Nonlinear programming and scientific computing visualization in the optimization design of electron optical system

Citation
Cx. Gu et al., Nonlinear programming and scientific computing visualization in the optimization design of electron optical system, NUCL INST A, 427(1-2), 1999, pp. 321-328
Citations number
9
Categorie Soggetti
Spectroscopy /Instrumentation/Analytical Sciences","Instrumentation & Measurement
Journal title
NUCLEAR INSTRUMENTS & METHODS IN PHYSICS RESEARCH SECTION A-ACCELERATORS SPECTROMETERS DETECTORS AND ASSOCIATED EQUIPMENT
ISSN journal
01689002 → ACNP
Volume
427
Issue
1-2
Year of publication
1999
Pages
321 - 328
Database
ISI
SICI code
0168-9002(19990511)427:1-2<321:NPASCV>2.0.ZU;2-E
Abstract
In part I of this paper, five different nonlinear programming methods, impr oved Powell method, axis directional search method with fixed step, Hook-Je eves method, axis directional search method with variable step and Rosenbro ck method, are discussed for the optimization design of electron optical sy stems. Major aspects of these techniques in their applications are systemat ically studied, which include convergence rate, optimal objective function value and global convergence. The calculation results indicate that Rosenbr ock method should be recommended in electron optical system optimization. I n part II of this paper, an efficient method, reverse-match interpolation m ethod, is presented for the visualized optimization design of electron opti cal systems, which is a new development of CAD in electron optics. This met hod can produce the density distribution of electron beam from electron tra jectories worked out in the optimization search process. It makes real-time control system possible. (C) 1999 Elsevier Science B.V. All rights reserve d.