作者: Edmond C. Roelof , Andrew J. Skinner
关键词: Simplex algorithm 、 Extreme ultraviolet lithography 、 Physics 、 Algorithm 、 Optics 、 Population 、 Plasmasphere 、 Conjugate gradient method 、 Energetic neutral atom 、 Error function 、 Image (mathematics)
摘要: Energetic neutral atom (ENA) and extreme ultra-violet photon (EUV) imagers will soon be probing magnetospheric ion distributions from the NASA space missions IMAGE TWINS. Although ENA EUV images differ greatly, same basic mathematical approach can applied to deducing distributions: extracting parameters of a model distribution in magnetic field (and, case ENA, interacting with cold population). The is highly non-linear its many (as as 38 have been used) order describe strong spatial gradients intensities magnetosphere. We developed several new computer algorithms accomplish extraction by minimizing differences between simulated (instrument-specific) image an observed (or set images). Towards goal truly automated `hands-off’ algorithm, we combined three into Hierarchical Simplex Algorithm. At each step minimization, it first tries sophisticated efficient Adaptive Conjugate Gradient algorithm. Then, if error function not reduced, defers intermediate Analytic (if this also fails) finally defaults robust but inefficient Downhill Whenever successful, algorithm begins next at top hierarchy. offer completely different (without minimization) for interpretation sharp `edges’ (e.g., plasmapause He + 30.4 nm plasmasphere). demonstrate mathematically that equatorial shape constructed directly using simple graphical