Order 14: 211–228, 1997–1998.
© 1998 Kluwer Academic Publishers. Printed in the Netherlands.
211
Chains and Trees: ‘Strong’–‘Weak’ Order in Job
Scheduling
MOSHE DROR
MIS Department, College of Business, The University of Arizona, Tucson, AZ 85721, U.S.A.
(Received: 5 February 1997; accepted: 15 December 1997)
Abstract. We present a summary of recent NP-hardness and polynomial time solvability results for
the distinction between ‘strong’ and ‘weak’ precedence for chains and trees in scheduling. We dis-
tinguish between chains and proper trees which are not chains, and demonstrate that the strong-weak
precedence distinction for chains is not inclusive with regards to NP-hardness, and conjecture that the
same holds for strong-weak tree precedence. The objective is to show that different ‘interpretations’
for chain and tree order relations in scheduling might have far reaching computational implications.
Mathematics Subject Classifications (1991): 05, 06, 68.
Key words: partial order, scheduling, complexity.
1. Introduction
Ordered sets have been of great importance both in many applied and theoretical
problems in computer science and operations research, including ranging project
management, processors scheduling, and many others. The basic structure of these
problems requires partial orders such as precedence constraints in scheduling.
For at least the last two decades it has been common knowledge within the
computer science and combinatorial optimization community that most applied
problems which involve partial orders are NP-hard. Still, since much attention has
been paid to special classes of partial orders in search of “nice” structural properties
which could lead to the design of efficient solution procedures, or at least to obtain
bounds by structural relaxation, the delineation of the partial order in chain and
tree structures into ‘strong’ and ‘weak’ precedence is of considerable interest. The
‘strong’–‘weak’ precedence distinction has clear interpretations for chain and tree
precedence; this is not the case when series-parallel partial order is involved. Thus,
we not pursue those concepts beyond the tree structure.
This paper is a summary of the research results on the ‘strong’–‘weak’ prece-
dence distinction presented in Dror et al. (1996a,b,c and 1997) written for the
scheduling community, and its aim is to introduce those findings to a mathematical
community interested in new concepts related to partial orders (precedence).