Longest Processing Time rule for identical parallel machines revisited

TL;DR

This paper revisits the Longest Processing Time (LPT) scheduling rule for identical parallel machines, providing new insights, improved approximation ratios, and an efficient heuristic that outperforms traditional LPT on benchmark instances.

Contribution

The paper offers a refined analysis of LPT's approximation ratio using Linear Programming and introduces a novel heuristic that enhances scheduling performance.

Findings

Improved approximation ratios for LPT modifications.

LP-based analysis as an alternative to formal proofs.

Heuristic significantly outperforms traditional LPT on benchmarks.

Abstract

We consider the Pm || Cmax scheduling problem where the goal is to schedule n jobs on m identical parallel machines to minimize makespan. We revisit the famous Longest Processing Time (LPT) rule proposed by Graham in 1969. LPT requires to sort jobs in non-ascending order of processing times and then to assign one job at a time to the machine whose load is smallest so far. We provide new insights on LPT and discuss the approximation ratio of a modification of LPT that improves Graham's bound from 4/3 - 1/(3m) to 4/3 - 1/(3(m-1)) for m >= 3 and from 7/6 to 9/8 for m = 2. We use Linear Programming (LP) to analyze the approximation ratio of our approach. This performance analysis can be seen as a valid alternative to formal proofs based on analytical derivation. Also, we derive from the proposed approach an O(n log n) heuristic. The heuristic splits the sorted jobset in tuples of m consecutive jobs (1,...,m; m+1,...,2m; etc.) and sorts the tuples in non-increasing order of the difference (slack) between largest job and smallest job in the tuple. Then, List Scheduling is applied to the set of sorted tuples. This approach strongly outperforms LPT on benchmark literature instances.