Abstract
We explore systematic biases in the identification of dark matter in future direct detection experiments and compare the reconstructed dark matter properties when assuming a self-consistent dark matter distribution function and the standard Maxwellian velocity distribution. We find that the systematic bias on the dark matter mass and cross-section determination arising from wrong assumptions for its distribution function is of order ~1\sigma. A much larger systematic bias can arise if wrong assumptions are made on the underlying Milky Way mass model. However, in both cases the bias is substantially mitigated by marginalizing over galactic model parameters. We additionally show that the velocity distribution can be reconstructed in an unbiased manner for typical dark matter parameters. Our results highlight both the robustness of the dark matter mass and cross-section determination using the standard Maxwellian velocity distribution and the importance of accounting for astrophysical uncertainties in a statistically consistent fashion.
Introduction
Direct detection experiments aim to identify Dark Matter (DM) by observing the nuclear recoils produced when DM particles scatter off target atoms in terrestrial detectors. To extract fundamental particle properties—specifically the DM mass ($m_\chi$) and its scattering cross-section ($\sigma_p^{SI}$)—from these signals, researchers must rely on assumptions about the local galactic environment. Specifically, the local DM density ($\rho_0$) and the velocity distribution ($f(w)$) are critical inputs. The Standard Halo Model (SHM) typically assumes a simple Maxwell-Boltzmann distribution, but real-world galactic dynamics are far more complex. This study investigates how incorrect astrophysical assumptions bias DM property reconstruction and how these biases can be mitigated.
Methodology and Modeling
The researchers employed a Bayesian inference framework using the MultiNest algorithm to analyze mock data for future ton-scale experiments (Xenon, Germanium, and Neon). They compared the standard Maxwellian distribution (SHM) against a physically rigorous Self-Consistent Distribution Function (SCDF) derived from the Milky Way's gravitational potential.
To test the impact of varying levels of knowledge, three modeling setups were used:
- Fixed: Astrophysical parameters are held constant at fiducial values.
- Pheno (Phenomenological): Local parameters like circular velocity ($v_0$) and escape velocity ($v_{esc}$) are allowed to vary.
- Full: The entire underlying mass model of the Milky Way, including disk and halo parameters, is varied within prior ranges.
Key Insights and Results
The analysis revealed that the Maxwellian assumption is remarkably robust; assuming the SHM when the true distribution is SCDF only leads to a bias of approximately $1\sigma$. However, the study identified a more significant threat: rigid assumptions about the underlying galactic mass model. Using a bias factor $\alpha$ (where $\alpha > 1$ indicates a biased result), the authors found that an incorrect mass model could lead to $\alpha > 4$.
Crucially, the systematic bias is substantially mitigated through marginalization. By treating astrophysical parameters as "nuisance parameters" and integrating over their uncertain ranges, the bias factor dropped to $\alpha < 1$ for most benchmarks. While this increases the final uncertainty (error bars), it provides a statistically honest and accurate reconstruction of the DM particle properties.
Conclusion
This work demonstrates that while astrophysical uncertainties pose a challenge to direct DM detection, they are not an insurmountable barrier. By adopting a self-consistent modeling approach and marginalizing over galactic parameters, future experiments can reliably extract DM properties. The results underscore the importance of integrating astronomical data and dark matter searches into a unified statistical framework to ensure robust discoveries.
Summary updated