This study presents a novel method for improving the resolution and linearity of proton energy reconstruction in isotropic calorimeters. The proposed approach utilizes the Virtual Layer Algorithm to allocate energy deposition, followed by longitudinal fit to quantify the total electromagnetic energy component. The proton energy is subsequently reconstructed by applying a correction factor. The method improves the energy resolution from approximately 70
To investigate the temperature dependence of dark current in radiation detectors based on p-Si/n-InP heterostructures and to identify the dominant charge-transport mechanism under reverse bias. Reverse current–voltage characteristics were measured in the temperature range of 308–333 K at reverse biases from 5 to 30 V. Arrhenius analysis was employed to determine the activation energy of the dark current. The dependences of ln(I) versus ln(V) and ln(I/V) versus √(V) were additionally analyzed to assess the possible contributions of space-charge-limited current (SCLC) and Poole–Frenkel emission mechanisms. The dark current increased monotonically with temperature throughout the investigated voltage range without evidence of breakdown or avalanche multiplication. The activation energy was found to range from 0.72 to 0.88 eV, with an average value of approximately 0.78 eV. Analysis of the current–voltage characteristics showed that neither the SCLC nor the Poole–Frenkel mechanism provides a dominant contribution to the dark current. Owing to the thin InP layer ( 20 nm) and the highly asymmetric carrier concentrations of the heterojunction, the depletion region extends predominantly into the high-resistivity p-Si substrate. The results indicate that the dark current in p-Si/n-InP heterostructure detectors is predominantly governed by thermally activated generation–recombination processes associated with deep levels. The investigated detectors maintained stable electrical characteristics throughout the studied temperature range of 308–333 K, demonstrating their potential for ionizing-radiation detection under variable-temperature operating conditions without active cooling.
This study aims to propose a machine learning-based cell-by-cell linear optics calibration method for storage rings, using the storage ring of the Ultrafast Transient Experimental Facility (UTEF) as the physical model. It addresses whether local cell-level data can replace full-ring fitting to achieve faster and more accurate optics calibration in compact lattices that are highly sensitive to magnet errors. A cell-by-cell optics calibration method was established by exploiting paired correctors and beam position monitors (BPMs) located at both ends of a straight section to lock boundary conditions for the linear optics functions and convert global ring optics fitting into local cell calculations. On this basis, a convolutional neural network (CNN) enhanced with residual blocks, an attention mechanism, and multi-scale pooling was trained to predict quadrupole gradient errors. A composite loss incorporating higher-order statistics and correction-direction constraints, together with a variance-decomposition-based post-processing algorithm, was introduced to improve prediction quality. Simulation results show rapid and stable convergence. After applying the predicted corrections in beam-dynamics simulations, the 1-sigma β -beating distributions were reduced to about 4% in the horizontal plane and 7% in the vertical plane and exhibited a certain robustness against BPM noise, supporting the feasibility of the method for practical optics tuning. Shapley Additive exPlanations (SHAP) analysis further quantified the contributions and interaction effects of input features, particularly orbit response matrix (ORM) features. The proposed approach reduces data dimensionality and improves calibration efficiency while maintaining satisfactory correction performance. It offers a promising strategy for fast storage ring linear optics calibration.
The barrel yoke of the Circular Electron Positron Collider (CEPC) detector faces a loading condition in which the gravitational load of the internal subdetectors exceeds its self-weight. Conventional polygonal staggered laminated barrel yoke structures struggle to meet the requirements for stiffness, strength, and installation precision. This paper presents an inter-supporting barrel yoke structure and numerically evaluates its mechanical performance and feasibility. The inter-supporting barrel yoke structure consists of 12 modules and 2 end flanges. The end flanges serve three functions of structural reinforcement, assembly tooling, and precision datum. Using the ANSYS finite element analysis software, static, modal, and seismic analyses were carried out for both the conventional barrel yoke structure (Structure I) and the inter-supporting barrel yoke structure (Structure II) under the coupled action of self-weight, internal load, and electromagnetic force. Structure II achieves substantial performance gains over the conventional Structure I. The maximum deformation decreases from 7.8 to 2.4 mm (a 69.2 125.6 MPa (a 66.8 5.03 Hz (an 87.7
