Description:
Many algorithms in mathematical optimisation work on the principle of breaking a complex problem into smaller, computationally tractable pieces and then exploiting the relative
simplicity of these smaller pieces as part of an iterative process. For algorithms that use first-order (i.e. gradient) information, operator splitting provides the unifying mathematical framework.
In recent times, there has been an explosion of interest in splitting algorithms due, on one hand, to their ideal suitability for large-scale problems in areas such as machine learning and, on the other hand, to a number of recent practically important mathematical developments such as variance reduction, acceleration, and explicit step size rules.
The aim of this workshop is to bring together international experts and Australian researchers working in the field of operator splitting and mathematical optimisation, creating an opportunity to meet, discuss and collaborate. The program is designed to provide an accessible introduction to the field for early-career academics as well as seasoned researchers alike.
This MATRIX Research Program is partially supported by AMSI and AustMS through the AMSI-AustMS Workshop Funding.
