CMS-PAS-BPH-25-005

CMS-PAS-BPH-25-005
Search for the rare decay $ {\rm K}^0_\mathrm{S}\to\mu^+\mu^- $ in proton-proton collisions at $ \sqrt{s}= $ 13.6 TeV
CMS Collaboration
2026-07-23

Abstract: Rare kaon decays have long been a subject of significant interest, with most previous studies conducted using fixed-target experiments as well as B-factories. The implementation of a new trigger strategy during LHC Run 3 enables the CMS experiment to contribute to this area of research by exploiting the unprecedented luminosity of the LHC. This study reports on a search for the rare decay of $ {\rm K}^0_\mathrm{S} $ mesons into two muons using proton-proton collision data at $ \sqrt{s}= $ 13.6 TeV collected by the CMS experiment during the 2022 to 2025 operation of the CERN LHC, with an integrated luminosity of 287 fb$ ^{-1} $. No significant excess is observed. A limit on the branching fraction of $ \mathcal{B}({\rm K}^0_\mathrm{S}\to\mu^+\mu^-) < 4.6 (5.5)\times10^{-10} $ at 90% (95%) confidence level is set.
Links: CDS record (PDF) ; CADI line (restricted) ;

Figures & Tables	Summary	References	CMS Publications

Figures
png pdf	Figure 1: The distributions of the dimuon mass $ m_{\mu\mu} $ for 2022--2025 data sample after the baseline selection.
png pdf	Figure 2: The distributions of the dimuon mass $ m_{\mu\mu} $ for 2022--2024 data sample before and after BDT selection. The orange and grey shaded areas represent the signal and background samples used for DNN training, respectively.
png pdf	Figure 3: The distributions of the dipion mass $ m_{\pi\pi} $ for the $ \mathrm{K^0_S}\to\pi^+\pi^- $ normalization channel using data collected from 2022 to 2025, along with the total fit (solid curve), the $ \mathrm{K^0_S}\to\pi^+\pi^- $ contribution (hatched area), and the background contributions (dashed curve).
png pdf	Figure 3-a: The distributions of the dipion mass $ m_{\pi\pi} $ for the $ \mathrm{K^0_S}\to\pi^+\pi^- $ normalization channel using data collected from 2022 to 2025, along with the total fit (solid curve), the $ \mathrm{K^0_S}\to\pi^+\pi^- $ contribution (hatched area), and the background contributions (dashed curve).
png pdf	Figure 3-b: The distributions of the dipion mass $ m_{\pi\pi} $ for the $ \mathrm{K^0_S}\to\pi^+\pi^- $ normalization channel using data collected from 2022 to 2025, along with the total fit (solid curve), the $ \mathrm{K^0_S}\to\pi^+\pi^- $ contribution (hatched area), and the background contributions (dashed curve).
png pdf	Figure 3-c: The distributions of the dipion mass $ m_{\pi\pi} $ for the $ \mathrm{K^0_S}\to\pi^+\pi^- $ normalization channel using data collected from 2022 to 2025, along with the total fit (solid curve), the $ \mathrm{K^0_S}\to\pi^+\pi^- $ contribution (hatched area), and the background contributions (dashed curve).
png pdf	Figure 3-d: The distributions of the dipion mass $ m_{\pi\pi} $ for the $ \mathrm{K^0_S}\to\pi^+\pi^- $ normalization channel using data collected from 2022 to 2025, along with the total fit (solid curve), the $ \mathrm{K^0_S}\to\pi^+\pi^- $ contribution (hatched area), and the background contributions (dashed curve).
png pdf	Figure 4: The efficiency comparison between 2024 (left) and 2025 (right) to 2023 for dimuon and dipion final states as a function of transverse displacement $ L_{\mathrm{xy}} $. Smooth functions, implemented via cubic splines, are used to extract the correction values continuously as a function of $ L_{\mathrm{xy}} $.
png pdf	Figure 4-a: The efficiency comparison between 2024 (left) and 2025 (right) to 2023 for dimuon and dipion final states as a function of transverse displacement $ L_{\mathrm{xy}} $. Smooth functions, implemented via cubic splines, are used to extract the correction values continuously as a function of $ L_{\mathrm{xy}} $.
png pdf	Figure 4-b: The efficiency comparison between 2024 (left) and 2025 (right) to 2023 for dimuon and dipion final states as a function of transverse displacement $ L_{\mathrm{xy}} $. Smooth functions, implemented via cubic splines, are used to extract the correction values continuously as a function of $ L_{\mathrm{xy}} $.
