Investigating Research Design Optimisation in Cross-flagship Research

Investigating Research Design Optimisation in Cross-flagship Research using NICTA Optimization Research Group (ORG) Algorithms and Tools

Funded by CSIRO Digital Fusion Strategic Funds
Partners:	R. Gaire
Duration:	2015 (3 Months)

Balanced incomplete block design (BIBD) generation is a combinatorial problem. The concept was originally developed in the design of statistical experiments. In CSIRO, various researchers conduct experiments whose designs often need BIBD solutions. For example, in Agriculture research, specifically plant research, field experiments need optimisation of parameters such as genotypes, number of repeats, field plot size and plot location. This design is important to ensure the experiment is not affected by artefacts. Currently this problem is being solved by using third party tools, which might be unable to provide the global optimal solutions. Data61 (former NICTA) Optimisation Research Group (ORG) carries out fundamental and applied research that addresses grand challenges faced by our society in environmental and societal resilience, future energy systems, logistics, and supply chains. The tools developed in ORG for solving constraint programing problems have been applied across various domains. In this project, we investigated the applications of ORG tools for research design optimisation problems across different flagships. Specifically, we worked closely with the Agriculture Business Unit to develop proof of concept that can then be used in other research domains.