Brad Alexander

Lecturer
School of Computer Science
University of Adelaide
Adelaide, South Australia, 5005

Email: brad@cs.adelaide.edu.au

Research Interests

I am a member of the Evolutionary Computation Group at the University of Adelaide.

My research interests include program optimisation, rewriting, genetic-programming (GP) - specifically as it relates building program optimisers and search-based-software-engineering I have also supervised successful projects in the evolution of control algorithms for teams of autonomous agents and using background optimisation to improve the performance of instruction set simulators (ISS)'s.

I also have a strong interest the teaching of problem-solving skills in Computer Science curricula. In particular the learning of problem-solving skills through deliberate practice. I am a member of the Education Research Group of Adelaide (ERGA). I am one of the holders of a Google Research Award to introduce a problem-solving curriculum into Computer Science. A little more informaton on my interests in education research can be found here.

Current Research Projects

Enhancing a Fast Instruction Set Simulator

This project, carried out in collaboration with an industry partner, the Australian Semiconductor Technology Company (ASTC) uses background compilation on a multicore machine to further enhance the peformance of an, already fast, instruction set simulator. The project is relatively new and has, thus far, yielded very promising results and work is ongoing. In future we hope to progress this work into a modular and integrated framework for fast cross-platform simulation. If you have an interest in virtual machines, low-level programming and compiler technology please feel free to contact me regarding potential projects.

Control for Teams of Autonomous Ground Surveillance Vehicles

This project contributes to the building of a team of autonomous vehicles to map and patrol urban areas. The project is ongoing. We need students who are good programmers interested the development of novel control strategies for teams of robots. The work staff from across the faculty of Engineering Maths and Computer Science. Development of prototypes is ongoing and early results are encouraging. This work can form the foundation for ongoing undergraduate or graduate research.

Cloud Computing for Geoscience Modelling

Part of the new C3L group at Adelaide, this work uses the cloud resources to get much faster results computationally intensive geoscience applications. Our initial application domain will be magneto-tellurics, a cost-effective but computationally intensive ways to map large areas to great depth. This is of strong relevance to minerals exploration in Australia where resources can be obscured by deep layers of regolith which expensive to probe by other means.

Using Genetic Programming (GP) to evolve Optimisation Phases for a Compiler

Optimisers work in a complex design space and are very difficult to construct manually, not least because it is difficult to predict the optimiser design choices that will work best for each application domain. This work extends previous successful work in automatically evolving the strategies used to drive a detailed optimisation phase in the Adl compiler - eventually targetting FPGA's. The slides for a very non-technical introduction to this project I delivered in Singapore earlier this year can be found here[pdf 4.6MB]. The talk is based on work published in our CEC 2009 paper below.

Tracable Compilation from a Functional Language using Strategic Rewriting

This work is part of the Adl Project to take a small functional language, through point-free intermediate form, to an FPGA. This work contrasts with most work mapping high-level-languages to FPGA by avoiding the need for the programmer to "wire-up" the transfer of values through their code or to use restricted idioms or heavy use of pragmas. Communication is implicit in the source code. Leaving the compiler with the unenviable task of making this communication explicit and then reducing the quantity of this communication (data-movement optimisation). Because the compiler is based on rewriting (using the excellent rewriting language Stratego) it is extremely robust and also traceable. Current information about the project including a tracable prototype - complete with traces can be accessed here. This prototype forms the basis of later experiments in evolving an optimiser using GP. The approach is highly robust with every one of the hundreds of indivdual optimisers tested not breaking any benchmark code.

Other research activities include the development of infrastructure for the more efficient support of deliberate practice in teaching computing and the development of genetic grouping algorithms for cooperative learning. A prototype genetic grouping application (GGACL, developed as part of student summer research project with Brendon Henessy and Yip Chong) that works very well on large classes can be downloaded from this link. I am happy to sponsor a small student project to build a Java GUI for this application to enhance its usability.