png pdf	Figure 5: Left: the distribution of the vertices of $ \omega\to\mu^{+}\mu^{-} $ decays in the transverse plane on the beam pipe and CMS pixel layers from the data, using the $ _\mathrm{{s}}\mathcal{P}\mathrm{lot} $ technique [30]. Right: the $ L_{\mathrm{xy}} $ distributions for the $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ simulation before and after the MVA selection, normalized to unity based on the preselection distribution.
png pdf	Figure 5-a: Left: the distribution of the vertices of $ \omega\to\mu^{+}\mu^{-} $ decays in the transverse plane on the beam pipe and CMS pixel layers from the data, using the $ _\mathrm{{s}}\mathcal{P}\mathrm{lot} $ technique [30]. Right: the $ L_{\mathrm{xy}} $ distributions for the $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ simulation before and after the MVA selection, normalized to unity based on the preselection distribution.
png pdf	Figure 5-b: Left: the distribution of the vertices of $ \omega\to\mu^{+}\mu^{-} $ decays in the transverse plane on the beam pipe and CMS pixel layers from the data, using the $ _\mathrm{{s}}\mathcal{P}\mathrm{lot} $ technique [30]. Right: the $ L_{\mathrm{xy}} $ distributions for the $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ simulation before and after the MVA selection, normalized to unity based on the preselection distribution.
png pdf	Figure 6: The distributions of the dimuon mass $ m_{\mu\mu} $ for the signal channel in categories of $ \sigma_m < $ 0.005 GeV (left) and $ \sigma_m > $ 0.005 GeV (right), along with the associated projections of the full fit (solid curve), signal contribution (hatched area), and background contributions (other curves). The pull is defined as the difference between the data and the fit, divided by the statistical uncertainty in the data.
png pdf	Figure 6-a: The distributions of the dimuon mass $ m_{\mu\mu} $ for the signal channel in categories of $ \sigma_m < $ 0.005 GeV (left) and $ \sigma_m > $ 0.005 GeV (right), along with the associated projections of the full fit (solid curve), signal contribution (hatched area), and background contributions (other curves). The pull is defined as the difference between the data and the fit, divided by the statistical uncertainty in the data.
png pdf	Figure 6-b: The distributions of the dimuon mass $ m_{\mu\mu} $ for the signal channel in categories of $ \sigma_m < $ 0.005 GeV (left) and $ \sigma_m > $ 0.005 GeV (right), along with the associated projections of the full fit (solid curve), signal contribution (hatched area), and background contributions (other curves). The pull is defined as the difference between the data and the fit, divided by the statistical uncertainty in the data.
png pdf	Figure 7: The profile likelihood as a function of the $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ branching fraction
png pdf	Figure 8: The upper limits on the $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ decay branching fraction using the $ \text{CL}_\text{s} $ method. The dashed line represents the expected median value of the quantity 1 $ -\text{CL} $ for the background-only hypothesis, while the solid line shows the observed value.

Tables
png pdf	Table 1: The fitted yields of $ \mathrm{K^0_S}\to\pi^+\pi^- $ in zero bias datasets after the final selection. The columns ``Comb'', ``Lumi'', and ``Prescale'' denote the combinatorial background, the integrated luminosity, and the trigger prescale factor, respectively. The observed numbers of events are given in the Data column.
png pdf	Table 2: Summary of systematic uncertainties for the $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ branching ratio measurement.
png pdf	Table 3: The post-fit event yields for signal, the combinatorial background (``Comb''), and the $ \mathrm{K^0_S}\to\pi^+\pi^- $ background are summarized. The observed event yields are given in the Data column.

