机器学习在大型粒子加速器中的应用回顾与展望

机器学习技术在近十几年发展迅猛,并被广泛地用于解决复杂的科学和工程问题。最近十年间,基于机器学习的粒子加速器相关研究也开始呈现出井喷式发展趋势。国际上许多加速器实验室开始尝试用机器学习和大数据技术处理加速器中的海量复杂数据,以期解决加速器及其子系统中的诸多物理和技术问题。不过,迄今为止,机器学习在加速器中的应用仍处于初步探索阶段,不同机器学习算法在解决具体加速器问题的效果及其适用范围尚待摸索,机器学习在实际加速器中的应用仍非常有限。因此,有必要对加速器领域中的机器学习研究做一个整体回顾和总结。将回顾机器学习在大型粒子加速器（以储存环加速器和直线加速器为主）中的加速器技术、束流物理以及加速器整体性能优化等研究方向中已取得的研究成果,并探讨机器学习在加速器领域的未来发展方向和应用前景。     

Beam dynamics studies on BEPC-Ⅱ storage rings at the commissioning stage

During the 1 st and 2 nd stages of the commissioning of the upgrade project of the Beijing Electron Positron Collider (BEPCⅡ), which started on Nov. 12, 2006 and Oct. 24, 2007, respectively, we got the luminosity one tenth of its design value, provided beams to synchrotron radiation users for about 4 months, and studied beam dynamics as well. In this paper, some beam dynamics studies on the storage rings and their preliminary results are given.     

Analytical transfer matrix of a quadrupole fringe

The analytical linear transfer matrices for different quadrupole fringes including quadratic, high order power and exponential models are deduced in this paper. As an example, the transfer matrices of the quadrupole BEPCⅡ/ 105Q are computed for the above three models and compared with hard edge and slice-by-slice models in cases of near 60° and 90° FODO cells. These models' results are much better than the hard edge model's, and can meet the requirement of accurate calculation.     

高能光源增强器束流横向尺寸测量系统设计

高能同步辐射光源的增强器将直线加速器注入的束流加速到储存环所需的能量,为储存环提供高品质的电子束。为了对增强器的束流横向截面尺寸、发射度及能散进行测量,设计了两条可见光-紫外波段的束测光束线。两条光束线分别选取无色散和色散较大的两处弯铁位置作为光源点,使用两套同步光成像系统来监测光源点的束流截面尺寸,并计算束流发射度及能散。介绍了同步光引出真空室及光学成像系统,对影响成像质量的空间分辨率进行了分析,并针对升能过程中不同能量下束流光斑变化的测量进行了设计。