⚛️ phenomenology

Constraints on Annihilating Dark Matter from Gamma-Ray Background-Galaxy Shape Correlations: Model-independent Null Results and Moderate Template-based Signals

Using 14 years of Fermi-LAT gamma-ray data and DES Y3/DECADE galaxy shapes to perform a model-independent cross-correlation analysis over 12,000 deg², this study finds null results that constrain wino-like dark matter and thermal annihilation scenarios, while a template-based approach reveals a moderate 3σ signal with deviations from standard background expectations.

Original authors: Masato Shirasaki, Deheng Song, Oscar Macias, Shunsaku Horiuchi, Naoki Yoshida

Published 2026-07-14

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Masato Shirasaki, Deheng Song, Oscar Macias, Shunsaku Horiuchi, Naoki Yoshida

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe is a giant, dark ocean. We know there's a massive amount of invisible "dark matter" swimming in it, but we can't see it directly. Scientists have a hunch that these invisible particles might bump into each other and vanish in a flash of light, creating high-energy gamma rays. If this happens, the sky should glow with a faint, fuzzy background of these rays, and the places where the light is brightest should match up with the places where the dark matter is most crowded.

To test this, the authors of this paper acted like cosmic detectives. They grabbed two massive datasets: 14 years of gamma-ray snapshots from the Fermi telescope and a detailed map of the shapes of 270 million galaxies from the Dark Energy Survey and the DECADE project. Think of the galaxy shapes as a way to trace the invisible dark matter; just as wind bends the grass to show where the breeze is strongest, the gravity of dark matter stretches the shapes of distant galaxies.

The team tried to find a "dance" between these two datasets: do the fuzzy gamma-ray patches wiggle in sync with the stretched galaxy shapes? They used a new, super-precise math tool (Fourier-space analysis) to look for this connection across a huge patch of sky, about 12,000 square degrees—roughly the size of 30,000 full moons.

The Main Finding: The Silence of the Dark
When they looked for the dance, the music was silent. In a strict, model-independent test (a "null result"), they found no significant evidence that the gamma rays and galaxy shapes are connected. The data is consistent with there being no signal at all.

Because they didn't find the signal, they were able to draw a very strict "Do Not Enter" line on a map of dark matter possibilities. They explicitly ruled out certain scenarios:

They excluded the idea that "wino-like" dark matter particles (a specific type of heavy particle) with a mass between 2 and 3 TeV are annihilating at an enhanced rate, even if we assume a moderate boost from substructures in our galaxy.
If we assume the dark matter clumps together even more intensely (a boost factor of about 100), they ruled out the "standard" thermal annihilation rate (a value of 3 × 10⁻²⁶ cm³/s) for lighter dark matter particles (between 7 and 40 GeV) that turn into bottom quarks or tau leptons.

The "Maybe" Signal: A Ghost in the Machine?
Here is where it gets tricky. While the strict test said "nothing here," the authors tried a different approach. They used a "template" method, which is like asking, "If the signal looked exactly like this specific shape we predicted, would we see it?"

When they used this template method, they found a signal that looked like a 3.9σ detection (which is a statistical way of saying it's pretty unlikely to be a fluke, but not quite a "gold medal" discovery). However, the authors are very careful to note that this signal depends entirely on the shape of the template they chose.

A "power-law" template (a specific mathematical curve) fit the data well.
A "log-parabola" template did not.
The signal seemed to get stronger as they looked at galaxies further away, which makes sense if it's coming from the large-scale structure of the universe.

But because the strict, model-independent test came up empty, the authors conclude that this "detection" is likely just a statistical fluctuation or a quirk of the specific model used, rather than a confirmed discovery of dark matter. They also noticed that the signal's energy behavior didn't perfectly match what you'd expect from the average brightness of the gamma-ray background around 100 GeV, adding another layer of doubt.

The Bottom Line
The paper doesn't claim to have solved the mystery of dark matter. Instead, it acts as a very strong filter. It says, "We looked very hard over a huge area, and we didn't find the smoking gun." This allows scientists to confidently cross off certain heavy dark matter candidates and certain levels of clumpiness from their list of suspects.

They also looked at the possibility of dark matter decaying (fall apart) rather than annihilating. Based on their data, they set a lower limit on how long a dark matter particle must live before it decays: it has to be at least 10²⁶ to 10²⁷ seconds long. That's a really, really long time—far longer than the age of the universe.

In short, the universe is still keeping its dark matter secrets. The "dance" between the gamma rays and the galaxy shapes wasn't found in the strictest test, but the search continues, and the rules of the game have just gotten a little tighter.

Technical Summary: Constraints on Annihilating Dark Matter from Gamma-Ray Background–Galaxy Shape Correlations

Problem and Motivation
The nature of cosmic dark matter (DM) remains a primary mystery in modern physics. Weakly Interacting Massive Particles (WIMPs) are a leading candidate, predicting a spatial cross-correlation between the products of DM annihilation and the large-scale structure (LSS) of the universe. While the unresolved extragalactic $\gamma$ -ray background (IGRB) is expected to contain a component from DM annihilation, disentangling this signal from astrophysical sources (e.g., active galactic nuclei, star-forming galaxies) is challenging. Previous observational searches have yielded mixed results: early studies reported null detections due to limited sky coverage, while a recent analysis using Dark Energy Survey (DES) Year 1 data and 9 years of Fermi-LAT data reported a detection with a signal-to-noise ratio of 8.9 using a matched-filter approach. However, this detection relied on specific model assumptions. This paper revisits the cross-correlation between the IGRB and galaxy shapes (cosmic shear) to provide a more robust, model-independent constraint on DM annihilation, utilizing significantly larger datasets and a Fourier-space analysis framework.