Summary

A search for $ \mathrm{K^0_S}\to\mu^{+}\mu^{-} $ decays by the CMS experiment, using proton-proton collision data at $ \sqrt{s} = $ 13.6 TeV corresponding to an integrated luminosity of 287 fb$ ^{-1} $, is presented. As one of the first CMS analyses to utilize the 2025 dataset, this work provides a deeper understanding of the Run 3 di-track reconstruction performance across both the 2024 and 2025 data-taking periods. A two-step multivariate analysis is performed, utilizing a boosted decision tree to preselect data samples that are then used to train a deep neural network to suppress the background by orders of magnitude. No significant excess above the fitted background is observed. An upper limit of $ {\mathcal B}(\mathrm{K^0_S}\to\mu^{+}\mu^{-}) < 4.6 (5.5) \times 10^{-10} $ is set at 90 (95)% confidence level. The resulting upper limit is about a factor of two higher than the world's most sensitive constraint established by the LHCb experiment [6], providing an independent probe of a highly suppressed flavor-changing neutral current process in the strange sector.

References
1	V. Chobanova et al.	Probing SUSY effects in $ k_s^0\rightarrow\mu^+\mu^- $	JHEP 05 (2018) 024	1711.11030
2	M. Endo et al.	Gluino-mediated electroweak penguin with flavor-violating trilinear couplings	JHEP 04 (2018) 019	1712.04959
3	C. Bobeth and A. J. Buras	Leptoquarks meet $ \varepsilon'/\varepsilon $ and rare kaon processes	JHEP 02 (2018) 101	1712.01295
4	G. D'Ambrosio and T. Kitahara	Direct $ cp $ violation in $ k \to \mu^+ \mu^- $	PRL 119 (2017) 201802	1707.06999
5	A. Dery, M. Ghosh, Y. Grossman, and S. Schacht	$ k\to\mu^{+}\mu^{-} $ as a clean probe of short-distance physics	JHEP 07 (2021) 103	2104.06427
6	LHCb Collaboration	Constraints on the $ k^0_s \rightarrow \mu^+ \mu^- $ branching fraction	PRL 125 (2020) 231801	2001.10354
7	Particle Data Group Collaboration	Review of particle physics	PRD 110 (2024) 030001
8	CMS Collaboration	The CMS experiment at the CERN LHC	JINST 3 (2008) S08004
9	CMS Collaboration	Development of the CMS detector for the CERN LHC Run 3	JINST 19 (2024) P05064	CMS-PRF-21-001 2309.05466
10	CMS Collaboration	Performance of the CMS Level-1 trigger in proton-proton collisions at $ \sqrt{s} = $ 13 TeV	JINST 15 (2020) P10017	CMS-TRG-17-001 2006.10165
11	CMS Collaboration	The CMS trigger system	JINST 12 (2017) P01020	CMS-TRG-12-001 1609.02366
12	CMS Collaboration	Performance of the CMS high-level trigger during LHC Run 2	JINST 19 (2024) P11021	CMS-TRG-19-001 2410.17038
13	CMS Collaboration	Electron and photon reconstruction and identification with the CMS experiment at the CERN LHC	JINST 16 (2021) P05014	CMS-EGM-17-001 2012.06888
14	CMS Collaboration	Performance of the CMS muon detector and muon reconstruction with proton-proton collisions at $ \sqrt{s} = $ 13 TeV	JINST 13 (2018) P06015	CMS-MUO-16-001 1804.04528
15	CMS Collaboration	Description and performance of track and primary-vertex reconstruction with the CMS tracker	JINST 9 (2014) P10009	CMS-TRK-11-001 1405.6569
16	CMS Tracker Group Collaboration	The CMS Phase-1 pixel detector upgrade	JINST 16 (2021) P02027	2012.14304
