PhD projects offered
We have positions available for PhD's in the areas of Modelling, Solving, BioInformatics, and Interactive Optimisation. For more information, please follow the link below.
Current Featured PhD students
Ahmad’s research lies in the Mathematical Optimization, particularly Mixed Integer Programming methods. Under the supervision of Andreas Ernst, Mohan Krishnamoorthy, and Pierre Le Bodic, Ahmad seeks novel ways to improve the efficiency of freight transportation in synchromodal networks by employing Network and Large-Scale Optimization techniques.
Jip J. Dekker
Jip’s research is focused on enhancing the functionality of MiniZinc. Supervised by Guido Tack, Maria Garcia de la Banda, and Andreas Schutt, Jip integrates advanced solving techniques, such as Auto-Tabling, Half-Reification, and Large Neighbourhood Search, into the MiniZinc toolchain. The result of his work allows modellers to exploit state of the art optimisation concepts with ease.
Shan Dora He
Shan works on providing new ways to keep the lights on at lower costs. Her research has lead to “Demand Manager” — a tool that schedules millions of household devices in a second to flatten the peak electrical demand, minimise the overall electricity supply cost and better utilise renewable energies such as wind or solar power. This tool will make the electricity supply more reliable, cheaper and greener. Her research is interdisciplinary that involves mathematics, computer science and energy management. Her work is supported by Ariel Liebman, Mark Wallace and Campbell Wilson.
The transition to a new low emission energy future, with a changing mix of generation and load types with significant growth in renewable energy generation and the intermittent nature of renewable energy sources, increases technical challenges for electrical grid investment and operation planning, aiming to minimise the cost of investment and operation while respecting dispatch, network, and security constraints of a power system. Stability and security of operation without interruption of consumer supply is of paramount importance and machine learning methods provide promising options to be embedded within power system planning. Therefore, the main focus of my research is on developing intelligent tools for fast and secure planning of future power systems.
Peter’s PhD is about designing the most cost-effective solution on how to accommodate more solar PV installations in distribution networks while providing safe power system operation. He says that instead of undertaking expensive network upgrades, we can increase the maximum PV capacity limits through coordinated control of power electronics, so customers can install solar PV systems and facilitate the decarbonisation of electricity grid at minimal cost. Peter is supervised by Ariel Liebman, Guido Tack, and Lachlan Andrew.
Rejitha Nath Ravindra
Rejitha’s research focuses on developing a holistic optimisation tool that improves the efficiency of our public transport timetables. Her research proposes a constraint based optimisation model that minimises the excessive passenger transfer waiting times between public transport services. Written in MiniZinc, the model accommodates a wide range of real world planning and operational scenarios that address the practical concerns of public transport users and schedulers alike. Her research is interdisciplinary and has a strong industry focus, with an aim to aid transit agencies with faster decision making-to realise accurate, realistic and cost-efficient timetable solutions. With the support of her supervisors- Mark Wallace, Graham Currie, Daniel Harabor, Ilankaikone Senthooran and Chris Loader, her research seeks to achieve a synergy between scheduling in principle and scheduling in practice.
Saman’s research focus is to design a framework for on-demand transport operated by state-of-the-art Autonomous Electric Vehicles (AEV). With the supervision of Guido Tack, Daniel Harabor and Philip Kilby, through various models and solvers developed in Minizinc, Saman investigates innovative methods of optimally addressing transit demands while taking vehicle dynamics and technical limitations into account.
Alexander J. Ek
Alexander's research is focused on extending MiniZinc to enable solving and coherent modelling of online optimisation problems. An online optimisation problem is a problem where new information about it is revealed over time (in real-time) and irreversible decisions have to be (or at least can be) made at each time based on this information. His research is supervised by Guido Tack, Peter J. Stuckey, Maria Garcia de la Banda, and Andreas Schutt. Alexander's work will lower the skill level for solving and modelling of online optimisation problems and increase the utility of constraint modelling (and MiniZinc).
Shizhe is working on single-agent search. Under the supervision of David Taniar, Daniel Harabor and Muhammad Cheema. Shizhe works on improving the compression of Compressed Path Database for grid maps and road network; he also extends the Polyanya, a fast point to point pathfinding on euclidean plan with obstacles, to multi-target scenario.
Bojie’s research is focused on single-agent pathfinding in Euclidean space, under the supervision of Muhammad Aamir Cheema and David Taniar. Bojie’s research aims to develop a fast, optimal algorithm that finds the euclidean shortest path from any given start to target in the plane. His research is motivated by a variety of real-world applications including computational geometry, robotics, and computer games.
Past Featured PhD Students
David Hemmi (2018)
David is working on algorithms to solve optimisation problems that are subject to uncertainty, so called Stochastic Programs. Most approaches to model and solve Stochastic Problems require expert knowledge. This prevents the widespread adoption of decision making under uncertainty in practice. With the support of his supervisors Guido Tack and Mark Wallace, David proposes a framework to solve Stochastic Problems formalised in MiniZinc.
