Mathematical Optimization for
Scalable Scientific Visualization

S. Frey

University of Groningen, The Netherlands

Scientific data is growing rapidly in memory footprint, resolution, parameter space, and ensemble size. This offers new opportunities for insight, but also introduces visualization challenges regarding scalability, interactivity, and human perception. Mathematical optimization provides a powerful framework for addressing these issues by framing visualization tasks as problems that can be solved efficiently and systematically.

In this talk, we explore how optimization is used in scientific visualization. We discuss the kinds of problems tackled in this context, such as data reduction, layout computation, and tuning of mapping and rendering parameters—and examine the objectives and evaluation criteria involved, from technical measures like runtime to quality metrics. We also consider the requirements and trade-offs across scenarios, from real-time interaction to offline processing, and how optimization results are communicated to users.

We complement our overview with concrete examples from the visualization community and own prior work on optimization-driven visualization. Looking ahead, we highlight emerging challenges and trends, such as tighter integration with machine learning, that will enable visualization systems to handle increasingly complex and large datasets more efficiently.

Keywords: Scientific Visualization, Optimization, Large Data.