude all viable scenarios. The authors project that CTAO will be able to probe all the DM parameter space of the Inert Doublet model with small mass-splitting (Δo,+ = m A,H± − m H ≲ 10 GeV).

The paper discusses the potential impact of the Sommerfeld Enhancement (SE), a non-perturbative effect occurring when the velocities of initial-state particles are non-relativistic. For a broad region of parameter space, the thermal cross section in the Galactic Center can be enhanced by factors as large as 104 for masses between 5 and 20 TeV. The authors note that "a significant region of the parameter space with m H > 10 TeV is expected to be excluded by the HESS 546h data when the non-perturbative effects are included."

The main annihilation channels in the high mass regime are the bosonic channels W+W−, ZZ, and hh. The W+W− and ZZ states dominate, accounting for almost 90% of annihilation channels, with both channels having branching fractions around 50% in the lower mass sample. The W+W− state is dominant over ZZ for m DM > 1.5 TeV for most scenarios. The Higgs hh channel does not surpass ≈ 20% of contribution, while the electroweak radiative γW+W− correction reaches contributions of at most ≈ 10%.

The paper concludes that "after imposing all limits, including theoretical and direct and collider constraints, the most recent Inner Galaxy Survey by the H.E.S.S. telescope excludes nearly the entire viable parameter space of the IDM, choosing the Einasto halo profile. Furthermore, considering the most recent IRFs, we computed the projected limits for the CTAO and showed that it will probe virtually all remaining viable dark matter scenarios within the high mass regime of the IDM."

The authors emphasize the complementarity between direct detection and indirect detection limits: "for small mass-splittings, Δ+, Δo ≲ 1 GeV, LZ exclusion dominates in the 0.5 TeV−2 TeV mass range, while H.E.S.S. is stronger for m DM in 2 − 20 TeV. For greater mass splittings, H.E.S.S. always dominates over direct detection."

The paper also notes important caveats: the limits are inherently tied to the spatial distribution of the chosen dark matter density profile, and assuming a cored density profile could weaken H.E.S.S. limits by up to two orders of magnitude, while CTAO sensitivity deterioration in cored scenarios is much less severe, scaling down by a factor of only a few. The CTAO projections represent an optimistic baseline projection using statistical errors only, and full integration of systematic errors and Galactic Diffuse Emission contamination would degrade limits by factors of approximately 2 to 6.

Improvements for AI systems

Based on the scientific paper, here are specific improvements that can be made to AI systems, particularly those used for particle physics phenomenology, dark matter model scanning, and gamma-ray data analysis:

Improvement: Replace uniform random sampling with an AI-driven active learning loop that uses Bayesian optimization or Gaussian processes to concentrate samples in regions where the relic abundance constraint (Ωh2 ≈ 0.1200) is satisfied.

What the improved AI system can do:

  • Dynamically allocate computational resources to high-priority regions (e.g., mass splittings Δ < 10 GeV, m H between 500 GeV and 25 TeV)

  • Reduce the number of required micrOMEGAs evaluations by 70–90% while maintaining full coverage of viable parameter space

  • Automatically identify and flag fine-tuned regions (e.g., where co-annihilation dominates) for targeted high-resolution scans

Improvement: Train a deep neural network (or gradient-boosted tree ensemble) as a fast surrogate for micrOMEGAs' relic density computation, using the 250,000 scanned points as training data.

Improvement: Use a conditional generative model (e.g., a normalizing flow or VAE) to produce annihilation spectra dN/dE for arbitrary (m H, Δ+, Δo, λ345) combinations, trained on the PPPC4DMID outputs.

Improvement: Implement a GPU-accelerated, vectorized computation of the Poisson log-likelihood (Eq. 9) across all spatial ROIs and energy bins simultaneously, using JAX or PyTorch.

Improvement: Extend the current statistical-only analysis to include nuisance parameters (e.g., J-factor uncertainty, background normalization, energy scale systematics) and use a neural network to approximate the profile likelihood ratio.

Improvement: Use a multi-objective optimization algorithm (e.g., NSGA-II) to identify the most constraining benchmark points that maximize coverage of the viable IDM parameter space while minimizing computational cost.

Improvement: Train a neural network to predict the non-perturbative enhancement factor as a function of (m H, Δ+, Δo, velocity) using the results of Ref. [51] as training data.

Improvement: Build an AI pipeline that automatically re-casts limits from different experiments (H.E.S.S., CTAO, LZ, Fermi-LAT) onto a common (m DM, ⟨σv⟩) grid with consistent halo profile assumptions.

Improvement: Deploy a web-based AI assistant (e.g., a fine-tuned LLM with access to the trained surrogates) that answers queries like Is m H = 5 TeV, Δ+ = 2 GeV, Δo = 8 GeV viable? in under a second.

Improvement: Use an LLM fine-tuned on physics papers to automatically extract the model definitions, constraints, and analysis procedures from papers like this one, generating executable Python/C++ code.

Summary of Impact: These improvements would reduce the computational cost of IDM (and similar model) analyses by 10–100×, enable real-time exploration of parameter space, produce more robust and realistic exclusion limits, and make the entire pipeline reproducible and accessible to the broader physics community. The most impactful single improvement is the surrogate model for relic abundance (item 2), which would enable scans that are currently computationally prohibitive.

Sources

Related papers