在Belle II實驗中用ECL資訊辨別K0L介子

Abstract

我們可以透過從ECL束或KLM束所重建的K0L粒子列表來研究K0L介子。粒子識別(PID)可以幫助我們將K0L從其它粒子，如光子、中子、反中子，和其他電中性粒子區分開來。本論文使用Belle II實驗中ECL的資訊並描述識別K0L粒子之建立、方法、表現及比較。 在這篇論文中，我們使用快速提升決策樹(FastBDT)以獲得訓練過後的多變量分析方法。此方法可以將訊號(K0L)以及背景(K0L以外的所有電中性粒子)。我們呈現重建B0 → J/ψ K0L衰變的初步結果，其中J/ψ會衰變成μ+μ−，而K0L介子是從ECL裡的中性強子束重建而來。在Belle II實驗中使用蒙地卡羅方法單純產生訊號，我們可以從100萬個事件裡得到25%的效率。使用蒙地卡羅方法生成各種粒子時，我們可以從100/fb的光度獲得681±71個訊號事件。此外，我們也使用Belle II實驗在SuperKEKB的非對稱能量e+e−對撞機所收集到光度189.049/fb的數據。在J/ψ → μ+μ−的情況下，我們總共量測到1031±95個訊號事件。我們也比較了上述的結果與從KLM裡的K0L介子重建B0 → J/ψ K0L衰變的結果。透過在訊號生成與真實數據中比較ECL和KLM在全範圍以及訊號範圍的結果，我們發現ECL和KLM並沒有太大的重複性。因此本論文提供了K0L識別的另一個方法。

We can study K0L through K0L particle list reconstructed from ECL cluster or KLM cluster. Particle identification (PID) can help separate K0L from other particles, like γ, neutron, anti-neutron and other neutral particles. This thesis shows the establishment, method, performance and the comparison of K0L particle identification by using only ECL information in Belle II experiment. In this thesis, we use Fast Boosted Decision Tree (FastBDT) to get a trained multivariate analysis (MVA) method which can separate signal (K0L) and background (neutral particles excluding K0L). We present the results on the reconstruction of the B0 → J/ψ K0L decay, where J/ψ → μ+μ−. K0L mesons are reconstructed from the hadronic neutral clusters in the Electromagnetic Calorimeter (ECL). Using signal Monte Carlo (MC) with one millon events in Belle II, we have about 25% efficiency. In generic MC corresponding to a luminosity of 100/fb, we get 681 ± 71 candidates. Moreover, we use the Belle II experiment dataset corresponding to a luminosity of 189.049/fb collected in the asymmetric energy e+e− collider at SuperKEKB. We measure 1031±95 candidates with J/ψ → μ+μ− in total. We also compare the results above with the reconstruction of the B0 → J/ψ K0L decay, while K0L mesons are reconstructed from the hadronic neutral clusters in K0L and muons (KLM) subdetector. By the comparison of the result in ECL and KLM in whole region and signal region with signal MC and Data, we notice that the overlap of ECL and KLM is not large. This provides another way for K0L identification.