Maxim Shishmarev (2018)
Maxim's thesis provides the design, implementation and evaluation of a profiling system for constraint programs that (a) is efficient and thus scales for large problems, (b) is solver-independent and thus can easily integrate different solvers, and (c) can support modern solvers and be extended to those developed in the future. In addition, the thesis proposes and integrates several proling techniques into this system: search-tree visualisations that are suitable for large problems; a technique for automatically finding recurring patterns in large problems that are often the reason for slow executions; two complimentary techniques that allow modellers to evaluate model modifications; and a technique for analysing so-called learning solvers that can aid the modeller in discovering weak constraints, as well as redundant constraints that if added to the model, will speed up the execution. Together, these techniques allow modellers to find high quality solutions faster by aiding them in making effective model modifications.
Kevin Leo (2017)
Kevin's thesis explores the structure present in high-level constraint models. It shows how this structure can be exploited to improve the compilation and solving process. In particular, it proposes a framework for extracting and using structure to solve problems faster and to help users write better models and debug them.
More specifically, the thesis had three main contributions:
- First, a method that preserves model structure deeper into the compilation process, thus enabling more efficient programs to be produced.
- Second, a method for finding implicit structure that can be made explicit in a model, thus making modelling easier
- And third, a method that uses preserved model structure to help explain why a model is unsatisfiable, thus helping users produce correct models.
These three methods have been implemented and evaluated for the MiniZinc language. The experimental results show them to be effective in practice. Together, these contributions help make high-level modelling a more powerful and attractive approach for solving hard combinatorial problems.
James Collier (2016-Vice-Chancellor’s commendation for Thesis Excellence)
James’ thesis proposed a fundamental shift in the way protein structural alignment quality is formalised and measured, and in the way biologically-meaningful alignments are identified. It brought together ideas from fields of information theory, data compression, optimisation and statistical inductive inference to develop a statistically rigorous framework to measure structural alignment quality. The resulting alignment quality measure, called I-value, is used by an optimisation algorithm also developed by James to consistently identify high quality and statistically significant structural alignments. This optimisation method is also able to identify significant alternative structural alignments of comparable quality. The culmination of James’ work is an open-source pairwise structural alignment program called MMLigner (available from http://lcb.infotech.monash.edu.au/mmligner ). Extensive benchmarking of MMLigner against popular alignment programs and alignment scoring functions, found it is highly-competitive and consistently outperforms these methods in identifying alternative structural alignments, a challenging problem when aligning oligomeric proteins and protein complexes.
Parthan Kasarapu (2016)
Parthan’s thesis explores the use of information-theory and compression principles to address the problem of optimal model selection, as these principles enable a better trade-off between the model's complexity and its goodness-of-fit to the data. The core of Parthan’s thesis explores the inference of models for some of the commonly used probability distributions and their mixtures. The results allow for accurate modelling of data in the Euclidean space and data that is directional in nature. Therefore, they have widespread uses in statistical machine learning tasks. To demonstrate this, Parthan’s thesis developed a method to determine the optimal number of mixture components and their parameters that describe the given data in a completely unsupervised setting. The method was benchmarked in detail on a variety of real-world data, specifically on directional text data and on the spatial orientation data of protein three-dimensional structures. In addition, the mixtures of directional probability distributions have facilitated the design of reliable computational models for protein structural data. Finally, the inference framework has been used for concise representations of protein folding patterns using a combination of non-linear parametric curves.
Richard Kelly (2015)
Richard's thesis focuses on transport efficiency in supply chains. Transport is a major cost especially in Australia and has significant impact on the environment. Transport efficiency can be improved by optimisation of routing and scheduling in a supply chain or by integrating components of different supply chains. This thesis addresses both forms of transport efficiency. A core driver of the research is the introduction of shared depots that allow multiple logistics companies to combine resources. By designing improved vehicle routes to effectively use these shared depots, a sharp reduction in overall truck kilometres can be achieved. The two optimisation problems have never been addressed in an integrated way. Typically, depot placement is performed before vehicle routing for a single supply chain. Finding solutions to both optimisation problems in an integrated way is a novel challenge because the techniques for depot placement and vehicle routing techniques are very different.
Chris Mears (2010)
Chris's thesis explores automatic methods for detecting and exploiting symmetries in constraint programs. Symmetry in an optimisation problem represents an opportunity to reduce the effort required to solve it, since parts of the search space can be safely eliminated. The thesis presents a new method for finding symmetries in small constraint programs quickly, and uses this as the foundation of a framework for detecting symmetries in higher-level constraint models. This allows the finding the symmetries in a whole class of problems. The thesis also presents a new method of symmetry breaking that focuses on practical performance without interfering with any problem-specific search heuristics. Together, these contributions form an automatic system for detecting symmetries across a class of problems and exploiting those symmetries when solving different instances of the that class.