Pontus Giselsson

Pontus Giselsson is an Associate Professor in the Department of Automatic Control at Lund University's Faculty of Engineering (LTH). His work bridges theoretical optimization and its practical applications in machine learning and control systems.

His research interests include optimization, convex and large-scale optimization, operator splitting methods, and numerical algorithms. These areas are central to modern control theory and machine learning, where efficient and scalable solutions are critical. He has made significant contributions to the understanding of convergence properties in first-order methods, particularly through Lyapunov analysis and performance estimation frameworks.

The recent publications highlight a strong focus on advancing the theoretical foundations of optimization algorithms, including the Chambolle-Pock method, forward-backward splitting, and generalized alternating projections. These works explore convergence conditions, momentum corrections, and deviations in algorithmic behavior, often in non-standard or nonlinear settings. The keywords across these articles reflect a deep engagement with mathematical programming, control theory, and computational mathematics.

Pontus Giselsson teaches a master's level course titled Optimization for Learning, offered annually, and previously taught a PhD-level course on Large-Scale Convex Optimization. He actively supervises research and collaborates with scholars such as M. Fält, S. Banert, and A.B. Taylor.

• Email: pontusg@control.lth.se
• Phone: 046-222 28 784