Viability of $Δm^2\sim$ 1 eV$^2$ sterile neutrino mixing models in light of MiniBooNE electron neutrino and antineutrino data from the Booster and NuMI beamlines

TL;DR

The paper re-evaluates short-baseline sterile-neutrino models in (3+1) and (3+2) frameworks using the latest MiniBooNE ν, ν̄, and NuMI data alongside LSND, KARMEN, NOMAD, Bugey, CHOOZ, CCFR84, CDHS, and atmospheric constraints. It employs a (3+n) oscillation formalism with CP phases (notably φ_{45} in the 3+2 case) and a Monte Carlo χ^2-based fitting approach with a PG test to assess cross-dataset compatibility. The results show that while (3+2) CP-violating fits better describe the full SBL data than (3+1), substantial tension remains between appearance and disappearance data, and between neutrino and antineutrino channels, with the main drivers being MBν, CDHS, and atmospheric constraints; CPT-conserving interpretations struggle to reconcile all results, suggesting the need to explore CPT-violating or alternative new-physics scenarios. The NuMI-MB data currently provide weak constraints due to systematics, but future updates and complementary disappearance measurements from MiniBooNE and MINOS are expected to sharpen the picture. Overall, the study reinforces that a single CPT-conserving sterile-neutrino framework is insufficient to harmonize all short-baseline oscillation signals, motivating exploration of more complex or CPT-violating dynamics.

Abstract

This paper examines sterile neutrino oscillation models in light of recently published results from the MiniBooNE Experiment. The new MiniBooNE data include the updated neutrino results, including the low energy region, and the first antineutrino results, as well as first results from the off-axis NuMI beam observed in the MiniBooNE detector. These new global fits also include data from LSND, KARMEN, NOMAD, Bugey, CHOOZ, CCFR84, and CDHS. Constraints from atmospheric oscillation data have been imposed.

Figures (13)

• Figure 1: Allowed regions (filled areas) at 90% and 99% CL from BNB-MB($\nu$)-only, BNB-MB($\bar{\nu}$)-only, and LSND-only (3+1) fits. These fits are, by construction, CP-conserving. The stars indicate the three respective best-fit points. All three data sets show closed contours at 90% CL. See text for more details.
• Figure 2: Left: Null and (3+1) best-fit predicted event distributions ($\Delta m_{41}^2,\sin^2(2\theta_{\mu e})$) $=$ (3.12, 0.0018) for BNB-MB($\nu$). Right: Null and (3+1) best-fit predicted event distributions ($\Delta m_{41}^2,\sin^2(2\theta_{\mu e})$) $=$ (4.46, 0.0065) for BNB-MB($\bar{\nu}$). The event distributions are shown as functions of reconstructed neutrino energy, $E^{QE}_{\nu}$. The data are shown in black points with statistical uncertainty. The null (no-oscillation) prediction is shown by the light gray histogram with (solid) systematic error band. The best-fit prediction (signal and background) is shown by the blue (dark gray) histogram with (shaded) systematic error band.
• Figure 3: Left: Allowed 90% and 99% CL regions (light and dark filled areas, respectively) from a combined analysis of BNB-MB($\nu$), BNB-MB($\bar{\nu}$) and LSND data sets, and 90% and 99% exclusion limits (light and dark curves, respectively) from each of the null appearance experiments, NUMI-MB (solid curves), KARMEN (dashed curves) and NOMAD (dotted curves). Middle: The same allowed region with overlayed 90% and 99% exclusion limits from a combined analysis of all null appearance experiments. Right: Allowed region obtained by a combined analysis of all appearance data sets, signal and null. See text for more details.
• Figure 4: Left: The allowed 90% and 99% CL regions (light and dark filled areas, respectively) from a combined analysis of BNB-MB($\bar{\nu}$) and LSND data sets, and 90% and 99% exclusion limits (light and dark curves, respectively) from KARMEN. A comparison of only these three experiments is interesting, as these three experiments have searched for antineutrino oscillations at short baselines. Right: The allowed regions obtained from a combined analysis of all three experiments (BNB-MB($\bar{\nu}$), LSND, and KARMEN). See text for more details.
• Figure 5: Null and (3+1) best-fit predicted event distributions ($\Delta m_{41}^2,\sin^2(2\theta_{\mu e})$) $=$ (7.36, 0.019) for NUMI-MB. The data are shown in black points with statistical uncertainty. The null (no-oscillation) prediction is shown by the light gray histogram with (solid) systematic error band. The best-fit prediction (signal and background) is shown by the blue (dark gray) histogram with (shaded) systematic error band.