New complexity and approximability results for minimizing the total weighted completion time on a single machine subject to non-renewable resource constraints

TL;DR

This paper investigates the complexity and approximation algorithms for scheduling jobs on a single machine with non-renewable resource constraints, focusing on minimizing total weighted completion time, and introduces new results and algorithms for special cases.

Contribution

It provides new NP-hardness results even with simplified resource requirements, develops an FPTAS for certain cases, and offers approximation guarantees for greedy algorithms.

Findings

NP-hardness even with unit resource requirements and weights

An FPTAS exists when the number of supply points is bounded

Greedy algorithms have provable approximation guarantees

Abstract

In this paper we consider single machine scheduling problems with additional non-renewable resource constraints. Examples for non-renewable resources include raw materials, energy, or money. Usually they have an initial stock and replenishments arrive over time at a-priori known time points and quantities. The jobs have some requirements from the resources and a job can only be started if the available quantity from each of the required resources exceeds the requirements of the job. Upon starting a job, it consumes its requirements which decreases the available quantities of the respective non-renewable resources. There is a broad theoretical and practical background for this class of problems. Most of the literature concentrate on the makespan, and the maximum lateness objectives. This paper focuses on the total weighted completion time objective for which the list of the approximation algorithms is very short. In this paper we extend that list by considering new special cases and obtain new complexity results and approximation algorithms. We show that even if there is only a single non-renewable resource, and each job has unit weight and requires only one unit from the resource, the problem is still NP-hard, however, in our construction we need a high-multiplicity encoding of the jobs in the input. We also propose an FPTAS for a variant in which the jobs have arbitrary weights, and the number of supply time points is bounded by a constant. Finally, we prove some non-trivial approximation guarantees for simple greedy algorithms for some further variants of the problem.