Search for D0 -> mu+ mu- Decays With 360 pb-1

Search for the Rare Decay D⁰&rarr&mu⁺&mu^-

This web page summarizes the results for the search for D⁰ &rarr &mu⁺ &mu^- decays using 360 pb^-1 of data. This measurement supersedes the previous CDF result which was based on an integrated luminosity of 69 pb^-1. In addition to increasing the integrated luminosity of the data sample used in the measurement, we expand the coverage of the muon system and utilize a strict muon identification likelihood to further reduce combinatorial background. We find that the dominant remaining background is due to real B &rarr &mu⁺&mu^- X decays. We further reduce this background using a likelihood ratio based on the displacement and pointing of the two-track system. The combined analysis improvements result in an improvement of the analysis sensitivity by a factor of almost 7. We observe no significant excess over the predicted background in the signal search region. Using a Bayesian approach, we set the upper limits on the branching fraction B(D⁰ &rarr &mu⁺&mu^-) < 5.3 x 10^-7 at 95% CL and B(D⁰ &rarr &mu⁺&mu^-) < 4.3 x 10^-7 at 90% CL. This is an improvement of roughly a factor 6 over the previous CDF result, and a factor of 3 over the current world best result from BaBar.

These results have been approved in the CDF B Group Meeting on February 28, 2008.

A more detailed summary of the results can be found in this public note.

Approved Figures and Numbers:

Feyman Diagrams for D⁰&rarr&mu⁺&mu^- decays:

Short distance interaction [eps] [jpeg]

Long distance interaction [eps] [jpeg]

SUSY contribution diagram [eps] [jpeg]

Reference sample D⁰ &rarr K^- &pi⁺ illustrates the amount of charm decays we extract from our datasets. The selection requirement that strongly reduces combinatorial background under the peak is the tight requirement on the mass difference m(D^*0) - m(D⁰).

D⁰ &rarr K^-&pi⁺ mass distribution [eps] [jpeg]

Mass difference distribution m(D^*+) - m(D⁰) [eps] [jpeg]

D⁰ &rarr K^-&pi⁺ mass distribution after mass difference cut [eps] [jpeg]

Reference sample D⁰ &rarr &pi⁺ &pi^- is the sample to which we normalize the result of our search. The selection requirement on the mass difference m(D^*0) - m(D⁰) is already applied; events are reconstructed under the &mu⁺&mu^- mass hypothesis. No muon identification is required on either track.

CMU-CMU [eps] [jpeg]

CMU-CMX [eps] [jpeg]

CMX-CMX [eps] [jpeg]

Muon Misidentification Rates as a function of momentum, pseudorapidity and azimuthal angle, for pions and kaons in the CMU and CMX subdetectors.

Pion Misidentification Rate as a function of Pt, CMU subdetector [eps] [jpeg]	Pion Misidentification Rate as a function of Eta, CMU subdetector [eps] [jpeg]	Pion Misidentification Rate as a function of Phi, CMU subdetector [eps] [jpeg]
Kaon Misidentification Rate as a function of Pt, CMU subdetector [eps] [jpeg]	Kaon Misidentification Rate as a function of Eta, CMU [eps] [jpeg]	Kaon Misidentification Rate as a function of Phi, CMU subdetector [eps] [jpeg]
Pion Misidentification Rate as a function of Pt, CMX subdetector [eps] [jpeg]	Pion Misidentification Rate as a function of Eta, CMX subdetector [eps] [jpeg]	Pion Misidentification Rate as a function of Phi, CMX subdetector [eps] [jpeg]
Kaon Misidentification Rate as a function of Pt, CMX subdetector [eps] [jpeg]	Kaon Misidentification Rate as a function of Eta, CMX subdetector [eps] [jpeg]	Kaon Misidentification Rate as a function of Phi, CMX subdetector [eps] [jpeg]

Expected background composition in a sample with one leg tagged as a muon:

Predicted background levels for different background components in a sample with only one decay leg tagged as a muon [eps] [jpeg]

Removing the B &rarr &mu⁺&mu^- Background: We find that the dominant source of background after muon tagging are real dimuons from B meson decays. These are removed using a likelihood ratio based on the displacement and pointing of the two-track pair

CMU-CMU [eps] [jpeg]

CMU-CMX [eps] [jpeg]

Dimuon invariant mass distributions after full selection is applied. Overlaid are background estimates from different sources. The signal search window is also shown, along with the corresponding event count after unblinding (blue) and predicted background (red). The dotted black line is a fit of the background rate assuming a flat background distribution in the high-mass sideband (simple crosscheck).

CMU-CMU [eps] [gif]

CMU-CMX [eps] [gif]

CMX-CMX [eps] [gif]

Summary table of expected background and observed events:

Detector	CMU-CMU	CMU-CMX	CMX-CMX
Combinatorial Background	0.040 ± 0.007	0.008 ± 0.001	0.0007 ± 0.0001
D⁰ &rarr &pi&pi Double Tags	0.530 ± 0.005	0.057 ± 0.001	0.012 ± 0.002
D⁰ &rarr K&pi Double Tags	< 0.01	< 0.01	< 0.01
Semileptonic D⁰ Decays	< 0.36	< 0.20	< 0.10
B Decays Involving One Real Muon	0.54 ± 0.06	0.13 ± 0.03	0.07 ± 0.02
B Decays Involving Two Real Muons	3.8 ± 1.3	2.5 ± 1.0	1.0 ± 0.5
Total Expected Background	4.9 ± 1.3	2.7 ± 1.0	1.0 ± 0.5
Observed Events	3	0	1

Statistical analysis: we utilize a Bayesian approach to set 90% and 95% confidence limits. The plots below show the distribution of limits which would be set in the absence of signal, for 1000 pseudo-experiments simulating the current measurement configuration. The median limit represents the sensitivity of the experiment (green arrow). The probability to observe limits as good as, or better than those seen by our measurement, is 15% (red histogram).

90% CL Limits [eps] [gif]

95% CL Limits [eps] [gif]

Result: We observe no excess of events in the search window. Using a Bayesian approach we derive the following upper limits on the branching fraction:

B(D⁰ &rarr &mu⁺&mu^-) < 5.3 x 10^-7 at 95% CL
B(D⁰ &rarr &mu⁺&mu^-) < 4.3 x 10^-7 at 90% CL

this translates into a limit on the R parity violating couplings

&lambda_21k&lambda_22k = 1.5 &radic B(D⁰ &rarr &mu⁺&mu^-) < 9.8 × 10^-4