Evidence for dark matter modulation in CoGeNT?

Chiara Arina, Jan Hamann, Roberto Trotta, Yvonne Y. Y. WongJournal of Cosmology and Astroparticle Physics 2012 · DOI · arXiv

Abstract

This study investigates whether the annual modulation signal reported by the CoGeNT experiment is best explained by dark matter (DM) using a Bayesian model comparison framework. We evaluate five phenomenological models, ranging from a null hypothesis (no modulation) to DM models with varying degrees of physical constraints. While DM models are only weakly preferred over the null hypothesis, they are strongly favored (odds up to 560:1) over generic "other physics" models with free phases or periods. These results remain robust under astrophysical uncertainties. Furthermore, we find that incorporating anisotropic velocity distributions into the DM halo model may help reconcile the tension between CoGeNT's total and modulated event rates.

Introduction: The CoGeNT Controversy

The search for Dark Matter (DM) often hinges on identifying an annual modulation in detection rates—a "smoking gun" signature caused by Earth's orbital motion through the galactic DM halo. The CoGeNT experiment has reported such a signal, but its interpretation remains contentious. The observed modulation amplitude is higher than expected relative to the total event rate, and the phase is slightly shifted from standard predictions. This research applies a rigorous Bayesian lens to determine if these data truly support a DM origin or if they are better explained by statistical noise or unknown periodic backgrounds.

Methodology: Bayesian Model Comparison

Rather than relying solely on frequentist p-values, this study employs Bayesian model comparison. This method evaluates the "evidence" (marginal likelihood) for a model, naturally penalizing excessive complexity—a principle known as Occam’s Razor. We compare five scenarios:

  1. Model 0: No modulation (Null hypothesis).
  2. Model 1a (Pheno-DM): DM with fixed period (1 year) and phase (June 2nd), but free amplitude.
  3. Model 1b (Consistent DM): DM model constrained by the total CoGeNT event rate.
  4. Model 2a: Generic periodic signal with a 1-year period but free phase.
  5. Model 2b: Generic periodic signal with both free period and free phase.

To interpret the resulting Bayes factors, we utilize a standard scale of evidence strength to categorize the results from "inconclusive" to "strong."

Table of evidence strength for Bayesian model comparison based on the log-Bayes factor.

Data and Energy Dependence

The analysis utilizes 458 days of CoGeNT time-stamped data, binned into three energy ranges: $\Delta E_1$ (0.5–0.9 keVee), $\Delta E_2$ (0.9–3.0 keVee), and $\Delta E_3$ (3.0–4.5 keVee). A key prediction of WIMP-based dark matter is that the modulation should be most prominent at low energies and disappear at higher energies.

CoGeNT event rates over time across three energy bins, showing fits for various modulation models.

As seen in the figure, the modulation is clearly visible in the lower energy bins ($\Delta E_1$ and $\Delta E_2$) but effectively vanishes in the highest bin ($\Delta E_3$), aligning with theoretical expectations for light WIMPs.

Key Results and Insights

  • DM vs. Noise: The DM models are preferred over the "no modulation" null hypothesis with odds of approximately 8:1. According to the evidence scale, this constitutes "weak" evidence.
  • DM vs. Generic Physics: The most striking result is the comparison between DM and models with free phases (2a/2b). The DM hypothesis is favored with overwhelming odds ranging from 185:1 to 560:1. This is because the DM model's fixed phase and period make it highly predictive; generic models are penalized for their extra degrees of freedom.
  • Resolving Tensions with Anisotropy: The study explores whether the standard assumption of a spherical dark matter halo is to blame for the high observed modulation amplitude. By introducing an anisotropic (triaxial) halo, the predicted modulation increases, potentially resolving the discrepancy between the total rate and the modulated signal.

Conclusion

While the current evidence for dark matter in CoGeNT is not yet "definitive" by strict Bayesian standards (falling into the "weak" category against the null hypothesis), it is significantly more robust than any generic periodic explanation. The results suggest that the "tension" in CoGeNT data might not be a failure of the DM hypothesis, but rather a hint that our local dark matter environment is more complex and anisotropic than previously assumed. Future work focusing on non-standard velocity distributions will be critical in finalizing this interpretation.

Summary updated

Abstract

We investigate the question of whether the recent modulation signal claimed by CoGeNT is best explained by the dark matter (DM) hypothesis from a Bayesian model comparison perspective. We consider five phenomenological explanations for the data: no modulation signal, modulation due to DM, modulation due to DM compatible with the total CoGeNT rate, and a signal coming from other physics with a free phase but annual period, or with a free phase and a free period. In each scenario, we assign to the free parameters physically motivated priors. We find that while the DM models are weakly preferred to the no modulation model, but when compared to models where the modulation is due to other physics, the DM hypothesis is favoured with odds ranging from 185:1 to 560:1. This result is robust even when astrophysical uncertainties are taken into account and the impact of priors assessed. Interestingly, the odds for the DM model in which the modulation signal is compatible with the total rate against a DM model in which this prior is not implemented is only 5:8, in spite of the former's prediction of a modulation amplitude in the energy range 0.9 3.0 keVee that is significantly smaller than the value observed by CoGeNT. Classical hypothesis testing also rules out the null hypothesis of no modulation at the 1.6σ to 2.3σ level, depending on the details of the alternative. Lastly, we investigate whether anisotropic velocity distributions can help to mitigate the tension between the CoGeNT total and modulated rates, and find encouraging results.
BibTeX
@article{arina2012evidence,
  author = {Chiara Arina and Jan Hamann and Roberto Trotta and Yvonne Y. Y. Wong},
  title = {{Evidence for dark matter modulation in CoGeNT?}},
  journal = {Journal of Cosmology and Astroparticle Physics},
  year = {2012},
  doi = {10.1088/1475-7516/2012/03/008},
}
Figure