About This Issue . . . This issue contains two articles, one on trace-driven memory simulation, by Richard Uhlig and Trevor Mudge, and the other on approximately optimal solutions to some covering and packing problems by Vangelis Paschos. Trace-driven memory simulation is about simulating patterns of memory access to develop caching algorithms and other methods for reducing the discrepancy between memory and processing speeds. This comprehensive survey examines the accuracy, speed, memory, flexibility, expense, and ease of use of memory-simulation techniques â concluding that no single method is best and that changes in the content of traces are needed to simulate multiprocessing and distributed applications. This topic is part of the emerging area of input/output for large-scale computation that was one of the strategic directions in the December 1996 issue of Surveys. Covering a set C and a collection of subsets S is concerned with finding a (minimal) set of subsets of S whose union is the set C, while packing is about finding a maximum number of mutually disjoint subsets of S. This paper surveys approximation algorithms for set covering and packing, and related problems such as vertex cover, independent set, and clique, since exact algorithms are intractably hard. Though this paper is mathematically challenging, it provides insights pertinent to the relations among problems as well as some proofs simpler than those in the literature. The editors thank readers for their email on our two special 50th-anniversary issues. Readers wishing to express an opinion about either perspectives or strategic directions in computing are encouraged to write a letter or submit a two-page position statement to the editors. ACM Computing Surveys, Vol. 29, No. 2, June 1997
/lp/association-for-computing-machinery/editorial-leJOkxjDBb