Determination of impact parameter for CEE with Digi-input neural networks

Impact parameter is an important quantity which characterizes the centrality in nucleus-nucleus collision geometry. The determination of impact parameter in real experiments takes use of the hits on detector system or the reconstructed tracks of the secondary particles. As a task of feature recognition, methods such as sharp cut-off, Bayesian methods and Neural Networks (NN) has been studied and applied. However, in the situation of the Cooler-storage-ring External-target Experiment (CEE), the low beam energy brings a lapse of dependency between impact parameter and charged particle multiplicity, which decreases the validity of the explicit determination methods. This work proposes a regressor constructed with Graph Attention neural network, which takes the hit-level data as input. This model has shown a mean absolute error of 0.496 fm for the IQMD collision data of the UU system at 0.5 AMeV. The performance of such a model is compared with reference models, showing its capacity in handling the original but potentially interrelated digi information.Comment: 13 pages, 9 figure