Implementation of a Fast Approximation Algorithm for Precedence Constrained Scheduling

Sammanfattning

We present an implementation of a very recent approximation algorithm for scheduling jobs on a single machine with precedence constraints, minimising the total weighted completion time. We also evaluate the performance of this implementation. The algorithm was published by Shi Li in 2021 and is a (6+ε)-approximation algorithm for the multiprocessor problem P|prec|∑j wjCj. We have implemented a version which is a (2+ε)-approximation algorithm for the single processor problem 1|prec|∑j wjCj. This special case can easily be generalised to the multiprocessor case, as the two algorithms are based on the same LP relaxation of the problem. Unlike other approximation algorithms for this and similar problems, for example, those published by Hall, Schulz, Shmoys and Wein in 1997, and by Li in 2020, this algorithm has been developed with a focus on obtaining a good asymptotic run time guarantee, rather than obtaining the best possible guarantee on the quality of solutions. Li’s algorithm has run time O((n+κ) · polylog(n+κ) · log3 pmax · 1/ε2), where n is the number of jobs, κ is the number of precedence constraints and pmax is the largest of the processing times of the jobs. We also present a detailed explanation of the algorithm aimed at readers who do not necessarily have a background in scheduling and/or approximation algorithms, based on the paper by Li. Finally, we empirically evaluate how well (our implementation of) this algorithm performs in practice. The performance was measured on a set of 96 randomly generated instances, with the largest instance having 1024 jobs and 32 768 precedence constraints. We can find a solution for an instance with 512 jobs and 11 585 precedence constraints in 25 minutes.