Reconstruction of a 4D Particle Distribution Using UnderdeterminedPhase-Space Data

Rostamizadeh, Afshin

Lawrence Berkeley National Laboratory

关键词: Data Analysis;    Phase Space;    Programming Languages;    Java;    Ion Beams;   

DOI  :  10.2172/877634
RP-ID  :  LBNL--58929
RP-ID  :  DE-AC02-05CH11231
RP-ID  :  877634

美国|英语

来源: UNT Digital Library

【 摘 要 】

A well defined 4D distribution that describes the transverse spatial coordinates (x,y) and momenta (x',y') of the particles that make up an intense ion beam is of great value to theorists in the field of particle beam physics. If such a distribution truthfully captures the characteristic of the actual beam, it can be used to initialize an extensive simulation, and can yield insight into the processes that affect beam quality. Creating a proper representative distribution of particles is a challenge because the problem is, in general, quite underdetermined. Data is collected through a pair of ''optical slit'' diagnostics which provide two 3D distributions, f(x,y,x') and f(x,y,y'); the challenge is to coalesce these into a full 4D distribution f(x,y,x',y'). Further difficulties are introduced because the data is collected at different longitudinal planes and must be ''remapped'' to a common plane, taking into account the convergence or divergence of the beam as well as any off-centering. This challenge was met by developing a suitable algorithm and implementing it as a ''plug-in'' for the popular scientific image analysis program ImageJ, written entirely in the Java programming language. The algorithm accomplishes the desired remapping and synthesizes a 4D particle distribution, using Monte-Carlo techniques. Preliminary results show that this reconstructed distribution is consistent with actual data that was gathered from the same experiment using a different diagnostic. Also, ''forward'' particle-in-cell (PIC) simulations, that use the reconstructed distribution, match actual data gathered downstream in the experiment. Both these results give us some indication that the reconstruction is being done correctly. In addition to the multi-particle synthesis, the plug-in allows for the easy loading of digital data and the output of various plots that are useful to both experimenters and theorists. It also provides a framework by which its applicability can be extended to other types of experiments for which data analysis and simulation-particle synthesis are required.