Recent Publications

• Relating Blob Model Complexity to Model Error (Abstract/Poster), Brad Alexander, Stephan Thiel, Jared Peacock, 21emiw, Darwin Australia, July 2012 (accepted).
• Application of Evolutionary Methods to 3D Geoscience Modelling, Brad Alexander, Jared Peacock, Stephan Theil, GECCO 2012, corrected version (fixed typo in background resistivity).
• Boosting Instruction Set Simulator Performance with Parallel Block Optimisation and Replacement, Brad Alexander, Sean Donnellan, Andrew Jeffries, Travis Olds, Nicholas Sizer, Australasian Computer Science Conference (ACSC), 2012.
• Evolving Path Planner for A Multi- Robot Exploration System Using Grammatical Evolution, Faisal Ibrahim, Brad Alexander, ISSNIP 2011.
• Evolving Patches for Software Repair, Thomas Ackling, Brad Alexander and Ian Grunert, GECCO 2011 - Smallworld example from the publication can be accessed here.
• Engaging Weaker Programmers in Problem Solving, Brad Alexander and Cruz Izu, IEEE International Conferernce on Engineering Education, EDUCON 2010.
• Constructing an Optimisation Phase Using Grammatical Evolution, Brad Alexander and Michael Gratton, 2009 IEEE Congress on Evolutionary Computation, CEC 2009. talk slides[pdf]
• Data Movement Optimisation in Point-Free Form [pdf], Presented at The 11th International Conference on Algebraic Methodology and Software Technology, AMAST '06.
• A Simple Programming Model for New-Generation Hardware[pdf], Presented at The Second International Workshop on Object Systems and Software Architectures (2006)
• Abstract of Thesis [pdf]
• Final Draft Thesis [pdf]
• Automated Transformation of BMF Programs[pdf], Presented at The First International Workshop on Object Systems and Software Architectures (2004)

Some Honours/Masters Theses

• Andrew Jeffery,Using the LLVM Compiler Infrastructure for Optimised, Asynchronous Dynamic Translation in Qemu[pdf], Honours Thesis (2009)
• Lee Wang Hao, ARM Instruction Set Simulation on Multi-Core x86 Hardware[pdf], Honours Thesis (2009)
• Ryan Craig, Evolution of Teams of Maze Solving Agents Using Grammatical Evolution, Honours Thesis (2008)
• Nisarg Patel, Visualising Optimiser [pdf] Masters thesis (2008)
• Sean Pettge, A Fast Code Generator for Point-Free Form [pdf] Honours thesis (2005)
• Joe Windows, Automated Parallelisation of code written in the Bird-Meertens Formalism [pdf] Honours Thesis (2003) University of Adelaide.
• Dean Philps, Mapping Parallel BMF Constructs to a Parallel Machine [pdf] Masters Thesis (2003) University of Adelaide.
• Paul Martinaitis, Simulation and Visualisation of Parallel BMF code [pdf] Honours Thesis (1998) University of Adelaide.

Honours/Masters projects

My current Honours project proposals can be found in the school's honours proposal list. Likewise, my current Masters project proposals can be found in the school's masters proposal list.

Teaching

2012 Courses

• Introduction to Programming(First Semester)
• Topics in Computer Graphics (First Semester)
• Advanced Topics in Computer Graphics (First Semester)
• Algorithm and Data Structure Analysis (Second Semester)
• Operating Systems (Second Semester)
• Computer Science Concepts (NAAEC Term 3)

Past Courses

• Computer Science I (A/B)
• Parallel Functional Programming (Honours/Masters).
• Systems Programmin in C and C++
• Programming Paradigms.
• Programming Techniques
• Data Structures and Algorithms
• Operating Systems
• Introduction to Software Engineering (old incarnation)

I have also presented short Ada and Java courses for industry.

Teaching interests

My primary teaching interests are:

• Introductory programming
• Problem Solving
• Functional programming
• Parallelism - High Performance Computing
• Optimisation and Performance

ACM Programming Contest

I am one of the of the Adelaide teams competing in ACM International Collegiate Programming Contest. The contest has enjoyed phenomenal growth in recent years, attracting very large numbers of competitors and strong corporate interest and sponsorship.

The regional round of the contest is held each year around mid-september. The South Australian Site this year was Flinder's University. Adelaide teams did very well again, claiming first and second places at the local site and 11th overall in the South Pacific Region. An Adelaide team qualified for the world final the previous two years running. Adelaide teams have a good record in this contest. If you a University of Adelaide student and you are interested in entering the contest in future email me. We also have a forum set up each year for the contest with information and resources for local teams.

Teaching Development

I am strongly involved in the development of core first year curricula and course material and, in particular, new practical exercises. I have also made major changes to the course content for Advanced Programming Paradigms, Parallel Functional Programming and in the development of the new Systems Programming course. I am a co-developer a new masters course, Specialised Programming, as part of the future development of a problem-solving curriculum in Computer Science.

Administration and Service

Current administrative roles

• School Assessment Coordinator
• Records and Marks System coordinator
• School NAAEC Coordinator
• Software Infrastructure and Web Support
• Schools Liason (Semester 1)
• Member of School Computing Committee
• Member of School Learning and Teaching Committee

Past Administrative roles

• Course Advisor
• First-Year Coordinator
• Occupational Health and Safety Officer
• Level A teaching coordinator
• Representative to the Faculty of Science
• Member of IT Subject Advisory Committee Senior Secondary Assessment Board of South Australia (SAABSA)