Speaker
Description
Reliable track reconstruction is vital for the success of the Belle II experiment, especially as detector conditions evolve. The full tracking chain in the Belle II Analysis Software Framework combines Central Drift Chamber (CDC)-seeded and Silicon Vertex Detector (SVD)-seeded tracking stages, which are merged and cleaned to produce final tracks using a Combinatorial Kalman Filter (CKF). We focus on optimising the SVD-seeded CKF extension into the CDC, where SVD-only tracks are extrapolated to attach CDC hits. This step is sensitive to CDC inefficiencies caused by ageing electronics and inactive wires. A grid search is used to tune key CKF hyperparameters—such as allowed layer skips—using a F1 score built from hit efficiency and hit purity as the optimisation objective. The resulting parameter set, which yields the highest F1 score, is integrated into the full tracking chain, thereby improving tracking robustness and performance under degraded detector conditions. In this task, a description of the Belle II tracking chain and results from the optimisation will be presented.