Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.sao.ru/precise/Midas_doc/doc/94NOV/vol2/node316.html
Дата изменения: Fri Feb 23 14:02:28 1996 Дата индексирования: Tue Oct 2 18:50:25 2012 Кодировка: |
This package is descended from a program originally written for the IBM 704 in the late 1950's [26]. It used all constant stars as extinction stars, combined multiple nights with separate extinction coefficients, and employed King's correct representation of color terms in the extinction. That program was written in assembly language after an initial try in FORTRAN II would not fit into the 8192 words of the machine's memory. Data were packed, two to a word, on magnetic drums. Computers at that time were so small that the program needed every cell of the machine; even so, it had to be split into overlays.
When the 7090 came along, the program was stored on a loadable tape that displaced the FMS operating system from core memory, and pulled the system back in when it finished. This special system tape was labelled PEPSYS (for Photo-Electric Photometry SYStem). This original PEPSYS died when the IBM 709x systems became obsolete.
A second attempt was made on a Cyber 175 many years later. The planning program was added in the 1980's. This was ported to a VAX, and later to other machines (AT&T 3B2 and Sun 4). These ports shook a lot of portability bugs out of the common subroutines. The Manfroid -- Heck program [15], and a program used at that time by Peter Stetson were investigated, but found to be too specialized to particular photometric systems and data formats to be generally useful, so a proposal was made to NSF to develop something along the lines of the present system. The proposal was rejected on the grounds that (a) it was too expensive, and (b) such a thing is impossible anyway.
The heart of the present reduction program --- the fast matrix inversion using partitioning -- was developed during a simulation study in the late 1980's. However, it was used only on phony data, and lacked the extensive user interface of the present version. Some user interface, especially material now embedded in the MAKE/PHOTOMETER command, was developed on an Intel 80386 system running SCO Xenix System V.
The current system was sponsored by ESO; I thank Chris Sterken and Preben Grosbø l for setting up my visit to Garching. This version was developed within ESO's Image Processing Group. Although an image-processing system offers the photometrist as many inconveniences to work around as useful tools, it has been possible to hide most of the problems from the user.