Use this URL to cite or link to this record in EThOS:
Title: Metaheuristic approach for solving scheduling and financial derivative problems
Author: Lawrance Amaldass, Nareyus I.
ISNI:       0000 0004 7972 8597
Awarding Body: Brunel University London
Current Institution: Brunel University
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
The objective of this thesis is to implement metaheuristic approaches to solve di erent types of combinatorial problems. The thesis is focused on neighborhood heuristic optimisation techniques such as Variable Neighborhood Search (VNS) and Ant Colony Optimisation (ACO) algorithms. The thesis will focus on two diverse combinatorial problems. A job shop scheduling problem, and a nancial derivative matching problem. The rst is a NP-hard 2-stage assembly problem, where we will be focussing on the rst stage. It consists of sequencing a set of jobs having multiple components to be processed. Each job has to be worked on independently on a speci c machine. We consider these jobs to form a vector of tasks. Our objective is to schedule jobs on the particular machines in order to minimise the completion time before the second stage starts. In this thesis, we have constructed a new hybrid metaheuristic approach to solve this unique job shop scheduling problem. The second problem has arisen in the nancial sector, where the nancial regulators collects transaction data across regulated assets classes. Our focus is to identify any unhedged derivative, Contract for Di erence (CFD), with its corresponding underlying asset that has been reported to the corresponding component authorities. The underlying asset and CFD transaction contain di erent variables, like volume and price. Therefore, we are looking for a combination of underlying asset variables that may hedge the equivalent CFD variables. Our aim is to identify unhedged or unmatched CFD's with their corresponding underlying asset. This problem closely relates to the goal programming problem with variable parameters. We have developed two new local search methods and embedded the newly constructed local search methods with basic VNS, to attain a new modi ed variant of the VNS algorithm. We then used these newly constructed VNS variants to solve this nancial matching problem. In tackling the Vector Job Scheduling problem, we developed a new hybrid optimisation heuristic algorithm by combining VNS and ACO. We then compared the results of this hybrid algorithm with VNS and ACO on their own. We found that the hybrid algorithm performance is better than the other two independent heuristic algorithms. In tackling the nancial derivative problem, our objective is to match the CFD trades with their corresponding underlying equity trades. Our goal is to identify the mismatched CFD trades while optimising the search process. We have developed two new local search techniques and we have implemented a VNS algorithm with the newly developed local search techniques to attain better solutions.
Supervisor: Lucas, C. ; Roman, D. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Mathematics ; Operational research ; Finance