17	CMS Collaboration	Track impact parameter resolution for the full pseudo rapidity coverage in the 2017 dataset with the CMS phase-1 pixel detector	CMS Detector Performance Summary CMS-DP-2020-049, 2020 CDS
18	CMS Collaboration	Pixel auto-masking, occupancy, and cluster size versus $ \eta $ in 2025	CMS Detector Performance Summary CMS-DP-2026-016, 2026 CDS
19	T. Sjöstrand et al.	An introduction to PYTHIA 8.2	Comput. Phys. Commun. 191 (2015) 159	1410.3012
20	CMS Collaboration	Extraction and validation of a new set of CMS PYTHIA8 tunes from underlying-event measurements	EPJC 80 (2020) 4	CMS-GEN-17-001 1903.12179
21	GEANT4 Collaboration	GEANT 4---a simulation toolkit	NIM A 506 (2003) 250
22	D. J. Lange	The EvtGen particle decay simulation package	NIM A 462 (2001) 152
23	N. Davidson, T. Przedzinski, and Z. Was	PHOTOS interface in C++: technical and physics documentation	Comput. Phys. Commun. 199 (2016) 86	1011.0937
24	CMS Collaboration	Enriching the physics program of the CMS experiment via data scouting and data parking	Phys. Rep. 1115 (2025) 678	CMS-EXO-23-007 2403.16134
25	K. Prokofiev and T. Speer	A kinematic fit and a decay chain reconstruction library	Computing in High Energy and Nuclear Physics 200 (2005) 411
26	A. Blum, A. Kalai, and J. Langford	Beating the hold-out: bounds for k-fold and progressive cross-validation	in the Ann. Conf. on Computational Learning Theory, COLT '99, 1999 Proc. 1 (1999) 203
27	T. Chen and C. Guestrin	XGBoost: A scalable tree boosting system	in and ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD '16, ACM, New York, NY, USA, 2016 Proceedings of the 2 (2016) 785
28	D. Mart \'\i nez Santos and F. Dupertuis	Mass distributions marginalized over per-event errors	NIM A 764 (2014) 150	1312.5000
29	P. D. Dauncey, M. Kenzie, N. Wardle, and G. J. Davies	Handling uncertainties in background shapes: the discrete profiling method	JINST 10 (2015) P04015	1408.6865
30	M. Pivk and F. R. Le Diberder	$ _\mathrm{{s}}\mathcal{P}\mathrm{lot} $: A statistical tool to unfold data distributions	NIM A 555 (2005) 356	physics/0402083
31	CMS Collaboration	Performance of CMS muon reconstruction in pp collision events at $ \sqrt{s} = $ 7 TeV	JINST 7 (2012) P10002	CMS-MUO-10-004 1206.4071
32	CMS Collaboration	Muon id and isolation efficiencies with muons in proton-proton collisions at $ \sqrt{s} = $ 13.6 TeV	CMS Detector Performance Summary CMS-DP-2024-067, 2024 CDS
33	CMS Collaboration	The CMS statistical analysis and combination tool: Combine	Comput. Softw. Big Sci. 8 (2024) 19	CMS-CAT-23-001 2404.06614
34	W. Verkerke and D. Kirkby	The RooFit toolkit for data modeling	in the International Conference on Computing in High Energy and Nuclear Physics (CHEP ): La Jolla CA, United States, March 24--28 Proc. 1 (2003) 3	physics/0306116
35	L. Moneta et al.	The RooStats project	in the International Workshop on Advanced Computing and Analysis Techniques in Physics Research (ACAT ): Jaipur, India, February 22--27 Proc. 1 (2010) 3	1009.1003
36	T. Junk	Confidence level computation for combining searches with small statistics	NIM A 434 (1999) 435	hep-ex/9902006
37	A. L. Read	Presentation of search results: The $ \text{CL}_\text{s} $ technique	JPG 28 (2002) 2693