Evolutionary algorithms for solving bilevel optimization problems

Bilevel optimization involves solving an upper level "leader" optimization problem subject to the optimality of a nested lower level "follower" problem. Both levels have their associated objective(s), variable(s) and constraint(s) with interaction between them. Such problems model real-life scenarios where there exist decision-makers are multiple hierarchical levels. A number of practical applications in the diverse fields such as logistics, transportation, engineering, finance and others are suitable to be modelled as bilevel problems due to their inherent nested structure.

Bilevel optimization problems contain several characteristic challenges not found in standard (single level) problems, such as (a) requirement of a large number of design evaluations to achieve good solutions (b) difficult-to-navigate problem landscape (c) conflict between the leader and follower problems that can mislead search algorithms and (d) in-applicability of standard performance metrics for assessment and benchmarking. 

This research aims to develop methods to address the above challenges and deal with bilevel optimization with using low computational expense. The developed methods will be primarily metaheuristic (such as evolutionary algorithms) in nature, complemented by improvements in mathematical formulations and use of metamodels.

Required Background:

• Good programming (e.g. Matlab/Python) and analytical skills, preferably with a Masters Degree in Engineering / Computer Science
• Prior research experience in optimization is desirable but not necessary
• Demonstrated competence in academic writing and oral presentation skills will be beneficial.

You can find more details of the research conducted in our Multidisciplinary Design Optimization (MDO) group at http://www.mdolab.net/. Please feel free to reach out and discuss regarding this project, or have a discussion about other potential topics for undertaking Masters (research) or PhD with us.   

How to Apply

Express your interest in this project by emailing Associate Professor Hemant Kumar Singh at h.singh@unsw.edu.au. Include a copy of your CV and your academic transcript(s).