Methodology
The authors analyze a common sky area of approximately $12,000 \text{ deg}^2$ , combining:

$\gamma$ -ray Data: 14 years of Fermi Large Area Telescope (LAT) Pass8 observations (August 2008–August 2022). The data covers energies from 100 MeV to 1 TeV, binned into 40 logarithmic micro-bins and subsequently grouped into 9 macro energy bins (0.63–1000 GeV). The unresolved IGRB is extracted by subtracting Galactic foregrounds and resolved sources (using the 4FGL-DR4 catalog) from the masked high-latitude region ( $|b| \ge 30^\circ$ ).
Galaxy Shape Data: Weak lensing shear catalogs from DES Year 3 (Y3) and the DECam All Data Everywhere (DECADE) project. These datasets contain $\sim 100$ million and $\sim 170$ million galaxies, respectively. The catalogs are split into four tomographic redshift bins (mean redshifts $\bar{z} \approx 0.34, 0.53, 0.75, 0.95$ ).

Analysis Framework:

Fourier-Space Estimator: Unlike the previous DES Y3 analysis which used real-space correlations, this study employs a Fourier-space estimator for the cross-power spectrum ( $C_{\gamma E}$ ) between the IGRB intensity and the E-mode cosmic shear. This approach allows for a controlled quantification of statistical uncertainties and systematic effects.
Power Spectrum Measurement: Using the pseudo- $C_\ell$ algorithm (NaMaster), the authors compute cross-power spectra for all combinations of 9 energy bins and 12 shape catalogs (4 redshift bins $\times$ 3 survey windows), totaling 108 power spectra. The analysis is restricted to multipoles $30 < \ell < 512$ to avoid pixelization artifacts.
Models Tested:
- Model-Independent: A $\chi^2$ test comparing the measured power spectra against a null hypothesis (no correlation).
- DM Annihilation Models: Theoretical predictions for WIMP annihilation into $b\bar{b}$ , $\tau^+\tau^-$ , $\mu^+\mu^-$ , and $W^+W^-$ channels. These models include primary $\gamma$ -rays and secondary emission from inverse-Compton (IC) scattering of upscattered background photons. Substructure boost factors ( $b_{sh}$ ) of 3 (LOW), 30 (MID), and 100 (HIGH) are considered.
- Phenomenological Templates: Two models from Ref. [11] (a power-law and a log-parabola parameterization) are used to test consistency with previous claims of detection.

Key Results

Null Detection in Model-Independent Test: The primary result is a null detection of the cross-correlation signal. The signal-to-noise ratio ( $S/N$ ) per degree of freedom is consistent with zero across all survey regions (DES Y3, DECADE NGC, DECADE SGC). The stacked B-modes (indicative of systematics) are also consistent with null, suggesting controlled systematic errors.
Constraints on Annihilation Cross Section: Based on the null result, the authors derive 95% confidence upper limits on the DM annihilation cross section $\langle \sigma v \rangle$ $⟨ σ v ⟩$ for DM masses between 7 GeV and 10 TeV.
- For a moderate substructure boost factor of $\sim 30$ , the analysis excludes an enhanced annihilation cross section for wino-like DM with masses of 2–3 TeV.
- For a higher boost factor of $\sim 100$ , the constraints become significantly stronger, excluding the canonical thermal relic cross section ( $\langle \sigma v \rangle = 3 \times 10^{-26} \text{ cm}^3/\text{s}$ ) for 7–40 GeV DM annihilating into $b\bar{b}$ or $\tau^+\tau^-$ .
- The inclusion of secondary IC emission significantly tightens these constraints, particularly for leptonic channels and heavier DM masses.
Decaying Dark Matter: The study also derives 2 $\sigma$ lower limits on the DM particle lifetime for decaying scenarios, ranging from $\sim 10^{26}$ to $10^{27}$ seconds, depending on the decay channel.
Template-Based Analysis: When applying the phenomenological templates from Ref. [11], the data shows a preference for the power-law model at the $\sim 3.9\sigma$ level ( $q = 0.62 \pm 0.16$ ), while the log-parabola model is disfavored. A matched-filter analysis assuming a simple $\ell^{-1}$ scaling yields a $p$ -value of $\sim 0.01$ (corresponding to a total $S/N \approx 7.6$ ). However, the authors note that the inferred significance is highly dependent on the assumed signal model, and the power-law fit shows deviations from the expected mean intensity scaling around 100 GeV.

Significance and Claims
The paper claims to provide the most stringent cosmological constraints on DM annihilation to date using cross-correlations between $\gamma$ -rays and galaxy shapes, leveraging the largest sky coverage ( $\sim 12,000 \text{ deg}^2$ ) achieved in such studies. The primary significance lies in the model-independent null result, which challenges the previous detection claims that relied heavily on specific halo-model templates. The authors demonstrate that while template-based analyses can yield signals at the $\sim 3\sigma$ – $4\sigma$ level, these signals are not robust when tested against a null hypothesis without prior assumptions.

The study concludes that cosmological cross-correlations offer constraints comparable to and complementary with local $\gamma$ -ray probes (e.g., dwarf spheroidal galaxies). The results specifically rule out certain parameter spaces for wino-like DM and thermal relic DM under high boost factor scenarios. The authors emphasize that the tension between the null model-independent result and the moderate template-based signal highlights the necessity for future joint analyses with multiple large-scale structure tracers to disentangle astrophysical and dark matter contributions through their environmental dependencies.

Technical Summary: Constraints on Annihilating Dark Matter from Gamma-Ray Background–Galaxy Shape Correlations

More like this