REANNZ develops and supports a range of products and services to support the specialist needs of our members in the R&E and innovation community.
We operate NZ's national research and education network and seamlessly connect with 120+ networks globally to enable researchers to collaborate.
REANNZ is proud to support the specialist needs of our members from NZ's research, education and innovation community.
Find out more about who REANNZ is, what we do and the people that operate New Zealand’s national research and education network.
REANNZ helped researchers optimise AI scheduling code, cutting runtime from 1,298 to 195 seconds and enabling larger experiments with fewer resources.
Yuan Tian and Qianhui Wang are PhD students at Te Herenga Waka—Victoria University of Wellington researching advanced optimisation algorithms for solving complex resource-constrained project scheduling problems. Their work focuses on developing AI-driven machine learning methods that can be applied to a wide range of planning scenarios.
One important application is medical dispatch, where ambulances and emergency response teams must be allocated to incidents while balancing constraints such as limited personnel, equipment availability, travel times, and treatment priorities. Faster and more effective scheduling decisions have the potential to improve emergency response times and patient outcomes. Beyond healthcare, similar optimisation techniques can benefit industries such as logistics, manufacturing, construction, and transportation by reducing planning time and improving operational efficiency.
The researchers' software is implemented in Python and uses genetic programming hyper-heuristics—a machine learning approach that evolves the scheduling rules themselves rather than learning fixed model parameters.
This flexibility enables the algorithm to discover highly effective scheduling strategies but comes at a significant computational cost, as evaluating thousands of candidate heuristics requires repeatedly solving large scheduling problems. As a result, execution times had become a major limitation to experimentation and algorithm development.
REANNZ worked with the researchers to profile the application, identify performance bottlenecks, and implement a series of targeted optimisations using a representative workload. The main improvements included:
The combined optimisations reduced the execution time of the representative test case from 1,298 seconds to 195 seconds—a 6.6× speedup. The substantial reduction in runtime will enable the researchers to evaluate many more candidate scheduling heuristics in the same amount of time, and save up to 3.4 million core-hours on REANNZ’s high performance platform.
This project has been a rare and highly valuable experience. Before working with REANNZ, we had been struggling with the computational efficiency of our research code for a long time. The consultancy provided both the expertise and the momentum we needed to address these challenges systematically. Communication throughout the project was highly effective. We are especially grateful to Alex and Thomas for taking the time to understand our research, listening carefully to our explanations, and being willing to explore a complex and sometimes untidy research codebase to identify the key performance bottlenecks. The collaboration has delivered substantial improvements to the efficiency of our software and will allow us to conduct larger and more comprehensive experiments within the same computational budget. Just as importantly, we have gained valuable practical knowledge about profiling and code optimisation. We hope to share these experiences with the wider research community so that other researchers facing similar computational challenges can also benefit.
This project has been a rare and highly valuable experience. Before working with REANNZ, we had been struggling with the computational efficiency of our research code for a long time. The consultancy provided both the expertise and the momentum we needed to address these challenges systematically. Communication throughout the project was highly effective.
We are especially grateful to Alex and Thomas for taking the time to understand our research, listening carefully to our explanations, and being willing to explore a complex and sometimes untidy research codebase to identify the key performance bottlenecks. The collaboration has delivered substantial improvements to the efficiency of our software and will allow us to conduct larger and more comprehensive experiments within the same computational budget. Just as importantly, we have gained valuable practical knowledge about profiling and code optimisation. We hope to share these experiences with the wider research community so that other researchers facing similar computational challenges can also benefit.
- Yuan Tian, Te Herenga Waka—Victoria University of Wellington
Find anything about our products, services, and more. Enter a query in the search input above.
