TL;DR
The BABAR experiment searched for the rare B meson decays $B \to K\nu\bar{\nu}$, which are highly suppressed in the Standard Model but sensitive to new physics. By developing a sophisticated multivariate analysis technique to overcome the challenge of identifying neutrinos, they set new upper limits on the branching fractions for these decays, constraining new physics scenarios.
What Problem Does This Paper Address?
The Standard Model (SM) of particle physics describes fundamental particles and their interactions. However, it is known to be incomplete, and physicists are actively searching for phenomena that deviate from its predictions. Rare decays of B mesons, which are produced in high-energy particle collisions, offer a powerful window into potential new physics. Specifically, the decays $B \to K\nu\bar{\nu}$ (where $\nu\bar{\nu}$ represents a neutrino-antineutrino pair) are flavor-changing neutral-current processes. These are forbidden at the most basic level (tree level) in the SM and can only occur through higher-order quantum loop diagrams. While the SM predicts very small branching fractions for these decays (on the order of $10^{-6}$), they are particularly sensitive to contributions from hypothetical new particles or forces that could also manifest at the loop level. Observing these decays or setting stringent limits on their rates can therefore provide crucial information about physics beyond the SM. The primary challenge in detecting these decays is the presence of two neutrinos in the final state, which are invisible to the detector, making it difficult to distinguish signal events from background.
Methods
To tackle the challenge of detecting the invisible neutrinos, the BABAR collaboration employed a sophisticated "tagging" strategy. They identified one B meson in the event through its decay into a $D^{(*)}l\nu$ final state (where $l$ is an electron or muon). This "reconstructed" B meson ($B_{rec}$) provides a clean signature. The remaining particles in the event, referred to as the "rest of the event" ($B_{roe}$), are then analyzed for the presence of the signal decay $B \to K\nu\bar{\nu}$.
A key innovation in this analysis is the use of a multivariate analysis technique, specifically a bagged decision tree (BDT). This algorithm leverages a large number of discriminating variables (up to 38 for the $B^0$ channel) that capture subtle differences between signal and background events. These variables include information about the missing energy and momentum (due to the neutrinos), overall event properties, signal kinematics, and reconstruction quality. To ensure robustness and minimize statistical fluctuations, the analysis used an ensemble of 20 BDTs, each trained on different random subsets of the simulated signal and background data. This approach allows for a more precise estimation of the background and a more reliable determination of the signal efficiency. The analysis also involved careful control of systematic uncertainties, including those related to signal efficiency, background prediction, and data-MC discrepancies.
Datasets
The analysis utilized a large dataset of approximately 459 million $B\bar{B}$ pairs collected by the BABAR detector at the SLAC PEP-II asymmetric-energy B Factory. This dataset corresponds to an integrated luminosity of about 418 fb$^{-1}$. Simulated Monte Carlo (MC) samples were extensively used for training the BDTs and estimating signal efficiencies and background contributions. These MC samples included simulated signal events and various background processes, such as non-resonant $e^+e^- \to q\bar{q}$, $B\bar{B}$, and $\tau^+\tau^-$ events, as well as simulated $B\bar{B}$ doubly semileptonic events.
Key Insights and Ideas
- The neutrino invisibility problem is a major hurdle: The core difficulty in searching for $B \to K\nu\bar{\nu}$ is the lack of direct detection of the neutrinos. This necessitates indirect methods like the $B_{rec}$ tagging strategy.
- Multivariate analysis is crucial for rare decay searches: The BDT approach, by combining many weak discriminants, significantly enhances the ability to separate signal from background compared to traditional cut-based analyses.
- Ensemble BDTs improve statistical precision: Using an ensemble of BDTs, rather than a single one, helps to average out statistical fluctuations and provides a more robust estimate of signal and background.
- Sensitivity to New Physics: The stringent limits set on these decays are important because they directly constrain theoretical models of new physics that predict enhanced branching fractions for these FCNC processes.
Key Results
The BABAR experiment set the following 90% confidence level upper limits on the branching fractions:
- For $B^+ \to K^+\nu\bar{\nu}$: $1.3 \times 10^{-5}$
- For $B^0 \to K^0\nu\bar{\nu}$: $5.6 \times 10^{-5}$
These limits are competitive with or improve upon previous measurements and are a few times larger than the SM predictions, thus providing significant constraints on new physics scenarios. The paper also reports upper limits on partial branching fractions for $B^+ \to K^+\nu\bar{\nu}$ in two ranges of di-neutrino invariant mass squared ($q^2$).
Limitations and Future Work
The authors acknowledge systematic uncertainties that limit the precision of their results. These include uncertainties in signal efficiency (14%), background predictions (5% for $K^+$, 25% for high-$q^2 K^+$, and 10% for $K^0$), and data-MC discrepancies. While the analysis is blind to signal events until the methodology is finalized, the statistical uncertainties on the number of excess events are still significant, particularly for the $K^0$ channel.
Future work could involve further refinement of the BDT variables and training, as well as exploring alternative tagging strategies or analysis techniques to improve sensitivity. With the advent of future B factories or other high-luminosity experiments, more precise measurements of these rare decays could provide even stronger tests of the Standard Model and probe new physics at higher energy scales.
Summary updated