新的AI算法提高了欧洲核子研究中心大型强子对撞机的粒子探测精确度和速度,为今后的升级做准备。 A new AI algorithm boosts particle detection accuracy and speed at CERN’s Large Hadron Collider, preparing for future upgrades.

由欧洲核研究组织CMS合作开发的新的机器学习算法MLPF改进了在大型强子对撞机上重塑质子-质子碰撞的工作。 A new machine learning algorithm, MLPF, developed by the CMS collaboration at CERN, has improved the reconstruction of proton-proton collisions at the Large Hadron Collider. 与传统的基于规则的方法不同,MLPF从模拟数据中学习,以更准确和更快速地识别粒子,在关键的动量范围中达到1020%的更高精度,并在GPU上运行更快. Unlike traditional rule-based methods, MLPF learns from simulated data to identify particles more accurately and quickly, achieving up to 10–20% better precision in key momentum ranges and running faster on GPUs. 它与长期的粒子流算法的性能相匹配或超过后者,从而能够进行更有效的数据分析。 It matches or exceeds the performance of the long-standing particle-flow algorithm, enabling more efficient data analysis. 在LHC2030升级期间,预计这一进展将至关重要。 这将使碰撞率提高五倍,支持对标准模型进行更深入的测试和寻找新的物理学。 The advancement is expected to be crucial during the LHC’s 2030 upgrade, which will increase collision rates fivefold, supporting deeper tests of the Standard Model and searches for new physics.