Using AI and Machine Learning to reveal the physics of magnetic reconnection, a universal plasma process

在大多数天体物理等离子体区域，磁扩散率很低，磁场被“冻结”，这意味着不同区域的等离子体不能相互渗透。这导致边界层的形成和磁能的储存。磁重联可能发生在这些边界层内，称为电流片，使能量的转换和释放，改变磁拓扑结构和粒子加速。磁重联的中心扩散区起着至关重要的作用，因为这是等离子体退磁的地方。该项目的目标是从磁层多尺度（MMS）和其他航天器任务中获取数据，并使用人工智能和机器学习中的新技术来发现和分析MMS在流体层面上的重连观测（流动逆转等），然后对潜在的等离子体分布函数进行分类。这将有助于解释复杂的相空间结构是如何由重联引起的，以及它们如何在空间和时间中演化，并作为边界条件的函数。在动力学尺度上，以分布函数为代表的等离子体知识的提高，将使我们对重连如何转换能量的理解发生一步变化。

In most regions of astrophysical plasma magnetic diffusivity is low and the magnetic field is "frozen in", meaning that different regions of plasma cannot interpenetrate. This leads to the formation of boundary layers, and the storage of magnetic energy. Magnetic reconnection may occur within these boundary layers, known as current sheets, enabling the conversion and release of energy, a changed magnetic topology, and particle acceleration. The central diffusion region of magnetic reconnection plays a crucial role, because this is where the plasma demagnetises. The goal of this project is to take data from Magnetospheric Multiscale (MMS) and other spacecraft missions, and use novel techniques in AI and machine learning to find and analyse MMS observations of reconnection at the fluid level (flow reversals etc.), and to then categorise the underlying plasma distribution functions. This will help explain how complex phase-space structures are caused by reconnection and how they evolve in space and time, and as a function of the boundary conditions. Improved knowledge of the plasma at kinetic scales, represented by the distribution function, will enable a step change in our understanding of how reconnection converts energy.