Scheduling university course examinations by computer

A new approach to the problem of scheduling course examinations is presented. In principle, an examination schedule which requires a minimum number of examination periods and satisfies the constraint that no student be required to take two examinations simultaneously can be found in two steps. First, courses which may have their examinations scheduled at the same period ore grouped together in all possible ways. Then a minimum number of these groups, such that each course is included at least once, are selected. By removing multiple occurrences of courses and then scheduling each group at a different period a minimal schedule can be obtained. Known algorithms for carrying out these procedures are prohlbitively expensive. Approximations to the ideal procedure outlined above are given which yield nonminimal but feasible schedules with a very small expenditure of time. Results of experiments using these techniques are given. These are encouraging and indicate that further experimentation would be worthwhile. In scheduling University course examinations, two restrictions predominate: absolute conflicts must not occur, and exams ought not 1o be bunched too tightly. To state them different, ly, no s(.udeng can be asked to take two ex-aminatiolls shmtllalteously and not too many students should have their examinations occurring too closely together. At Princeton University the precise definition of "leo Hose togelher" has varied over the years but, typi-(!ally, a. student could not be gives1 three examinations within any 2t: hotlr period. At times, a further restriction forttidclinK four exams within 48 hours has been added. Since this bttnching :restrictiou is very diffi<+ult to check, historical practice ires tolerated its occurrence in the schedule but has offered relief to the affected students in tltc Form of postponed examinations-a solution that is tin-popular wilt* both the faculty and the Registrar, (,hough fop different reasons! In 1his paper we describe an effort to Mleviate some of these dilticulties t)y transferring the process to a large computer (CI}C 1604). The present purpose is to indicate the major practical dirt*cult*as and to report on their expedient, if not always elegant, solution. We shall not be concerned with details of the programming nor the preparation of itU)u(, do.to, important thoug|l these items be, T h e G r a n d S t r a t e g y Our ge~eral approach has been to concentrate the com-i)lger effort on the a.bsolute conflict problents. We have * Del>nrt, n~e~at of ]+;lectrical F m …